Final Report PC2-FE2 - June 2005 - Land and Water - CSIRO · water. Beam Attenuation Coefficient...

Final Report PC2-FE2 - June 2005 1


CRC for Coastal Zone, Estuary and Waterway Management

PC2/FE2 Project Final Report June 2005

Port Curtis and Fitzroy River Estuary Remote Sensing Tasks

Editors A. Dekker and S. Phinn

From: A. Dekker, V. Brando, R. van Ede, P. Daniel, J. Anstee and A. Marks. Environmental Remote Sensing Group CSIRO Land and Water GPO Box 1666, Canberra, ACT 2601 Australia Tel: 61-2-62465821 Mobile: 61-0419411338 Fax: 61-2-62465815 Email: [email protected] S. Phinn, C. Roelfsema*, and P. Scarth Biophysical Remote Sensing Group School of Geography, Planning and Architecture *Centre for Marine Studies University of Queensland Brisbane, Queensland, AUSTRALIA, 4072 Tel: 61-7-3365-6526 Mobile: 0417-629765 Fax: 61-7-3365-6899 Email: [email protected] Associated Student: F. Manson Biophysical Remote Sensing Group School of Geography, Planning and Architecture University of Queensland Brisbane, Queensland, AUSTRALIA, 4072


Table of Contents -

Item Page

(i) Executive Summary 4 (ii) Glossary 7

1. Users’ Summary (GIS technicians, resource managers, scientists) 10

2. Supporting Science Report 17

2.1 An integrated approach for Mapping Water Quality and Substrate

Types in Coastal Environments – Lessons from Moreton Bay 17

2.2 Port Curtis and Fitzroy Estuary – Underwater Light Climate and Water Quality Mapping 19 2.2.1 Field and Image Data Collection 2000 - 2003 20 2.2.2 Field and Image Data Pre-Processing 25 2.2.3 Underwater Light Climate – Bio-optical Model Development 41 2.2.4 Mapping Water Quality 58

2.3 Port Curtis and Fitzroy Estuary – Substrate Mapping 81

2.3.1 Field and Image Data Pre-Processing 81 2.3.2 Mapping Substrate Variables 81

2.4 Terrestrial – Coastal Environment Mapping and Monitoring 92 2.4.1 Terrestrial Land-Cover Mapping Port Curtis – Fitzroy Estuary 92 2.4.2 Land-Cover Change and Trend Detection Port Curtis 1990 – 2002 101

3. Conclusions and Recommendations 120 4. References 123

5. Acknowledgements 127

6. Appendices: 128

Appendix 1 Communication of Results by the Remote Sensing Team Appendix 2 Student Achievements

Appendix 3 Image Processing Log

Appendix 4 Publications Produced in Association with the PC2 and FE2 Tasks Appendix 5 Full-page Water Quality Parameter Maps for Fitzroy Estuary and Port Curtis


(i) Executive Summary

Introduction and Structure From 2000 onwards research teams from CSIRO and the University of Queensland carried out integrated research to establish methods for remote sensing based mapping and monitoring variables of coastal environments that may be used to monitor their condition or health. For the context of this report coastal environments include mangrove wetlands, inter-tidal and sub-tidal sections of a coastal embayment. The format for this report is based on the four main tasks originally identified by the project team; each of these tasks is outlined below. Task PC/FE 2.1 Information needs for remote sensing of coastal environments.

Task PC/FE 2.2 Remote sensing products (data and processing techniques) addressing identified specific monitoring and management agency information requirements. This includes an inventory and assessment of availability and accessibility of remotely sensed data for Port Curtis and the Fitzroy Estuary. Task PC/FE 2.3 Documented techniques for acquiring and processing remotely sensed data in Port Curtis and the Fitzroy Estuary to produce reliable output information. This coastal ecosystem health information consists of relevant aquatic, inter-tidal and terrestrial variables. Task PC/FE 2.4 Development of remote sensing algorithms for the following environmental zones/variables: aquatic (turbidity, phytoplankton, algal blooms), intertidal, and mangroves. Task PC/FE 2.5 Calibrated and validated remote sensing data for assimilation into (non-remote sensing) models developed for Port Curtis and the Fitzroy Estuary. Task PC/FE 2.6 Publications and conferences. Methods This projects intended to build on the satellite image processing methods (using commercially available Landsat images) developed in the Moreton Bay project (Phinn & Dekker, 2005) to map the constituents (CDOM – coloured dissolved organic material, tripton – total suspended inorganic material - TSM) controlling the optical properties of a water body, to map the transparency and to map the substratum type and cover. This approach is based on measuring and understanding the underwater light climate characteristics (or properties) controlling the absorption, scattering and transmission of sun and sky light, in different sections of the Fitzroy River Estuary and Port Curtis waters. In simple terms, the approach selects f or each image element (pixel) the appropriate combination of optically active substances. Thresholds based on these specific inherent optical properties are applied using an automated routine that determines how to apply the algorithms for mapping water quality variables (e.g. TSM, CDOM concentration and Secchi disk transparency) and thresholds based on transparency determine where to map the appropriate substrate types. This project also included a terrestrial mapping and monitoring component. Vegetation index images were calculated for a series of commercially available Landsat images to map changes and trends in vegetation cover in the immediate area of the Fitzroy River Estuary and Port Curtis Harbour. A new form of image analysis was developed to map trends in vegetation cover from a time series of satellite images. The resultant maps serve as a demonstration product for future


environmental monitoring programs and can be applied to any image data set or image based products, e.g. this approach could also be applied to the water quality parameter maps developed in the project. Although preliminary in nature, both the water-based and vegetation-trend processing approaches represent significant improvements for remote sensing of the coastal zone due to: (1) a common international requirement to monitor and manage coastal estuaries around the world to enable sustainable management; (2) the reliable and accurate nature of the results produced; (3) extension and validation of the approaches beyond Moreton Bay and 4) a processing method that is easily adapted to other sensors. Conclusions

(1) Commercially available, multi-spectral, moderate spatial resolution satellite image data can be used to produce maps of aquatic and terrestrial environmental variables relevant to coastal ecosystem health monitoring in the Fitzroy Estuary and Port Curtis coastal areas.

The project team was able to apply the methods developed in the Moreton Bay project to map the following water quality variables from commercial satellite image data (Landsat 5 Thematic Mapper and Landsat 7 Enhanced Thematic Mapper) collected over Port Curtis and Fitzroy Estuary in 1986, 1990, 1995 and 2002: - coloured dissolved organic material (CDOM); - total suspended solids (tripton); and - Secchi disk transparency. Distinct optical domains, or zones of water colouration, were evident in the Fitzroy Estuary and Port Curtis water quality maps, and appeared to be controlled by tidal currents and stream flow, along with surface and sub-surface aquatic vegetation. Due to the predominantly turbid nature of both Port Curtis and Fitzroy Estuary waterways, substrate mapping was not completed using these available Landsat Thematic Mapper (TM)/Enhanced TM (ETM) images, as they were selected for optimal water and land vegetation assessment and not targeted for substratum mapping purposes. Substratum mapping may be possible in Port Curtis under conditions when the tide is low and stream-flow is low, resulting in exposed inter-tidal seagrass beds and visible sub-tidal beds. With the increasing availability of high spatial resolution satellite image data from sensors that can be programmed to image an area at a specific time (currently Quickbird, IKONOS and SPOT) the likelihood of acquiring suitable images for substrate mapping will increase. Terrestrial environments were also examined as part of this work, focussing on the use of “vegetation indices” to map vegetation cover throughout the immediate coastal areas of Fitzroy Estuary and Port Curtis.

(2) The image processing approach developed for optically complex waters in Moreton Bay

can be applied successfully to other coastal environments with different specific inherent optical properties.

The optimised matrix inversion model used to estimate the concentration of organic and inorganic water column constituents was the same model as was developed for Moreton Bay. In this case the model was driven by a set of specific inherent optical properties (SIOP) collected during fieldwork in 2002 in Fitzroy Estuary and Port Curtis areas. Realistic maps of CDOM, TSM and Secchi disk transparency were produced, with the associated error maps clearly indicating areas where substrate was visible or exposed.

(3) Multi-date image analysis techniques can be used to produce maps depicting trends in the

state or condition of an environmental parameter over time.


To present the results from multiple dates of vegetation index images in one image that would summarise the change in a pixel’s vegetation cover over time, a trend detection approach was developed and applied to a five date series of vegetation index images over the Gladstone area. This approach was applied and validated successfully using multi-date Landsat TM/ETM based normalised difference vegetation index (NDVI) images and aerial photography as a reference source. Areas that showed an increase or decrease in vegetation cover were identified. The approach could also be applied to multi-date substrate cover type and water quality parameter image maps. (4) Future work can be used to take the demonstration projects presented in this work to

operational, accurate and cost-efficient environmental mapping and monitoring programs.

Although the image data used here was the most commonly available commercial product when this project commenced, there are a range of different sensors now operating that could provide at least the same, or probably more information. For water quality monitoring there is a need to enable a high-frequency (e.g. daily) data collection to monitor changes in water quality distribution as its controlling processes act at this temporal scale. Confidence in the model results was assessed, however, more work needs to be completed on the validation of products. Products from the MODerate resolution Imaging Spectrometer (MODIS) and the MEdium Resolution Imaging Spectrometer (MERIS) sensors should be examined as new data sources, due to their high temporal, spectral and radiometric resolutions, specifically designed for measuring coastal and ocean waters. The same applies to terrestrial vegetation monitoring, where MODIS and MERIS products may also be useful. Potential improvement to seagrass mapping in this region requires assessment of the capabilities of high spatial and radiometric resolution satellite images, along with airborne and satellite hyperspectral imaging sensors. In each case these sensors will improve mapping ability in shallow waters and/or low density seagrass cover.


(ii) Glossary Term Description Absorption Coefficient Measurement of the amount of sun- and skylight absorbed per metre

depth of water. Air-Water Interface Ocean-atmosphere boundary which alters the direction and magnitude

of sun- and skylight. ALI Advanced Land Imager - next generation Landsat TM/ETM

experimental sensor on the EO-1 platform. Analytical Modelling Image processing approach used to estimate biophysical properties

(e.g. total suspended sediment concentration) by applying radiative transfer equations to image data.

AOP’S Apparent optical properties of water - those dependent on the strength and direction of incident sun- and skylight, e.g. reflectance.

Atmospheric Correction Image processing approach used to remove atmosphere scattering and absorption effects from image data so the signal remaining is that emanating from the air-water interface or from the land surface (thus not from the top of the atmosphere)

Attenuation Reduction of true reflectance or radiance levels with increasing depth in the water column due to absorption and scattering effects per metre depth of water.

Backscattering Coefficient

A measure of the proportion of light that will be backscattered - dependent on particle size, composition and shape per metre depth of water.

Beam Attenuation Coefficient

Measurement of the amount of light reduction by a collimated beam of light per metre depth of water.

Bio-Optical Modelling Application of radiative transfer based models to calculate light interactions in natural water to estimate water column constituents from remotely sensed data.

Biophysical Properties Biological and physical properties of an environment. CDOM Coloured dissolved organic matter. Coastal Environments Mangrove wetlands, inter-tidal and sub-tidal sections of a coastal

embayment. Collimated A parallel beam of light (it does not spread out)

C-WOMBAT-C Coastal Waters and Ocean MODTRAN-4 Based ATmospheric Correction software.

Diffuse Skylight Sun-light that has been scattered by aerosol, dusts and clouds, and is not in a direct beam from the sun.

Downwelling Light moving downwards and away from the sun and sky to the earth's surface or into the water.

Downwelling Irradiance Sun- and skylight incident upon a surface from all viewing angles. EO-1 TRW/NASA satellite Earth Observer -1 : experimental platform for

Hyperion and ALI sensors. Geometric Correction or Georeferencing

The process of converting an image into a form where each pixel has a projection, datum and coordinate, enabling it to be integrated with other spatial data.

Hydroscat-6 Instrument for measuring the backscattering coefficient of water bodies at a six wavelengths.

Hyperion Experimental satellite-borne hyperspectral imaging sensor with 30m pixels and 196 spectral bands on the EO-1 platform.

Hyperspectral A device is hyperspectral if it records reflected or emitted light using (usually > 10), narrow overlapping spectral bandwidths (cf. multispectral) covering the entire wavelength range required.

Ikonos Commercially operated high spatial resolution (4m) multispectral and panchromatic (1m) satellite imaging sensor.


Image Corrections Image processing operations designed to remove geometric and atmospheric distortions in an image data set.

Image Pixel Smallest spatial sampling unit of an imaging sensor to measure reflected or emitted energy from the surface of the earth.

Inherent Optical Domain

An area of water exhibiting similar specific absorption and specific scattering properties.

Inherent Optical Properties

Properties of the water column that do not change with illumination strength and direction, i.e. absorption and (back)scattering coefficients.

Irradiance Reflectance A ratio of (upwelling) reflected sun- and skylight measured over all directions to incident (downwelling) sun- and skylight.

Kd Vertical diffuse attenuation coefficient - an estimate of the amount of light attenuated by water per metre depth of water.

Landsat Thematic Mapper (TM) Enhanced Thematic Mapper (ETM)

General purpose multispectral, moderate spatial resolution imaging satellite sensors that have operated since 1984 (5TM) and April 1999 (7ETM).

Matrix Inversion Method

A mathematical solution to invert the bio-optical model equations to estimate concentrations of the water column constituents.

Multi-Date Normalization

An image correction approach used to remove the distorting effects of different types of atmospheric conditions over an image of the same area collected on different dates.

Multi-Spectral An image data collection system where reflected or emitted energy is measured in less than 10 broad discontinuous spectral bands.

Multi-Temporal Data that has been collected for the same area over a series of successive dates.

Near Infra Red (NIR) The non-visible 700-1000 nm portion of the EMR spectrum. Normalised Difference Vegetation Index

A normalised ratio transformation using the red and NIR pixel values that is related to vegetation cover.

Remote Sensing Reflectance Or R(0-)

Atmospheric and air-water interface correction of image data produces the ratio of upwelling radiance from the water surface to downwelling irradiance just below the water surface.

PIFs Pseudo-Invariant Features - landscape features that retain the same reflectance over time and can be used for image calibration.

Planimetric Projection A map projection where each map element is at its true location on the ground (cf. aerial photography and some satellite images which have a perspective projection).

PSICAM (Point Source Integrating Cavity Absorption Meter)

A device for measuring absorption properties of water samples based on spectrometer measurements of a water sample within a diffuse light environment.

QuickBird Commercially operated high spatial resolution (2.68m) multispectral and panchromatic (0.7m) satellite imaging sensor with similar VIS and NIR bands to Landsat.

Radiative Transfer Modelling Approach

All radiative transfer numerical models compute radiance distributions and related quantities (irradiance, reflectance, diffuse attenuation functions, etc.) in the water column as a function of the water absorption and scattering properties, the sky and air-water interface conditions and the bottom boundary conditions

Radiometric Correction Conversion of image pixel digital numbers from a relative scale (usually Volts in a detector) to an absolute physical units scale, e.g. radiance.

Radiometric Resolution Smallest difference in radiance (light) able to be detected by a sensor and its dynamic range.

Radio-Sonde Data A measured vertical profile through the atmosphere of temperature and humidity, used for atmospheric correction.

Reflectance Or Spectral Signature

Ratio of upwelling radiance to downwelling irradiance on a target, i.e. its characteristic reflectance response.


Secchi Depth Commonly used approach to measure the clarity of a water body by dropping a black/white disk and noting the depth at which the black and white is no longer distinguishable.

Spatial Resolution A measure of the scale of the smallest feature able to be detected on an image.

Spectral Angle Mapping

An image classification approach used to group pixels with similar reflectance signature.

Spot XS Commercially operated high-moderate spatial resolution (20m) multispectral and panchromatic (5m) satellite imaging sensor.

Striping Effects Artefacts in Landsat images due to scanner or detector variations (in the case of Landsat 5TM every 16th lines).

Substrate Types Types of material found on seafloor: sand, rock, seagrass, algae etc…(the exact terminology is substratum)

Subsurface Irradiance Reflectance

R(0-) measured as a ratio of upwelling to downwelling irradiance just below the water surface.

Supervised Classification

Image processing approach used to group pixels with a similar spectral signature as these are assumed to be the same land-cover class.

Suspended Sediment Sediment particles that are floating in suspension (as opposed to dissolved) in the water column.

Total Suspended Material (TSM)

All particulate matter suspended in the water column that does not pass through a 0.45 micron filter

Tripton The non-algal part of TSM

Underwater-Light-Climate

A physics-based description how light behaves in a water column.

Upwelling Irradiance A measurement taken using a field spectrometer with the sensor head pointing down to capture upwards directed (=reflected from the water column or substrate) direct and diffuse light over all directions.

Water Quality Variables Biological and physical properties of the water column that have been adopted by scientific and management agencies as indicative of the condition of a water body, e.g. Secchi depth, chlorophyll and total suspended sediment concentration.


1. Users Summary This document presents results from inter-disciplinary work over a period of four years by research teams from the University of Queensland and CSIRO to establish remotely-sensed methods for mapping and monitoring variables of coastal environments that may be used to monitor their condition or health. For the context of this report coastal environments include mangrove wetlands, inter-tidal and sub-tidal sections of a coastal embayment. The original aim of this project was: To develop remote sensing tools that can be used to assess the ecosystem health of the Port Curtis/Fitzroy Estuary and to produce output information on the aquatic, intertidal and terrestrial environments related to ecosystem health and condition in Port Curtis and the Fitzroy Estuary.

Remote sensing tools were developed through this task, and results from applying these tools were presented as demonstration products for selected water quality and terrestrial variables that could be used to determine the ecosystem health of Port Curtis, the Fitzroy River and Keppel Bay. Cost effective and repeatable monitoring techniques developed in the Moreton Bay project, that use off-the shelf commercial satellite image data, were applied to time-series of satellite images (Figure 1), from 1990 to 2002, in an attempt to produce maps of submerged aquatic vegetation, water column constituents (organic and non-organic suspended and dissolved material) and terrestrial vegetation cover.

This study matched available remote sensing data with the environments for which they are suited, to develop processing techniques capable of providing reliable estimates of select biophysical properties in coastal environments. These data sets may play a role in scaling-up of field based sampling and models to cover larger areas and to provide input and calibration/validation for biogeochemical and physical models related to estuarine/embayment dynamics, water quality and benthic flora/fauna. Results from this project will contribute to a number of modelling projects for the Fitzroy River Estuary and Keppel Bay area in Phase 2 projects of the Coastal CRC. Production of reliable maps of water-quality or substrate types in coastal waters from remotely sensed data is a complex task, due to varying optical properties and concentrations in the water column. Coastal embayments such as Port Curtis, the Fitzroy Estuary and Moreton Bay are highly dynamic and display optically complex features due to oceanic tidal influences and terrestrial inputs from rivers and creeks (McEwan et al. 1998). Case 1 waters, or oceanic waters, have optical properties (and thus remote sensing images) dominated by phytoplankton and associated pigments. Case 2 waters, typically coastal and lacustrine waters, contain constituents that do not co-vary with chlorophyll, i.e., coloured dissolved organic material (CDOM), total suspended sediments (TSM)/inanimate detritus (tripton) and bacteria (Morel and Prieur, 1977, Gordon and Morel 1983). Remote sensing techniques have been successfully developed and are now applied on an operational basis for mapping biophysical properties of case 1(open ocean waters) (IOCCG 2000). However, case 2 waters continue to represent a challenge to remote sensing techniques, and recent reviews on the state of the art in mapping water quality in case 2 waters highlight the need for a new approach to using remotely sensed data in the optically complex waters of these environments (IOCCG 2000, Dekker et al. 2001b, Malthus and Mumby, 2003). As part of the Moreton Bay project our project team developed and applied an approach applicable for case 2 waters, the Inherent Optical Domain based mapping of aquatic coastal ecosystem properties. This approach was used to map the dissolved organic material and suspended sediment present in the water column, the Secchi disk transparency, as well as seagrass density and a type of algal bloom, from commercially available Landsat 7 ETM images. As shown in the Landsat 7 image of the study area taken during the dry-season, at the top of a high-tide (Figures 1, 2 and 3), the Fitzroy River Estuary and Port Curtis harbour are complex coastal embayments, with a mix of near Case 1 and Case 2 waters. Near Case 1 waters dominate in the optically-deep off-shore areas, at least 5.0 km offshore. Inputs from the Fitzroy River provide the most dominant source of sediment and organic material for the area. In contrast, the Calliope and Boyne Rivers drain significantly smaller catchments, but still provide inputs of sediment and


organic material for Port Curtis harbour area. There have been no previous published attempts at mapping water quality variables across all of Port Curtis Harbour or within the Fitzroy River Estuary. Our approach has been to design an innovative method that can be used with commercially available satellite image data to map the constituents (CDOM – coloured dissolved organic material, tripton – total suspended inorganic material - TSM) controlling the optical properties of a water body, the Secchi disk transparency and to also map the substrate type and its characteristics (Phinn and Dekker 2005, Phinn et al. 2005). This approach is based on measuring and understanding the underwater light climate characteristics (or properties) controlling the absorption, scattering and transmission of sun and sky light, in different sections of the Fitzroy River Estuary and Port Curtis waters (Figure 4). In simple terms, the approach divides an image into segments based on the level of transparency in the water column as well as the source of the optically active substances. Thresholds based on specific inherent optical properties applied using an automated routine will then determine where to apply algorithms for mapping water quality variables (e.g. TSM concentration) and substrate types. Due to the dynamic nature of the water quality parameter maps, and the influence of these on substrate mapping, sufficient data were only able to be collected for calibration of the image processing routines. Validation of these products will be undertaken in Phase 2 of the Coastal CRC work in this region. A method was developed to assess the confidence of the models used to derive the water quality variables, and the output maps (confidence assessment maps) from this method provide a spatial assessment of where the models find gaps in our understanding of either the water column or the substrate. This project also included a terrestrial mapping and monitoring component. Vegetation index images were calculated for the same series of commercially available Landsat images to map changes and trends in vegetation cover in the immediate area of the Fitzroy River Estuary and Port Curtis Harbour (Figure 1). A new form of image analysis was developed to map trends in vegetation cover from a time series of satellite images. The resultant maps serve as a demonstration product for future environmental monitoring programs and can be applied to any image data set or image based products, e.g. this approach could also be applied to the water quality parameter maps developed in the project. Although preliminary in nature, both the water-based and vegetation-trend processing approaches represent significant improvements for remote sensing of the coastal zone due to: (1) a common international requirement to monitor and manage coastal estuaries around the world to enable sustainable management; (2) the reliable and accurate nature of the results produced; (3) extension and validation of the approaches beyond Moreton Bay (Phinn & Dekker, 2005) and 4) easy adaptability to remote sensing image data from other sensors. This methodology may be applied to any other optical remote sensing data with only minor modifications. Thus it introduces timesaving and a standardised methodology. This method is now ripe for operationalising for coastal water management.


Figure 1: Landsat 7 Enhanced Thematic Mapper true-colour scene of Port Curtis and the Fitzroy River Estuary, captured at high tide, 09:45 on July 24th 2002.

KKKeeeppppppeeelll BBBaaayyy

CCCuuurrrttt iiisss IIIssslllaaannnddd

RRRoooccckkkhhhaaammmppptttooonnn

FFFaaaccciiinnngggIIIssslllaaannnddd

CCCaaapppeee CCCaaappprrr iiicccooorrrnnn

CCCaaalll lll iiiooopppeee RRRiiivvveeerrr

BBBoooyyynnneee RRRiiivvveeerrr

PPPooorrrttt AAAlllmmmaaa

FFFiii tttzzzrrroooyyy RRRiiivvveeerrr

TTThhheee NNNaaarrrrrrooowwwsss

GGGlllaaadddssstttooonnneee

CCCaaasssuuuaaarrr iiinnnaaa CCCrrreeeeeekkk

CCCuuurrrttt iiisss CCChhhaaannnnnneeelll

PPPeeelll iiicccaaannn BBBaaannnkkksss

NNNooorrrttthhh EEEnnnttt rrraaannnccceee

KKKeeeppppppeeelll IIIssslllaaannndddsss

KKKeeeppppppeeelll SSSaaannndddsss

YYYeeeppppppoooooonnn


Figure 2: Tidal curves for Gladstone Harbour and Port Alma at the time of Landsat image acquisitions (usually between 09:30 and 10:00).

16-08-1990 Port Alma

-1.5

-1

-0.5

0

0.5

1

1.5

2

2.5

00:00

02:00

04:00

06:00

08:00

10:00

12:00

14:00

16:00

18:00

20:00

22:00

Time (Hours)

Rea

ding

(m)

16-8-1990 Gladstone

-1.5

-1

-0.5

0

0.5

1

1.5

2

00:00

02:00

04:00

06:00

08:00

10:00

12:00

14:00

16:00

18:00

20:00

22:00

Time (Hours)

Rea

ding

(m)

7-3-1995 Gladstone

-1.5

-1

-0.5

0

0.5

1

1.5

00:00

02:00

04:00

06:00

08:00

10:00

12:00

14:00

16:00

18:00

20:00

22:00

Time (Hours)

Rea

ding

(m)

7-3-1995 Port Alma

-1.5

-1

-0.5

0

0.5

1

1.5

2

00:00

02:00

04:00

06:00

08:00

10:00

12:00

14:00

16:00

18:00

20:00

22:00

Time (Hours)

Rea

ding

(m)

24-7-02 Gladstone

-2.5-2

-1.5-1

-0.50

0.51

1.52

2.5

00:00

02:00

04:00

06:00

08:00

10:00

12:00

14:00

16:00

18:00

20:00

22:00

Time (Hours)

Rea

ding

(m)

24-7-02 Port Alma

-3

-2

-1

0

1

2

3

00:00

02:00

04:00

06:00

08:00

10:00

12:00

14:00

16:00

18:00

20:00

22:00

Time (Hours)

Rea

ding

(m)


Figure 3: Stream discharges for the Fitzroy, Calliope and Boyne Rivers prior to each Landsat TM/ETM image acquisition (21st August 1986, 16th August 1990, 7th March 1995 and 24th July 2002). Note that the range of the flow is adjusted to 10, 100 or 1000 m-3 s-1 in each plot (data courtesy of Queensland Department of Natural Resources and Mines).

0102030405060708090

100

27-Jul1986

1-Aug1986

6-Aug1986

11-Aug1986

16-Aug1986

21-Aug1986

26-Aug1986

Riv

er fl

ow (

m3 s

-1)

Calliope River at Castlehope Boyne River at Awoonga Dam Headwater

0123456789

10

27-Jul1986

1-Aug1986

6-Aug1986

11-Aug1986

16-Aug1986

21-Aug1986

26-Aug1986

Riv

er fl

ow (

m3 s

-1)

Fitzroy River at The Gap Raglan Creek at Old Station

0123456789

10

22-Jul1990

27-Jul1990

1-Aug1990

6-Aug1990

11-Aug1990

16-Aug1990

21-Aug1990

Riv

er fl

ow (

m3 s

-1)


0102030405060708090

100

22-Jul1990

27-Jul1990

1-Aug1990

6-Aug1990

11-Aug1990

16-Aug1990

21-Aug1990

Riv

er fl

ow (

m3 s

-1)


0123456789

10

10-Feb1995

15-Feb1995

20-Feb1995

25-Feb1995

2-Mar1995

7-Mar1995

12-Mar1995

Riv

er fl

ow (

m3 s

-1)


0100200300400500600700800900

1000

10-Feb1995

15-Feb1995

20-Feb1995

25-Feb1995

2-Mar1995

7-Mar1995

12-Mar1995

Riv

er fl

ow (

m3 s

-1)


0123456789

10

29-Jun2002

04-Jul2002

09-Jul2002

14-Jul2002

19-Jul2002

24-Jul2002

29-Jul2002

Riv

er fl

ow (

m3 s

-1)


0123456789

10

29-Jun2002

04-Jul2002

09-Jul2002

14-Jul2002

19-Jul2002

24-Jul2002

29-Jul2002

Riv

er fl

ow (

m3 s

-1)



Figure 4: The key processes controlling the amount of scattered, absorbed and transmitted sun- and skylight from aquatic environments recorded by imaging sensors. The original objectives of this project were: Task PC/FE 2.1 Establish information needs for remote sensing of coastal environments.

Task PC/FE 2.2 Develop remote sensing products (data and processing techniques) addressing specific monitoring and management agency information requirements. This includes an inventory and assessment of availability and accessibility of remotely sensed data for Port Curtis and the Fitzroy Estuary. Task PC/FE 2.3 Document techniques for acquiring and processing remotely sensed data in Port Curtis and the Fitzroy Estuary to produce reliable output information. This coastal ecosystem health information consists of relevant aquatic, inter-tidal and terrestrial variables. Task PC/FE 2.4 Develop of remote sensing algorithms for the following environmental zones/variables: aquatic (turbidity, phytoplankton, algal blooms), intertidal, and mangroves.


Task PC/FE 2.5 Provide calibrated and validated remote sensing data for assimilation into (non-remote sensing) models developed for Port Curtis and the Fitzroy Estuary. Task PC/FE 2.1 was presented in previous annual reports and published in the following two publications; hence it is not discussed in detail here Phinn, S., Nightingale, J. and Stanford, M. (2001) Review of the Status of Remote Sensing for Environmental Monitoring and Management Applications in Australia. Research. Monograph, The University of Queensland Foundation Research Excellence Awards for Early Career Researchers, 2001. Biophysical Remote Sensing Group Report 2001-3 Phinn, S., Nightingale, J. and Stanford, M. (2002) Focus on Remote Sensing: A National Survey of Remote Sensing for Environmental Monitoring and Management Applications in Australia. GIS User, 51:26-27 The format for the remainder of the report is based on the main objectives. However, to provide a more coherent and useful document for the reader it has been restructured into sections that provide coverage of specific mapping goals in Port Curtis and Fitzroy Estuary. Section 2.1 contains a summary, sections 2.2 – 2.4 contain the key findings for each mapping goal (Tasks PC/FE 2.2 – 2.5), while the appendices contain a summary of communication products and student activities associated with the project. Section 2.1 outlines the approach originally developed in Moreton Bay (Phinn and Dekker, 2005) that was applied to map water quality and substrate variables in the optically complex coastal waters of Port Curtis and Fitzroy Estuary. Section 2.2 describes the approach used to measure the optical properties of Port Curtis and Fitzroy Estuary waters for parameterising a bio-optical model (or underwater light climate model). An inversion of this bio-optical model was used to produce maps depicting the concentration of suspended and dissolved materials and the Secchi disk transparency in the waters of Port Curtis and Fitzroy Estuary. Section 2.3 describes the approach used to map substrate types in Port Curtis, focussing on the mapping of intertidal and sub-tidal substrate types Section 2.4 describes the integration of terrestrial and environmental change analyses based on processed remotely sensed images.


2. Supporting Science Report 2.1 An Integrated Approach for Mapping Water Quality and Substrate

Types in Coastal Environments Lessons from Moreton Bay As noted in Section 1, application of optical remotely sensed data (i.e. where sensors rely on measurement of reflected sun- and skylight) to coastal waters with a mix of case 1 (ocean) and case 2 (coastal-riverine) types has prevented reliable mapping of coastal zone ecosystem health indicators from remotely sensed data (IOCCG 2000, Dekker et al. 2001b). Based on our work in the Moreton Bay (Phinn & Dekker, 2005) we established a sequence for processing satellite or airborne images of the coastal zone to produce maps depicting valid concentrations of organic and inorganic material in the water column and Secchi disk transparency. Concentrations of these materials and transparency are internationally recognised measurements of water quality in coastal, estuarine and riverine areas (IOCCG 2000). In addition, the process enables production of maps depicting the spatial distribution of substrate types if they are visible through the water column (e.g. sand, seagrass, algae, etc.) and in some cases maps of biophysical properties of the substrate type (e.g. seagrass density). Figure 5 provides an overview of the sequence referred to as an “inherent optical domain based mapping of aquatic coastal ecosystem properties.” This framework was the end-result of a number of trial and error based approaches to map water column and substrate type features, and addressed limitations of mapping in optically complex coastal waters (Dekker et al. 2001b; Phinn and Dekker, 2005). Preliminary results from Moreton Bay demonstrated that this approach will: (1) produce accurate maps of CDOM, TSM and Secchi disk transparency in areas where substrate is not visible through the water column; and (2) produce accurate maps of substrate cover type and density where the substrate is visible through the water column. The success of the approach in the Fitzroy River Estuary and Port Curtis harbour waters will depend on these two conditions. Preliminary assessment of the images indicates that substrate visibility often does not occur in this region. This project also had a focus on mapping changes to the terrestrial environment of the Port Curtis / Fitzroy Estuary area. Standard approaches for transforming image data to maps of vegetation cover per pixel were first applied. Then new techniques for mapping changes and trends in land-surface features were used to produce a single output map depicting increase or decrease in vegetation cover between two dates, and another map categorising the trend in vegetation cover change over multiple image dates.


Figure 5: Flowchart illustrating the sequence of image processing operations and data sets required in the Optical domain based mapping of aquatic coastal ecosystem properties from Phinn et al. (2004a, b)

Input unprocessed image data set of a coastal area

(satellite or airborne)

Coastal water body R(0-) image to be mapped

WATER QUALITY MAPPING Output = Concentration maps ofCDOM and tripton/ suspendedsediments

Corrected image of coastal area

R(0-) subsurface irradiance reflectance or remote sensing reflectance

Mask out all image areas above MHWM

(image or bathymetry based)

Image corrections - Geometric - Georeferencing - Radiometric - Atmospheric - Air-Water Interface

Application of supervised classification to delimit optical domains

(deep clear water, clear water + substrate, turbid water types, exposed substrate)

SUBSTRATE MAPPING

Automatic selection of SIOPs Application of an optimisation technique

to select appropriate SIOP’s for each image pixel and define image pixels

unable to be processed

Approach #2 Integration of bathymetry, imagedata, field signatures and aclassification process to mapsubstrate type within suitabledepth and optical domains, i.e. <5m and not affected by highsuspended sediment levels

Approach #1 Integration of SIOPs, image data,field signatures and inversiontechnique to retrieve depth andsubstrate type in the opticaldomains where sufficient lightpenetrates to -and is reflected from-the substrates

Image map of deep clear water, clear water + substrate, turbid water types, exposed

substrate in the coastal water body

Separate or combined maps of: - Exposed substrate types - Concentration of organic/inorganic water constituents - Substrate type and depth - Areas where algorithms cannot be applied

Output = Maps of substrate type (density) and depth

Field measurement of substrate type spectral

signature

Integration of SIOP, image dataand inversion technique toretrieve concentrations of organicand inorganic material in theoptical domains not affected bysubstrate influences

Field measurement of Specific Inherent Optical Properties (SIOPs) (specific absorption and

backscattering) for each optical domain


2.2 Port Curtis and Fitzroy Estuary Underwater Light Climate and Water Quality Mapping

2.2.1 Field and Image Data Collection 2000- 2003 2.2.2 Field and Image Data Pre-Processing 2.2.3 Underwater Light Climate – Bio-optical Model Development 2.2.4 Mapping Water Quality


2.2.1 Field and Image Data Collection 2000- 2003

Underwater-light-climate/Optical properties measurement Implementation of this new approach to map coastal water and substrate properties requires a detailed description of the relationship between the inherent and apparent optical properties of the water column and the water constituents. The inherent optical properties (IOPs) are the properties of the medium itself regardless of the ambient light field; the IOPs are measured by active (i.e. having their own light source) optical instruments. The radiometric variables are the basic properties of the light that are measured by passive optical instruments (using the sun as the light source). The apparent optical properties (AOPs) are combinations of radiometric variables that can be used as indicators for the colour or transparency of the water, for example the reflectance or vertical attenuation coefficients. During the project two field campaigns were carried out to characterise the optical properties of Port Curtis and the Fitzroy River Estuary. An intensive measurement campaign was conducted on 16-17th January 2002 to measure the optical properties of the water in the Fitzroy River Estuary during the wet season (Figure 6), in response to a record downpour of 400 mm in the Springsure-Comet areas of Central Queensland overnight on January 4 2002. A larger intensive measurement campaign was conducted from 21-24th June 2002 to characterize the optical properties for the winter dry season for: i) the Fitzroy River upstream the barrage; ii) the Fitzroy River Estuary from the barrage to Keppel Bay; and iii) Port Curtis waters. During the wet season field campaign apparent and inherent optical properties were measured at six locations (Figure 6) while for the “dry season” twelve locations were visited (Figure 6). For both the measurement campaigns a unique set of instruments was deployed to measure the IOP’s, AOP’s and concentrations of the water constituents at a set of locations which were considered to represent the different colours of water. The variables measured and instrumentation used to record them are listed in Table 1. The variables measured in Table 1 are explained in Definition Tables 1-3.


Figure 6: Sample sites visited during field campaigns conducted on 16-17th January 2002 and June 21-24th 2002 to measure specific inherent optical properties of Fitzroy River Estuary and the Port Curtis waters.


Table 1. Inherent optical properties (IOP), apparent optical properties (AOP) and biophysical measurements made in Fitzroy River Estuary and the Post Curtis during the 16-17th January 2002 and 21-24th June 2002 field programs. The variables used are presented in Definition Tables 1-3.

Instrument Measured parameters

Estimated Parameters

RAMSES MCC Ed air UQ- BRG

RAMSES MCC & MRC Ed/Lu R(0-), R(z), Kd, Ku

CLW

HydroScat 6 bb 145o bb CLW / EOC

PSICAM atot(λ), adetr(λ), aCDOM(λ) aphy(λ)

CLW

AC9 CLW / EOC

Secchi Depth CLW /UQ/EPA

Water samples Chl a, TSS CLW/CMR

HPLC CLW/CMR

CLW: CSIRO Land &Water; UQ- BRG: University of Queensland Biophysical Remote Sensing Group; UQ- MB: University of Queensland, Marine Botany; QLD EPA: Queensland Environmental Protection Agency; EOC: CSIRO Earth Observation Center; CMR: CSIRO Marine Research

Definition Table 1. Description and definition of the spectral radiometric variables, where Ξ is the unit sphere (the set of all directions s with solid angle dΩ ) and µ the cosine of the zenith angle of the horizontal plane. ∂ is the partial derivative. Source: Dekker et al. (2001b).

Symbol description/definition Units/reference

L (spectral) radiance W m-1 sr-1 nm-1

∂λ∂Ω∂∂∂≡

AtQL

4

(Mobley, 1995)

µ cosine zenith angle -

µ θ≡ cos (Mobley, 1995)

Ed downwelling irradiance W m-1 nm-1

E L dd

d

= ∫ µ ( )sΞ

Ω (Mobley, 1995)

Eu upwelling irradiance W m-1 nm-1

E L du

u

= ∫ µ ( )sΞ

Ω (Mobley, 1995)

E0 scalar irradiance W m-1 nm-1

E L d0 = ∫ ( )s

Ξ

Ω (Mobley, 1995)

E0d downward scalar irradiance W m-1 nm-1

E L d0d

d

= ∫ ( )sΞ

Ω (Mobley, 1995)

E0u upward scalar irradiance W m-1 nm-1

E L d0u

u

= ∫ ( )sΞ

Ω (Mobley, 1995)


Definition Table 2 . Description and definition of the apparent optical properties. Source: Dekker et al. (2001b).

Symbol description/definition units/reference

R irradiance reflectance -

R E

E≡ u

d

(Mobley, 1995)

R( )0− subsurface irradiance reflectance -

( )( ) 0

0

d

u

==

≡zEzE

R (Gordon et al. 1975)

RL ( )s radiance reflectance -

R L

EL ( ) ( )s s= π u

d

(Walker, 1994)

Rrs ( )s remote sensing reflectance sr-1

)()(

d

urs E

LR

ss =

(Mobley, 1994)

Kd diffuse attenuation coefficient of downwelling light m-1

K

EdEdzd

d

d≡ − 1

(Mobley, 1995)

Ku diffuse attenuation coefficient of upwelling light m-1

K

EdEdzu

u

u≡ − 1

(Mobley, 1994)

Q ratio of upwelling irradiance to upwelling radiance Sr

Q

EL

( )( )

ss

≡ u

u

(Mobley, 1994)

µd downwelling average cosine -

µd

d

d ≡ E

E0

(Mobley, 1995)

µu upwelling average cosine -

µu

u

u ≡ E

E0

(Mobley, 1994)

Kdnorm normalized diffuse attenuation coefficient of

downwelling light m-1

K Kdnorm

d d= µ (Gordon, 1989)

Definition Table 3. Description and definition of the inherent optical properties. Source: Dekker et al. (2001b).

Symbol description/definition units/reference a absorption coefficient m-1

ra

r ∆ΦΦ

≡→∆ i

a

0lim

(Mobley, 1995)

β volume scattering function sr-1 m-1

∆Ω∆ΦψΦ

≡ψβ→∆Ω→∆ rr i

s

00

)(limlim)(

(Mobley, 1995)

b scattering coefficient m-1

ψψψβπ≡ ∫π

db sin)(20

(Mobley, 1995)

bb backscattering coefficient m-1


ψψψβπ≡ ∫π

π

db sin)(22

b

(Mobley, 1995)

c beam attenuation coefficient m-1

bac +≡ (Mobley, 1995)

The water constituents used in this work are chlorophyll, coloured dissolved organic matter (CDOM), and tripton, where tripton is the “lifeless” component of the suspended matter. The TSM is a measure of the suspended matter and it can be divided into the live component, i.e. phytoplankton, and into tripton the “lifeless” component, i.e. organic detritus and minerals. Chlorophyll and total suspended matter (TSM) concentrations were estimated for all sites by CSIRO Marine Research. At each site several spectro-radiometers were used to measure radiometric variables of the atmosphere, the water column and the substrates (see Table 1). The reflectance and the vertical attenuation coefficients were retrieved from these radiometric variables. The absorption coefficients were estimated on samples of each site by using the spectrophotometric techniques at CSIRO Marine Research (Clementson et al. 2001) and using a PSICAM (Point Source Integrating Cavity Absorption Meter) (Kirk 1997, Dekker et al. 2002). The backscattering of the suspended matter was estimated in situ using a HydroScat-6 (with customized band settings) according to Maffione and Dana (1997).The total absorption and the total attenuation was estimated in situ using a AC9 according to Zaneveld et al. (1990) .

The resulting descriptions of the underwater light climate at each of the 18 sample sites are presented in Section 2.2.3 and include the spectral absorption and backscattering coefficients of phytoplankton, tripton and CDOM, the subsurface irradiance reflectance and the spectral diffuse vertical attenuation coefficient of the site. This information will provide a basis for stratifying the Fitzroy River Estuary and the Port Curtis waters into a series of optical domains or water colour types.


2.2.2 Field and Image Data Pre-Processing Data and Methodology The first step in remote sensing based image processing is to ensure that the distortions introduced, or inherent in the image acquisition process, are removed to produce a corrected image spatial data set with minimal geometric, radiometric and atmospheric distortions. A reliable, robust and well-documented correction procedure is essential to produce a comparable set of images that can then be subjected to the same image processing routines, enabling their output to be used to map and measure changes in coastal ecosystem health indicators over time. Geometric correction removes spatial distortions in the image to provide it with a planimetrically correct data set, with associated projection, datum and coordinate system. The geocorrected image data can then be integrated with other spatial information; in our case this information is field survey data, other image data sets of the same area collected at different dates and other spatial data. In the case of the main data sets used for this report, the three Landsat 5 TM images (21st August 1986, 16th August 1990, 7th March 1995) were acquired in path-oriented format with no systematic corrections and then georeferenced to the Landsat 7 ETM image (24th July 2002) as a georeferenced base. The images were georeferenced to the WGS84 datum, with a UTM projection and a nominal pixel size of 25m. Sub-pixel scale Root-Mean-Square-Error values were produced for each georeferenced image. Production of a radiometrically correct image data set requires two steps, (i) radiometric correction of relative sensor response (digital numbers) to standard physical units representing the amount of light reaching the sensor, i.e., at-sensor radiance and (ii) atmospheric correction of the at sensor data to at-surface radiance or reflectance units. Conversion to at-surface radiance or reflectance values is essential to enable integration of the image data with field spectrometry. More importantly, this conversion also enables correction for air-water interface effects and the accurate estimation of in water concentrations of organic and inorganic material.

The following paragraphs describe the atmospheric correction approach further improved by the remote sensing team and applied to three Landsat 5 TM images (21st August 1986, 16th August 1990, 7th March 1995) and Landsat 7 ETM image (24th July 2002) of the Fitzroy River Estuary and Port Curtis which were selected for use as demonstration products. This includes further development of a dedicated atmospheric and air-water interface correction code c-WOMBAT-c. Initial radiometric correction to at-sensor spectral radiance was carried out by application of gain and offset coefficients from Landsat 7 Science Data Users Handbook for the 24th July 2002 image (http://ltpwww.gsfc.nasa.Gov/IAS /handbook/handbook_menu.html). For the three Landsat 5 TM images, the radiometric to at-sensor spectral radiance was carried out by applying the Canadian Centre for Remote Sensing gain and offset coefficients specified for Landsat 5. The physics of atmospheric correction of remote sensing data over waters is essentially the same as for terrestrial targets. However, for any water-body it is the signal coming from within the water body that is the desired signal. On land it is the surface reflected signal that is of interest. For water bodies, sub-surface reflected signal (subsurface irradiance reflectance R(0-)) is the required information and the surface reflected signal is considered as noise, as it is composed of the reflected component of diffuse skylight and direct sunlight impinging on the water surface. Water bodies in general reflect (as subsurface irradiance reflectance) in the range of 1 to 15% of down-welling irradiance. The majority of water bodies reflect between 2 and 6% of down-welling irradiance. Thus to obtain at least 40 levels of irradiance reflectance in the range of 2 to 6% reflectance we need a minimal accuracy of atmospheric correction to 0.1% reflectance.

The coastal Waters and Ocean MODTRAN-4 Based ATmospheric Correction (c-WOMBAT-c)

procedure was implemented in the IDL/ENVI® image processing software based on the code developed by De Haan et al. (1997). The procedure consists of: (i) a three step atmospheric inversion from at-sensor-radiance to apparent reflectance (Rapparent); and (ii) a two step inversion of


the air-water interface from apparent reflectance to subsurface irradiance reflectance (R(0-). c-WOMBAT-c applies a full MODTRAN-4 atmosphere parameterization and characterization to run the inversion. Adjacency effects (photons transferring from adjacent pixels to the one being sampled) in the imagery were corrected for using a spatially averaged radiance image generated by convolving the radiance imagery with a one square kilometre spatial weighting function.

The atmospheric parameterization for the four images was based on: radiosonde data to estimate the water column contents and the actual and 24th hour average wind speed; and Total Ozone Mapping Spectrometer - TOMS data (http://toms.gsfc.nasa.gov/ozone/ozone.html) for estimating ozone content.

The atmospheric parameterization in the three Landsat 5 TM images was established iteratively by comparing Pseudo Invariant Features (PIFs) in both the Rapparent and R(0-) images to the R apparent and R(0-) Landsat 7 ETM image of the 24th July 2002 for which the most complete parameterization was available. The parameterization that was used for the final atmospheric correction is shown in Table 2. Table 2: Parameterisation of the variables used in the atmospheric correction applied to the four Landsat images. Image Date 21st August

1986 16th August 1990

7th March 1995 24th July 2002

Sensor Landsat 5TM Landsat 5TM Landsat 5TM Landsat 7ETM Water 2.04 g 2.04 g 1.05 g 1.13 g Ozone 289 DU 278 DU 264 DU 277 DU Horizontal Visibility

45 km 70 km 75 km 100 km

Atmosphere Model

tropical tropical tropical tropical

Aerosol Model navy maritime navy maritime navy maritime navy maritime Airmass Char. 9 9 9 9 Wind Speed 2.0 m/s 7.0 m/s 2.0 m/s 0.40 m/s Wind Speed (24h avg)

4.1 m/s 4.1 m/s 4.1 m/s 4.1 m/s

To evaluate whether the atmospheric correction was correct, i.e. providing realistic and consistent reflectance values in each image, a series of targets were selected that should have exactly the same Reflectance in each image. These targets are referred to as Pseudo Invariant Features (PIFs). If the atmospheric correction has been implemented correctly the signatures of the PIFs should match across the different image dates. A feature can be considered invariant if the spectral difference between dates is less than the observed average value of the environmental noise equivalent R(0-) difference (See Brando & Dekker, 2003). The use of PIFs in close proximity to the coast was not sensible because of differences in tidal conditions and water constituents in the images. These differences in water constituents are easily visible from the images in Figures 13-20 and the tidal curves for the day of acquisition are shown in Figure 7.


Tides at Port Alma, relative to AHD (2.854 Above Station Datum)

-3

-2

-1

0

1

2

3

0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24

Time

Hei

ght (

m)

21-Aug-86 (modeled) 16-Aug-90 7-Mar-95 24-Jul-02 Figure 7: Tidal curves at Port Alma for the four Landsat acquisition dates. All Landsat image overpasses were between 09:30 and 10:00 hours. The black line indicates the approximate time of acquisition. The 21st August 1986 curve was modelled using WXTide32 (www.WXTide32.com), all other curves were measured. Readings are referred to AHD which is 2.854 m above the lowest astronomical tide for the station. Deep clear ocean targets show very low reflectance and are therefore suitable as a low reflectance PIF (although care must be taken as the differences between the ocean portion of the images tend to be relatively high as noise may have a relatively large effect). Apart from the deep clear ocean, no suitable submerged PIFs were available in the images. The quality of the atmospheric correction was therefore evaluated using several PIFs on the land in the Apparent Reflectance images. Locations that were used as PIFs include a coal pile and a large patch of forest. These two targets were chosen because their spectral signatures were similar in range to the coastal waters. Spectral signatures for these PIFs are shown in Figures 8 and 9.


Mean Apparent Refectance Coal Pile

0.0%

0.5%

1.0%

1.5%

2.0%

2.5%

3.0%

450 500 550 600 650 700 750 800 850

Wavelength (nm)

Rap

pare

nte

Aug-86 Aug-90 Mar-95 Jul-02

Figure 8: Spectral signature for a coal pile [Map: 319587.50E, 7363762.50S Meters, LL: 23o49’43.36”S, 151o13’43.62”E]. Each symbol in the lines is the centre of a Landsat Spectral Band.

Apparent Refectance forest

0.0%

1.0%

2.0%

3.0%

4.0%

5.0%

6.0%

7.0%

8.0%

450 500 550 600 650 700 750 800 850

Wavelength (nm)

Rap

pare

nt

Aug-86 Aug-90 Mar-95 2-Jul

Figure 9: Spectral signature for a forest [Map: 305159.50E, 7379262.50S Meters, LL: 23o41’13.51”S, 151o5’21.37”E]. Each symbol in the lines is the centre of a Landsat Spectral Band.


A final stage in the radiometric processing is to determine the quality of the image data, expressed as the smallest possible levels of differences in reflected light that can be accurately and reliable measured. Establishing this information is critical as it will provide the lower bound to the magnitude of changes in the water constituents’ concentrations that can be estimated from image data sets. This is done by estimating the environmental noise equivalent radiance or reflectance. To determine this precision and accuracy in the estimate of an environmental variable with any remote sensing sensor, it is necessary to estimate the sensitivity of the sensor-atmosphere-air-water interface system for detecting changes in reflectance. The environmental noise equivalent R(0-) difference NE∆R(0-) and the environmental noise equivalent Rapp difference NE∆Rapp are image-based measures of such changes (Brando and Dekker, 2003). NE∆Rapp and NE∆R(0-) are dependent on the instrument signal to noise ratio (SNR) with added influences of noise in the image data such as atmospheric variability, as well as the effects of the adopted atmospheric correction procedure and parameterization. NE∆R(0-) also incorporates the effects of the air water interface (with swell, wave and wavelet induced reflections), and refraction of diffuse and direct sky and sunlight. The NE∆R(0-) is calculated from the atmospherically and air-water interface corrected image R(0-) according to Brando and Dekker (2003) and Wettle et al. ( 2004):

NE∆R(0-) =σ(R0-) (1) Where,

σ(R) is the standard deviation in each band over an “as homogeneous as possible” area of optically deep water within the image. The size of the uniform area can be determined by increasing the number of pixels (1, 3x3, 5x5, 7x7, etc.) until σ(R) reaches a first asymptotic limit. Care has to be taken that whilst increasing the size of the uniform area no actual water body heterogeneities are included in the sampled area. The sample sites used for calculation of the NE∆R(0-)E were located in the north-eastern portion of the Fitzroy Estuary and Port Curtis images, where the ocean target can be assumed quasi-homogeneous. Resulting values of NE∆RApp and NE∆R(0-) for the Landsat 7 ETM and the three Landsat 5 TM images used in this project are shown in Figures 10 and 11. The observed NE∆RApp

is almost spectrally flat between 550 to 850 nm with an average value of 0.3% – 0.4%. The observed NE∆R(0-)E is almost spectrally flat between 550 to 850 nm with an average value of 0.8%. The Landsat image of 1986 shows a higher noise level both in NE∆RApp and NE∆R(0-)

terms. Ideally, distinguishable levels of 0.1 - 0.2 % reflectance are required, hence attention needs to focus on how to decrease NE∆R(0-). This can be achieved by spatially binning the image (with an ensuing loss of spatial resolution) or by applying a low-pass filter on the image (increasing the covariance of the pixels). In order to preserve the 30 m Landsat ETM resolution we applied the low pass filter method. NE∆R(0-) of the 5x5 low pass filtered image is almost spectrally flat between 450 to 800 nm with an average value of 0.3% (see Figure 12). The environmental noise equivalent for the 1986 image remains high at 0.5% average, even after the low pass filtering. This level is adequate for optical water quality remote sensing purposes. The level of match between the PIFs across the different dates of images (Figures 10 and 12) is within the observed average value of NE∆RApp of 0.3% – 0.4% indicating the atmospheric correction was sufficiently successful as the resultant difference was within the noise limits of the Landsat data. From this, it can be concluded that the atmospheric correction was performed adequately and that no image is over or under corrected. Figure 10: Noise equivalent Rapp for the four Landsat images.


Noise equivalent Apparent Reflectance for four Landsat images

0.0%

0.1%

0.2%

0.3%

0.4%

0.5%

0.6%

450 500 550 600 650 700 750 800 850

Wavelength [nm]

NE∆

Rap

pare

nt



Figure 11: Noise equivalent R(0-) for the four Landsat images. Noise equivalent R(0-) for four Landsat images

0.0%

0.2%

0.4%

0.6%

0.8%

1.0%

1.2%

450 500 550 600 650 700 750 800 850

Wavelength [nm]

NE∆

R(0

-)

Aug-86 Aug-90 Mar-95 Jul-02 Figure 12: Noise equivalent R(0-) for the 5x5 low pass filtered Landsat images.

Noise equivalent R(0-) for four 5x5 low pass filtered Landsat images

0.0%

0.1%

0.2%

0.3%

0.4%

0.5%

0.6%

0.7%

450 500 550 600 650 700 750 800 850

Wavelength [nm]

NE∆

R(0

-)



Figures 13-20 present the pseudo true colour composites of the atmospherically corrected reflectance images (bands 1, 2 and 3 displayed as blue, green and red) for the 4 dates (21st August 1986, 16th August 1990, 7th March 1995 and 24th July 2002) for both Port Curtis and Fitzroy Estuary. The colour coding was fixed to the same range for all the images to enable a direct visual comparison. The two sub-scenes of the Port Curtis imagery acquired at high tide (Figures 13 and 19; 21st August 1986 and 24th July 2002) show a similar pattern. Dark green colours in figure 13 indicate higher organic matter concentrations. The white linear strips in the ocean in figures 17 and 18 are likely to be caused by Trichodesmium blooms. All Fitzroy Estuary images (14, 16,18 and 20) show the major part of the suspended matter curving towards the East, whilst there is a noticeable band of suspended matter (but at lower concentrations travelling North along the shore line. Figure 13: Landsat 5 Thematic Mapper image of Port Curtis captured at 09:30 on 21st August 1986, true colour composite (bands 1, 2 and 3 displayed as blue, green and red) at high tide. Figure 14: Landsat 5 Thematic Mapper image of Fitzroy Estuary captured at 09:30 on 21st August 1986, true colour composite (bands 1, 2 and 3 displayed as blue, green and red) at high tide. The white colour in the Fitzroy Estuary mouth is due to comparatively high suspended sediment loads as compared to the other dates (Figs 16, 18 and 20) as the colour coding was fixed to the range of the other images. Figure 15: Landsat 5 Thematic Mapper image of Port Curtis captured at 09:30 on 16th August 1990, true colour composite (bands 1, 2 and 3 displayed as blue, green and red). Figure 16: Landsat 5 Thematic Mapper image of Fitzroy Estuary captured at 09:30 on 16th August 1990, true colour composite (bands 1, 2 and 3 displayed as blue, green and red). Figure 17: Landsat 5 Thematic Mapper image of Port Curtis captured at 09:30 on 7th March 1995, true colour composite (bands 1, 2 and 3 displayed as blue, green and red). Figure 18: Landsat 5 Thematic Mapper image of Fitzroy Estuary captured at 09:30 on 7th March 1995, true colour composite (bands 1, 2 and 3 displayed as blue, green and red). The white linear strips in the ocean are likely to be caused by Trichodesmium blooms. Figure 19: Landsat 7 Enhanced Thematic Mapper image of Port Curtis captured at 09:45 on 24th July 2002, true colour composite (bands 1, 2 and 3 displayed as blue, green and red). Figure 20: Landsat 7 Enhanced Thematic Mapper image of Fitzroy Estuary captured at 09:45 on 24th July 2002, true colour composite (bands 1, 2 and 3 displayed as blue, green and red).


Figure 13: Landsat 5 Thematic Mapper image captured at 09:30 on 21st August 1986, true colour composite (bands 1, 2 and 3 displayed as blue, green and red) at high tide. Dark green colours indicate higher organic matter concentrations. Light green colours indicate high TSM loads.


Figure 14: Landsat 5 Thematic Mapper image captured at 09:30 on 21st August 1986, true colour composite (bands 1,2 and 3 displayed as blue, green and red) at high tide. The white colour in the Fitzroy Estuary mouth is due to comparatively high suspended sediment loads as compared to the other dates (Figs 16, 18 and 20) as the colour coding was fixed to the range of the other images.


Figure 15: Landsat 5 Thematic Mapper image captured at 09:30 on 16th August 1990, true colour composite (bands 1, 2 and 3 displayed as blue, green and red).


Figure 16: Landsat 5 Thematic Mapper image captured at 09:30 on 16th August 1990, true colour composite (bands 1,2 and 3 displayed as blue, green and red).


Figure 17: Landsat 5 Thematic Mapper image captured at 09:30 on 7th March 1995, true colour composite (bands 1,2 and 3 displayed as blue, green and red).The white linear to wavy strips in the ocean are likely to be caused by Trichodesmium blooms.


Figure 18: Landsat 5 Thematic Mapper image captured at 09:30 on 7th March 1995, true colour composite (bands 1, 2 and 3 displayed as blue, green and red). The white wavy strips in the ocean are likely to be caused by Trichodesmium blooms.


Figure 19: Landsat 7 Enhanced Thematic Mapper image captured at 09:45 on 24th July 2002, true colour composite (bands 1, 2 and 3 displayed as blue, green and red), at high tide. This image shows a similar pattern to Figure 13 for the 1986 image.


Figure 20: Landsat 7 Enhanced Thematic Mapper image captured at 09:45 on 24th July 2002, true colour composite (bands 1,2 and 3 displayed as blue, green and red).


2.2.3 Underwater Light Climate – Bio-optical Model Development

Data and Methodology Modified from Moreton Bay Report (Phinn and Dekker, 2005) As noted in the previous sections, measurement of IOP’s and AOP’s (Definition Tables 2-3) is an essential stage if a physics-based approach is to be taken for estimating quantities of organic and inorganic material in the water column and accurately mapping substrate features (Dekker et al. 2001b, Kirk, 1994). A key limitation of the majority of past approaches estimating biophysical properties of water bodies has been their empirical approach which restricts their algorithm to the image date and field data they were developed for. Examples of successful application of the physics based bio-optical modelling and inversion approach include: Dekker et al. (1999, 2001a and b), Brando and Dekker (2003), and Lee et al. (2001). There are two modelling approaches to determine the relationship between the inherent and apparent optical properties of the water column and the water constituents: the analytical modelling and the radiative transfer modelling approach. Radiative transfer equations can be solved by several numerical methods. They include Monte-Carlo ray tracing, invariant imbedding methods, discrete ordinates solution methods and others (Mobley, 1994). All radiative transfer numerical models compute radiance distributions and related quantities (irradiance, reflectance, diffuse attenuation functions, etc.) in the water column as a function of the water absorption and scattering properties, the sky and air-water interface conditions and the bottom boundary conditions. The analytical model is a simplification of the full radiative transfer equations, based on a set of given assumptions. Analytical models have the advantage that, due to their relative simplicity, they can be solved quickly. This is important in a remote sensing application where a model must be evaluated at every pixel of an image. An analytical model for water quality retrieval relates the subsurface irradiance reflectance R(0-), to the water constituent concentrations. Several models for coastal and inland waters were investigated as summarized by Dekker et al. (2001b). They are very similar to an analytical solution of the irradiance transfer equations given by Aas (1997):

ud11 ,)0(

µµ+=

+=− f

babfR

b

b , (2)

where a is the total absorption coefficient, bb is the total backscattering coefficient; f is the anisotropy factor of the downwelling light field; and µd and µu are the average cosines of the downwelling and upwelling light field respectively. All variables in the equation are spectra but the wavelength dependency was omitted for readability. In the general case, the absorption and backscattering coefficients are the sum of the contributions of N constituents and a constant coefficient for pure water:

∑=

∗+=N

Caaa1j

jj0 ; ∑=

∗+=N

bbb Cbbb1j

jj0 , (3)

where a0 and bb0 are usually determined by the absorption and backscattering of pure water, a*j

and bb*j are the specific inherent optical properties (SIOPs) of jth constituent with concentration Cj.

Since the water constituents used in this work are chlorophyll, coloured dissolved organic matter, and tripton, the total absorption coefficient can be written as:

*TRCDOM

*CHLw TRCDOMCHL aaaaa ⋅+⋅+⋅+= ∗ (4)

The underwater light-climate model describes how light is absorbed and scattered by components of the water column. The measurement of the variability of the SIOPs will reflect how the river and ocean waters interact in the Fitzroy River Estuary and the Port Curtis. Estimation of SIOPs was a major activity during the field sampling in of January 16-17th 2002 and June 21-24th 2002. During the two field campaigns 18 sample sites were visited (Table 3).


The absorption coefficient of CDOM (aCDOM) was estimated using a dual beam spectrophotometer (Clementson et al. 2001). The spectral absorption aCDOM was described as a normalized spectrum with an exponential slope:

( ) ( ) ( )( )00CDOMCDOM expCDOM λλλλ −−⋅= ∗ Saa , (5) and estimated from a non linear regression of the log-transformed absorption versus wavelength. λ0 was set at 440 nm, CDOM is the aCDOM at 440 nm thus a*CDOM (440) is 1, and S varies at the sampling stations. The particulate (algal and detrital) absorption was measured using a dual beam spectrophotometer with integrating sphere (see Clementson et al. 2001 for further details). The spectral absorption by tripton (aTR) was estimated after a pigment extraction. The aTR was retrieved following the same method as for aCDOM :

( ) ( ) ( )( )00*TRTR expTR λλλλ −−⋅= Saa , (6)

where λ0 was set at 440 nm, while a*TR (440) and S vary at the sampling stations.

The spectral absorption of phytoplankton (aph) was estimated subtracting the spectral absorption by tripton (aTR) from the total particulate absorption. Because of the low concentration of chlorophyll and high concentration of particulate matter in these turbid waters the estimate of specific spectral absorption of phytoplankton (a*ph) was not considered reliable. Thus the a*ph from another Australian estuarine site with a similar chlorophyll concentration range was adopted for the optical modelling.

The backscattering of the suspended matter was estimated in situ using a HydroScat-6 (with customized band settings) according to Maffione and Dana (1997). The backscattering is due to a small component from the algae (as algal concentrations are relatively low) and a larger component due to the tripton. As a separate measurement of the backscattering of phytoplankton cells was not feasible, the assumption was made that 1 µg l-1 CHL is approximately equal to 0.07 mg l-1 TSM (Dekker et al. 2001b). As we know the total backscattering, it is then possible to approximate the contribution of backscattering due to tripton and the contribution by phytoplankton and thus retrieve a specific backscattering for the two components. Several spectro-radiometers were used to measure radiometric variables of the atmosphere, the water column and the substrates (see Table 1). The reflectance and the vertical attenuation coefficients were retrieved from these radiometric variables. Validation of the SIOPs retrieval was achieved by comparing the simulated reflectance spectra with the spectra measured in situ each site. Numerical radiative transfer model (Hydrolight, Mobley, 1994) simulations as well as analytical model simulations were parameterized with all the relevant information for each in situ site: the SIOPs, the measured concentrations and the in situ measured downwelling irradiance. These simulations were used to simulate the reflectance R(0-) and the Vertical attenuation coefficient of the downwelling light Kd. Achievement of an overall optical closure for 500-700 nm range of the numerical radiative transfer model with the in situ measurements confirms the parameterization and estimate of the SIOPs. Results and Discussion During the wet season campaign, apparent and inherent optical properties were measured at six locations along the river and the estuary. Tables 3 and 4 report on conditions measured at the two sampling sites on Wednesday January 16th 2002 (upstream of the Fitzroy River Cut-through and in the Fitzroy River Mouth at the middle of Mud Island) and four sites on Thursday January 17th 2002 (in Keppel Bay near Sea Hill, in the Fitzroy River Mouth east of Mud Island, in the Fitzroy River Mouth near Egg Island and in the Fitzroy River at Rocky Point (~ 10 km upstream the mouth)) in different tidal conditions. The flood event flushed out saltwater from the river but the freshwater had not reached the bay. The tidal front going in and out at the river mouth presented an interface


between brown (chocolate) waters with a Secchi depth of few cm and a mixed brackish greenish water with a Secchi disk of 40-80 cm. During the dry season campaign we visited three sampling sites on Friday June 21st (Fitzroy River upstream the Barrage at the Water Treatment Plant; Fitzroy River just upstream the Barrage, Fitzroy River 5 km downstream the Barrage at Devil's Elbow), four sites in the Fitzroy Estuary in different tidal conditions on Saturday June 22 (in the middle of Keppel Bay, in the Fitzroy River Mouth, in the Fitzroy River at Pirate Point in the Fitzroy River Mouth in the Southern Channel), and five sites in Port Curtis in different tidal conditions on Sunday 23 and Monday 24th June (in Port Curtis Northern edge of facing island, in Port Curtis at Quoin Island, in Port Curtis at the outlet of South Trees Inlet, and in Port Curtis at Wiggins Island). Table 4 presents measured organic and inorganic matter concentrations, while Table 5 presents the IOPs and SIOPs parameterization for all of the 18 sites. In Table 5, the range of the IOPs and SIOPs are summarized. The measured range of the aTR slope and of the aCDOM slopes (Figures 21-23) are similar to those reported by Bukata et al. (1995) and Roesler et al. (1989) for inland and coastal waters. Representation of the underwater light climate at 18 locations throughout Port Curtis and the Fitzroy Estuary was provided by the measurement of two IOPs spectra: absorption; backscattering; and two AOPs spectra: vertical attenuation and subsurface irradiance reflectance R(0-). These quantities represent key components of hydrologic radiative transfer equations (Dekker et al. 2001b) that quantify how much sun and sky light is absorbed and scattered, and transmitted to an imaging sensor, when light interacts within a water body containing organic and inorganic material. The subsurface irradiance reflectance R(0-) is the physical unit which all of the Landsat ETM data sets have been corrected to. Hence, each of Figures 24th through 41 provides a quantitative definition for 18 locations in the Port Curtis and Fitzroy Estuary areas of how much incident light is absorbed and scattered, and what proportion of these interactions are due to the various organic and inorganic constituents of the water column. In simpler terms these measurements quantify the features responsible for the different water colours and optical domains identified in the previous section.

For Figures 24-41 the following variables are graphed to present the underwater light climate at each location: The order of graphics in each figure is starting from the top right and moving clockwise:

- The tidal stage at which samples and measurements were taken superimposed on a picture of the water at the site.

- Spectral absorption by phytoplankton, tripton, water and CDOM This graph indicates which constituents of the water column are absorbing specific wavelengths of light. Water that is phytoplankton rich will absorb highly in blue and red due to chlorophyll and not as much in green, hence its green colour, as it is mainly green light that is backscattered. - Spectral backscattering by water, phytoplankton and tripton; This graph indicates which constituents are scattering most sun- and skylight back to the sensor. Water bodies with a high suspended sediment load (often itself yellowish-brown coloured) will have high tripton concentrations and reflect more strongly in the yellow-red wavelengths. Clear water bodies scatter predominantly blue light, hence their blue colour. - Spectral subsurface irradiance reflectance and the vertical attenuation

coefficient(Kd); This plot defines the colour of sun- and skylight reflected from the sample site and the wavelengths at which most absorption takes place(=~highest Kd). - Environmental conditions at the time of underwater light climate measurements.


Figure 21: A scatter plot of the optical parameterisation for the tripton absorption: it relates the slope (the colour in terms of hue) to the intensity per concentration unit. Where groupings occur, a regional/local bio-optical parameterisation exists. The outlier (0.38 ~ -0.015) represents a closed-off arm at Pirate Point in the Fitzroy Estuary and thus represents a ponding optical parameterisation.

-0.016

-0.015

-0.014

-0.013

-0.012

-0.011

-0.010

-0.009

-0.0080.0000 0.0500 0.1000 0.1500 0.2000 0.2500 0.3000 0.3500 0.4000

aTR* @ 440 nm

aTR

Slo

pe FRE dryPC dryFRE wet

Figure 22: A scatter plot of the optical parameterisations for the CDOM absorption: it relates the slope (the colour in terms of hue) to the intensity of light absorption. Where groupings occur a regional/local bio-optical parameterisation exists indicative of freshwater to ocean water gradients in CDOM pools.

-0.022

-0.021

-0.020

-0.019

-0.018

-0.017

-0.016

-0.0150.000 0.500 1.000 1.500 2.000 2.500 3.000 3.500

aCDOM @ 440 nm

aCD

OM

Slo

pe

FRE dryPC dryFRE wet


Figure 23: A scatter plot of the two optical parameterisations for the tripton and the CDOM absorption: it relates the slopes (the colour in terms of hue) of these two optically active components. Where groupings or a pattern occur a regional/local bio-optical parameterisation exists.

-0.022

-0.021

-0.020

-0.019

-0.018

-0.017

-0.016

-0.015-0.016 -0.015 -0.014 -0.013 -0.012 -0.011 -0.010 -0.009 -0.008

aCDOM Slope

aTR

Slo

pe FRE dryPC dryFRE wet


Table 3: Port Curtis and Fitzroy Estuary – underwater light climate sample site characteristics for January and June 2002. Site Site description Lat (S) Lon (E) date Wet/Dry

season Depth (m)

Secchi (m)

Weather Cloud Coverage

Substrate type

FE_CUT Fitzroy River at Cut-through 23.49325 150.66511 16-Jan-02 Wet 6.7 < 5 cm cloudy sky / overcast

>75% not visible

FE_MUD Fitzroy River Mouth - Mud Island 23.53436 150.84383 16-Jan-02 Wet 6.5 0.8 cloudy sky/light rain

high not visible

FE_SEA Keppel Bay (near Sea Hill) 23.47700 150.96175 17-Jan-02 Wet 16.0 1.0 cloudy/ patchy / light rain /sunny

>75% not visible

FE_MUDE Fitzroy River Mouth - east of Mud Island

23.53898 150.85400 17-Jan-02 Wet 5.5 0.8 light rain >85% not visible

FE_EGG Fitzroy River Mouth - Egg Island 23.53515 150.85700 17-Jan-02 Wet 7.6 0.4 cloudy/ overcast

>85% not visible

FE_ROCK Fitzroy River at Rocky Point (~ 10 km upstream the mouth)

23.50420 150.78983 17-Jan-02 Wet 5.3 0.1 - 0.2 cloudy sunny patches

40-70% not visible

Note change to decimal

notation

FR_WTP Fitzroy River upstream the Barrage at the Water Treatment Plant

23.31800 150.48123 21-Jun-02 Dry 12.0 0.1 fine/breeze 10-60% not visible

FR_BAR Fitzroy River just upstream the Barrage 23.35922 150.49346 21-Jun-02 Dry 13.0 0.1 overcast 90% not visible FR_DEV Fitzroy River 5 km downstream the

Barrage at Devil's Elbow 23.37709 150.55042 21-Jun-02 Dry 10.0 0.1 overcast 100% not visible

FE_KEB Keppel Bay (in the middle) 23.42364 150.93512 22-Jun-02 Dry 8.7 0.7 fine and sunny

0% not visible

FE_RMO Fitzroy River Mouth 23.54409 150.88423 22-Jun-02 Dry 4.7 N/A fine and sunny

0% not visible

FE_PIR Fitzroy River at Pirate Point 23.50495 150.63758 22-Jun-02 Dry 2.1 0.35 fine and sunny

0% not visible

FE_SOC Fitzroy River Mouth - Southern Channel 23.54409 150.88423 22-Jun-02 Dry 4.5 N/A fine and sunny

0% not visible

PC_OAK Northern edge of Facing Island at "the

OAKES" 23.76136 151.33020 23-Jun-02 Dry 6.2 2.0 fine 0% Sandy / not

visible PC_QUO SW of Quoin Island 23.81569 151.28447 23-Jun-02 Dry 5.5 1.1 fine/ slight

haze 5% Muddy

(gray/black)

PC_QIS SW of Quoin Island 23.81640 151.28339 23-Jun-02 Dry 2.6 sunny/windy 20% Muddy PC_STI South Trees Inlet (upstream causeway

of QLD alumina) 23.86066 151.30442 23-Jun-02 Dry 2.9 0.8 variable 40-50% not visible

PC_WIG Wiggins Island (mouth of Calliope River)

23.81445 151.19567 24-Jun-02 Dry 1.7 1.0 mainly sunny

30-40% Mud / seagrasses


Table 4: Concentration of the organic and inorganic water column constituents for the 18 sites sampled in January and June 2002; a tripton to TSM ratio close to 1 indicates a dominance of TSM and a relatively very low chlorophyll contents.

Site

Site description TSM (mg/L)

Chl a (ug/L)

CHL as DW(mg/L)

Tripton (mg/L) Tripton/TSM

FE_CUT Fitzroy River at Cut-through 346.0 4.0 0.28 345.8 0.999 FE_MUD Fitzroy River Mouth - Mud Island 8.9 13.6 0.95 7.9 0.893 FE_SEA Keppel Bay (near Sea Hill) 22.7 2.2 0.15 22.5 0.993 FE_MUDE Fitzroy River Mouth - east of Mud Island 18.7 6.8 0.48 18.2 0.974 FE_EGG Fitzroy River Mouth - Egg Island 24.6 3.7 0.26 24.3 0.989 FE_ROCK Fitzroy River at Rocky Point (~ 10 km upstream the mouth) 74.6 4.4 0.31 74.3 0.996 FR_WTP Fitzroy River upstream the Barrage at the Water Treatment Plant 122.6 4.0 0.28 122.3 0.998 FR_BAR Fitzroy River just upstream the Barrage 95.2 2.6 0.19 95.0 0.998 FR_DEV Fitzroy River 5 km downstream the Barrage at Devil's Elbow 78.1 3.4 0.24 77.9 0.997 FE_KEB Keppel Bay (in the middle) 25.0 1.3 0.09 24.9 0.996 FE_RMO Fitzroy River Mouth 60.7 3.4 0.24 60.5 0.996 FE_PIR Fitzroy River at Pirate Point 27.0 2.6 0.18 26.8 0.993 FE_SOC Fitzroy River Mouth - Southern Channel 12.1 0.9 0.06 12.0 0.995 PC_OAK Northern edge of Facing Island at "the OAKES" 10.8 0.3 0.02 10.8 0.998 PC_QUO SW of Quoin Island 12.2 0.8 0.06 12.1 0.996 PC_QIS SW of Quoin Island 10.5 1.0 0.06 10.4 0.994 PC_STI South Trees Inlet (upstream causeway of QLD alumina) 10.3 1.2 0.08 10.2 0.992 PC_WIG Wiggins Island (mouth of Calliope River) 12.4 0.9 0.06 12.3 0.995


Table 5: Parameterization of the IOPs and SIOPs from the 18 sites sampled in January and June 2002

Site aCDOM @440 nm aCDOM Slope

aTR @440nm aTR Slope

bb P @546nm

bb P Gamma

bb TR @546nm aTR* bb TR * bb PHY *

FE_CUT 2.098 -0.0168 41.166 -0.0137 0.1191 FE_MUD 0.699 -0.0180 0.315 -0.0113 0.098 1.047 0.087 0.0396 0.0110 0.0008FE_SEA 0.076 -0.0159 0.212 -0.0118 0.127 0.820 0.126 0.0094 0.0056 0.0004FE_MUDE 0.378 -0.0176 0.331 -0.0115 0.188 1.076 0.183 0.0182 0.0101 0.0007FE_EGG 0.474 -0.0174 0.462 -0.0111 0.0190 FE_ROCK 1.397 -0.0171 3.427 -0.0118 0.271 1.293 0.270 0.0461 0.0036 0.0003 FR-WTP 3.108 -0.0166 11.699 -0.0139 0.0956 FR-BAR 3.045 -0.0166 9.994 -0.0136 0.1052 FR-DEV 2.885 -0.0167 28.249 -0.0149 0.3628 FE-KEB 0.089 -0.0205 0.949 -0.0106 0.365 0.979 0.364 0.0381 0.0146 0.0010FE-RMO 0.310 -0.0178 1.692 -0.0107 1.948 0.914 1.940 0.0280 0.0321 0.0022FE-PIR 1.390 -0.0178 1.140 -0.0121 2.919 0.341 2.900 0.0425 0.1081 0.0076FE-SOC 0.448 -0.0182 1.328 -0.0112 0.1103 PC_OAK 0.104 -0.0186 0.291 -0.0118 0.090 1.153 0.089 0.0270 0.0083 0.0006PC_QUO 0.094 -0.0213 0.481 -0.0103 0.146 1.226 0.145 0.0396 0.0119 0.0008PC_QIS 0.172 -0.0181 0.632 -0.0111 0.148 1.087 0.147 0.0605 0.0141 0.0010PC_STI 0.252 -0.0173 0.451 -0.0115 0.139 1.092 0.138 0.0441 0.0135 0.0009PC_WIG 0.271 -0.0159 0.362 -0.0128 0.157 1.124 0.156 0.0294 0.0126 0.0009


Figure 24: Underwater light climate parameters at Fitzroy River upstream of the Barrage at the Water Treatment Plant (site FR_WTP), 21/06/2002 11:30 (clockwise from top right): tidal stage at sample time, spectral absorption, spectral backscattering, subsurface irradiance reflectance, and vertical diffuse attenuation, and sample site characteristics. Fitzroy River upstream the Barrage at the Water Treatment Plant

Date 21 June 2002Time (AEST) 11:30Latitude 23.31800Longitude 150.48123Bottom Depth (m) 12Secchi Depth (m) 0.1Cloud cover 10-60%Weather conditions fine/breezeTSS (mg/L) 122.6Chl a (µg/L) 4.0substrate not visible

N

E

S

W

Kilometers0 5 10 20

0.0%

0.2%

0.4%

0.6%

0.8%

1.0%

1.2%

1.4%

400 450 500 550 600 650 700 750wavelength [nm]

Subs

urfa

ce Ir

radi

ance

Ref

lect

ance

R

(0-)

0.0

2.0

4.0

6.0

8.0

10.0

12.0

Verti

cal a

ttenu

atio

n K

d [m

-1]

R(0-) Kd

0.01.02.03.04.05.06.0

21:00 0:00 3:00 6:00 9:00 12:00 15:00 18:00 21:00

Tide

[m]

∗

0.000

0.020

0.040

0.060

0.080

0.100

0.120

0.140

0.160

0.180

450 500 550 600 650 700 750

wavelength [nm]

a*ph

, a*t

r

0.00.10.20.30.40.50.60.70.80.91.0

a*cd

om ,

aw

a*phy a* tr aw a*cdom

0.000

0.001

0.001

0.002

0.002

0.003

0.003

0.004

0.004

0.005

450 500 550 600 650 700 750

wavelength [nm]

bb*p

h, b

b*tr

, bbw

bbw bb*phy bb* tr Figure 25: Underwater light climate parameters at Fitzroy River just upstream of the Barrage (site FR_BAR), 21/06/2002 12:55 (clockwise from top right): tidal stage at sample time, spectral absorption, spectral backscattering, subsurface irradiance reflectance, and vertical diffuse attenuation, and sample site characteristics.

Fitzroy River Estuary Underwater Light ClimateFitzroy River just upstream the Barrage

Date 21 June 2002Time (AEST) 12:55Latitude 23.35922Longitude 150.49346Bottom Depth (m) 13Secchi Depth (m) 0.1Cloud cover 90%Weather conditions overcastTSS (mg/L) 95.2Chl a (µg/L) 2.6substrate not visible

N

E

S

W

Kilometers0 5 10 20

0.0%

1.0%

2.0%

3.0%

4.0%

5.0%

6.0%

400 450 500 550 600 650 700 750wavelength [nm]

Subs

urfa

ce Ir

radi

ance

Ref

lect

ance

R

(0-)

0.0

1.0

2.0

3.0

4.0

5.0

6.0

7.0

8.0

Verti

cal a

ttenu

atio

n K

d [m

-1]

R(0-) Kd

0.01.02.03.04.05.06.0

21:00 0:00 3:00 6:00 9:00 12:00 15:00 18:00 21:00

Tide

[m]

∗

0.0000.0200.0400.0600.0800.1000.1200.1400.1600.1800.200

450 500 550 600 650 700 750

wavelength [nm]

a*ph

, a*t

r

0.00.10.20.30.40.50.60.70.80.91.0

a*cd

om ,

aw


0.000

0.001

0.001

0.002

0.002

0.003

0.003

0.004

0.004

0.005

450 500 550 600 650 700 750

wavelength [nm]

bb*p

h, b

b*tr

, bbw

bbw bb*phy bb* tr


Figure 26: Underwater light climate parameters at Fitzroy River 5 km downstream the Barrage at Devil's Elbow (site FR_DEV), 21/06/2002 14:00 (clockwise from top right): tidal stage at sample time, spectral absorption, spectral backscattering, subsurface irradiance reflectance, and vertical diffuse attenuation, and sample site characteristics.

Fitzroy River Estuary Underwater Light ClimateFitzroy River 5km downstream the Barrage at Devil's Elbow

Date 21 June 2002Time (AEST) 14:00Latitude 23.37709Longitude 150.55042Bottom Depth (m) 10Secchi Depth (m) 0.05Cloud cover 100%Weather conditions overcastTSS (mg/L) 78.1Chl a (µg/L) 3.4substrate not visible

N

E

S

W

Kilometers0 5 10 20

0.0%

1.0%

2.0%

3.0%

4.0%

5.0%

6.0%

7.0%

400 450 500 550 600 650 700 750wavelength [nm]

Subs

urfa

ce Ir

radi

ance

Ref

lect

ance

R

(0-)

0.0

0.5

1.0

1.5

2.0

2.5

3.0

3.5

4.0

4.5

5.0

Verti

cal a

ttenu

atio

n K

d [m

-1]

R(0-) Kd

0.01.02.03.04.05.06.0

21:00 0:00 3:00 6:00 9:00 12:00 15:00 18:00 21:00

Tide

[m]

∗

0.000

0.100

0.200

0.300

0.400

0.500

0.600

0.700

450 500 550 600 650 700 750

wavelength [nm]

a*ph

, a*t

r

0.00.10.20.30.40.50.60.70.80.91.0

a*cd

om ,

aw


0.000

0.001

0.001

0.002

0.002

0.003

0.003

0.004

0.004

0.005

450 500 550 600 650 700 750

wavelength [nm]

bb*p

h, b

b*tr

, bbw

bbw bb*phy bb* tr Figure 27: Underwater light climate parameters at Keppel Bay (middle) (site FE_KEB), 22/06/2002 10:00 (clockwise from top right): tidal stage at sample time, spectral absorption, spectral backscattering, subsurface irradiance reflectance, and vertical diffuse attenuation, and sample site characteristics.

Fitzroy River Estuary Underwater Light ClimateKeppel Bay (in the middle)

Date 22 June 2002Time (AEST) 10:00Latitude 23.42364Longitude 150.93512Bottom Depth (m) 8.7Secchi Depth (m) 0.7Cloud cover 0%Weather conditions fine & sunnyTSS (mg/L) 25.0Chl a (µg/L) 1.3substrate not visible

N

E

S

W

Kilometers0 5 10 20

0.0%

2.0%

4.0%

6.0%

8.0%

10.0%

12.0%

14.0%

400 450 500 550 600 650 700 750wavelength [nm]

Subs

urfa

ce Ir

radi

ance

Ref

lect

ance

R

(0-)

0.0

0.5

1.0

1.5

2.0

2.5

Verti

cal a

ttenu

atio

n K

d [m

-1]

R(0-) Kd

0.01.02.03.04.05.06.0

0:00 3:00 6:00 9:00 12:00 15:00 18:00 21:00

Tide

[m]

∗

0.000

0.010

0.020

0.030

0.040

0.050

0.060

0.070

450 500 550 600 650 700 750

wavelength [nm]

a*ph

, a*t

r

0.00.10.20.30.40.50.60.70.80.91.0

a*cd

om ,

aw


0.000

0.005

0.010

0.015

0.020

0.025

450 500 550 600 650 700 750

wavelength [nm]

bb*p

h, b

b*tr

, bbw

bbw bb*phy bb* tr


Figure 28: Underwater light climate parameters at Fitzroy River Mouth (site FE_RMO), 22/06/2002 11:30 (clockwise from top right): tidal stage at sample time, spectral absorption, spectral backscattering, subsurface irradiance reflectance, and vertical diffuse attenuation, and sample site characteristics.

Fitzroy River Estuary Underwater Light ClimateFitzroy River Mouth

Date 22 June 2002Time (AEST) 11:30Latitude 23.54409Longitude 150.88423Bottom Depth (m) 4.7Secchi Depth (m) N/ACloud cover 0%Weather conditions fine & sunnyTSS (mg/L) 60.7Chl a (µg/L) 3.4substrate not visible

N

E

S

W

Kilometers0 5 10 20

0.0%

2.0%

4.0%

6.0%

8.0%

10.0%

12.0%

400 450 500 550 600 650 700 750wavelength [nm]

Subs

urfa

ce Ir

radi

ance

Ref

lect

ance

R

(0-)

0.0

1.0

2.0

3.0

4.0

5.0

6.0

7.0

8.0

Verti

cal a

ttenu

atio

n K

d [m

-1]

R(0-) Kd

0.01.02.03.04.05.06.0

0:00 3:00 6:00 9:00 12:00 15:00 18:00 21:00

Tide

[m]

∗

0.000

0.005

0.010

0.015

0.020

0.025

0.030

0.035

0.040

0.045

450 500 550 600 650 700 750

wavelength [nm]

a*ph

, a*t

r

0.00.10.20.30.40.50.60.70.80.91.0

a*cd

om ,

aw


0.000

0.005

0.010

0.015

0.020

0.025

0.030

0.035

0.040

0.045

450 500 550 600 650 700 750

wavelength [nm]

bb*p

h, b

b*tr

, bbw

bbw bb*phy bb* tr Figure 29: Underwater light climate parameters at Fitzroy River at Pirate Point (site FE_PIR), 22/06/2002 13:30 (clockwise from top right): tidal stage at sample time, spectral absorption, spectral backscattering, subsurface irradiance reflectance, and vertical diffuse attenuation, and sample site characteristics.

Fitzroy River Estuary Underwater Light ClimateFitzroy River at Pirate Point

Date 22 June 2002Time (AEST) 13:30Latitude 23.50495Longitude 150.63758Bottom Depth (m) 2.1Secchi Depth (m) 0.35Cloud cover 0%Weather conditions fine & sunnyTSS (mg/L) 27.0Chl a (µg/L) 2.6substrate not visible

N

E

S

W

Kilometers0 5 10 20

0.0%

2.0%

4.0%

6.0%

8.0%

10.0%

12.0%

14.0%

16.0%

400 450 500 550 600 650 700 750wavelength [nm]

Subs

urfa

ce Ir

radi

ance

Ref

lect

ance

R

(0-)

0.0

0.5

1.0

1.5

2.0

2.5

3.0

3.5

4.0

4.5

Verti

cal a

ttenu

atio

n K

d [m

-1]

R(0-) Kd

0.01.02.03.04.05.06.0

0:00 3:00 6:00 9:00 12:00 15:00 18:00 21:00

Tide

[m]

∗

0.000

0.010

0.020

0.030

0.040

0.050

0.060

0.070

0.080

450 500 550 600 650 700 750

wavelength [nm]

a*ph

, a*t

r

0.00.10.20.30.40.50.60.70.80.91.0

a*cd

om ,

aw


0.000

0.020

0.040

0.060

0.080

0.100

0.120

0.140

450 500 550 600 650 700 750

wavelength [nm]

bb*p

h, b

b*tr

, bbw

bbw bb*phy bb* tr


Figure 30: Underwater light climate parameters at Fitzroy River Mouth - Southern Channel (site FE_SOC), 22/06/2002 15:15 (clockwise from top right): tidal stage at sample time, spectral absorption, spectral backscattering, subsurface irradiance reflectance, and vertical diffuse attenuation, and sample site characteristics.

Fitzroy River Estuary Underwater Light ClimateFitzroy River Mouth - Southern Channel

Date 22 June 2002Time (AEST) 15:15Latitude 23.54409Longitude 150.88423Bottom Depth (m) 4.5Secchi Depth (m) N/ACloud cover 0%Weather conditions fine & sunnyTSS (mg/L) 12.1Chl a (µg/L) 0.9substrate not visible

N

E

S

W

Kilometers0 5 10 20

0.0%

5.0%

10.0%

15.0%

20.0%

25.0%

30.0%

35.0%

40.0%

400 450 500 550 600 650 700 750wavelength [nm]

Subs

urfa

ce Ir

radi

ance

Ref

lect

ance

R

(0-)

0.0

0.5

1.0

1.5

2.0

2.5

3.0

3.5

Verti

cal a

ttenu

atio

n K

d [m

-1]

R(0-) Kd

0.01.02.03.04.05.06.0

0:00 3:00 6:00 9:00 12:00 15:00 18:00 21:00

Tide

[m]

∗

0.0000.0200.0400.0600.0800.1000.1200.1400.1600.1800.200

450 500 550 600 650 700 750

wavelength [nm]

a*ph

, a*t

r

0.00.10.20.30.40.50.60.70.80.91.0

a*cd

om ,

aw


0.000

0.001

0.001

0.002

0.002

0.003

0.003

0.004

0.004

0.005

450 500 550 600 650 700 750

wavelength [nm]

bb*p

h, b

b*tr

, bbw

bbw bb*phy bb* tr Figure 31: Underwater light climate parameters at Fitzroy River at Cut-through (site FE_CUT), 16/01/2002 12:30 (clockwise from top right): tidal stage at sample time, spectral absorption, spectral backscattering, subsurface irradiance reflectance, and vertical diffuse attenuation, and sample site characteristics.

Fitzroy River Estuary Underwater Light ClimateFitzroy River at Cut-through

Date 16 January 2002Time (AEST) 12:30Latitude 23 29' 35.7" SLongitude 150 39' 54.4" EBottom Depth (m) 6.7Secchi Depth (m) < 5 cmCloud cover high (>75%)Weather conditions cloudly sky / overcastTSS (mg/L) 346.0Chl a (µg/L) 4.0substrate not visible

N

E

S

W

Kilometers0 5 10 20

0.0%

0.0%

0.0%

0.0%

0.0%

0.1%

0.1%

400 450 500 550 600 650 700 750wavelength [nm]

Subs

urfa

ce Ir

radi

ance

Ref

lect

ance

R

(0-)

0.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1.0

Verti

cal a

ttenu

atio

n K

d [m

-1]

R(0-) Kd

0.01.02.03.04.05.06.0

3:00 6:00 9:00 12:00 15:00 18:00 21:00 0:00 3:00

Tide

[m]

∗

0.000

0.050

0.100

0.150

0.200

0.250

450 500 550 600 650 700 750

wavelength [nm]

a*ph

, a*t

r

0.00.10.20.30.40.50.60.70.80.91.0

a*cd

om ,

aw


0.000

0.001

0.001

0.002

0.002

0.003

0.003

0.004

0.004

0.005

450 500 550 600 650 700 750

wavelength [nm]

bb*p

h, b

b*tr

, bbw

bbw bb*phy bb* tr


Figure 32: Underwater light climate parameters at Fitzroy River Mouth - Mud Island (site FE_MUD), 16/01/2002 14:40 (clockwise from top right): tidal stage at sample time, spectral absorption, spectral backscattering, subsurface irradiance reflectance, and vertical diffuse attenuation, and sample site characteristics.

Fitzroy River Estuary Underwater Light ClimateFitzroy River Mouth - Mud Island

Date 16 January 2002Time (AEST) 14:40Latitude 23 32' 03.7" SLongitude 150 50' 37.8" EBottom Depth (m) 6.5Secchi Depth (m) 0.8Cloud cover highWeather conditionsed during the samplingTSS (mg/L) 8.9Chl a (µg/L) 13.6substrate not visible

N

E

S

W

Kilometers0 5 10 20

0.0%

1.0%

2.0%

3.0%

4.0%

5.0%

6.0%

7.0%

400 450 500 550 600 650 700 750wavelength [nm]

Subs

urfa

ce Ir

radi

ance

Ref

lect

ance

R

(0-)

0.0

0.5

1.0

1.5

2.0

2.5

3.0

3.5

Verti

cal a

ttenu

atio

n K

d [m

-1]

R(0-) Kd

0.01.02.03.04.05.06.0

3:00 6:00 9:00 12:00 15:00 18:00 21:00 0:00 3:00

Tide

[m]

∗

0.000

0.010

0.020

0.030

0.040

0.050

0.060

0.070

450 500 550 600 650 700 750

wavelength [nm]

a*ph

, a*t

r

0.00.10.20.30.40.50.60.70.80.91.0

a*cd

om ,

aw


0.000

0.002

0.004

0.006

0.008

0.010

0.012

0.014

0.016

450 500 550 600 650 700 750

wavelength [nm]

bb*p

h, b

b*tr

, bbw

bbw bb*phy bb* tr Figure 33: Underwater light climate parameters at Fitzroy River Mouth - east of Mud Island (site FE_MUDE), 17/01/2002 12:20 (clockwise from top right): tidal stage at sample time, spectral absorption, spectral backscattering, subsurface irradiance reflectance, and vertical diffuse attenuation, and sample site characteristics.

Fitzroy River Estuary Underwater Light ClimateFitzroy River Mouth - east of Mud Island

Date 17 January 2002Time (AEST) 12:20Latitude 23 32. 339 S (boat)Longitude 150 51.240' E (boat)Bottom Depth (m) 5.5Secchi Depth (m) 0.8Cloud cover high (>85%)Weather conditions light rainTSS (mg/L) 18.7Chl a (µg/L) 6.8substrate not visible

0.01.02.03.04.05.06.0

3:00 6:00 9:00 12:00 15:00 18:00 21:00 0:00 3:00

Tide

[m]

N

E

S

W

Kilometers0 5 10 20

0.0%

2.0%

4.0%

6.0%

8.0%

10.0%

12.0%

400 450 500 550 600 650 700 750wavelength [nm]

Subs

urfa

ce Ir

radi

ance

Ref

lect

ance

R

(0-)

0.0

0.5

1.0

1.5

2.0

2.5

3.0

3.5

Verti

cal a

ttenu

atio

n K

d [m

-1]

R(0-) Kd

∗

0.000

0.005

0.010

0.015

0.020

0.025

0.030

0.035

0.040

0.045

450 500 550 600 650 700 750

wavelength [nm]

a*ph

, a*t

r

0.00.10.20.30.40.50.60.70.80.91.0

a*cd

om ,

aw


0.000

0.002

0.004

0.006

0.008

0.010

0.012

0.014

0.016

450 500 550 600 650 700 750

wavelength [nm]

bb*p

h, b

b*tr

, bbw

bbw bb*phy bb* tr


Figure 34: Underwater light climate parameters at Keppel Bay (near Sea Hill) (site FE_SEA), 17/01/2002 10:20 (clockwise from top right): tidal stage at sample time, spectral absorption, spectral backscattering, subsurface irradiance reflectance, and vertical diffuse attenuation, and sample site characteristics.

Fitzroy River Estuary Underwater Light ClimateKeppel Bay (near Sea Hill)

Date 17 January 2002Time (AEST) 10:20Latitude S / 23 28. 639 S (boat)Longitude E/ 150 57.699' E (boat)Bottom Depth (m) 16mSecchi Depth (m) 0.95Cloud cover high (>75%)Weather conditionsatchy / light rain /sunnyTSS (mg/L) 22.7Chl a (µg/L) 2.2substrate not visible

N

E

S

W

Kilometers0 5 10 20

0.0%

2.0%

4.0%

6.0%

8.0%

10.0%

12.0%

14.0%

400 450 500 550 600 650 700 750wavelength [nm]

Subs

urfa

ce Ir

radi

ance

Ref

lect

ance

R

(0-)

0.0

0.2

0.4

0.6

0.8

1.0

1.2

1.4

1.6

Verti

cal a

ttenu

atio

n K

d [m

-1]

R(0-) Kd

0.01.02.03.04.05.06.0

3:00 6:00 9:00 12:00 15:00 18:00 21:00 0:00 3:00

Tide

[m]

∗

0.000

0.005

0.010

0.015

0.020

0.025

0.030

0.035

0.040

0.045

450 500 550 600 650 700 750

wavelength [nm]

a*ph

, a*t

r

0.00.10.20.30.40.50.60.70.80.91.0

a*cd

om ,

aw


0.000

0.001

0.002

0.003

0.004

0.005

0.006

0.007

0.008

450 500 550 600 650 700 750

wavelength [nm]

bb*p

h, b

b*tr

, bbw

bbw bb*phy bb* tr Figure 35: Underwater light climate parameters at Fitzroy River Mouth - Egg Island (site FE_EGG), 17/01/2002 13:15 (clockwise from top right): tidal stage at sample time, spectral absorption, spectral backscattering, subsurface irradiance reflectance, and vertical diffuse attenuation, and sample site characteristics.

Fitzroy River Estuary Underwater Light ClimateFitzroy River Mouth - Egg Island

Date 17 January 2002Time (AEST) 13:15Latitude 23 32. 109 S (boat)Longitude 150 51.420' E (boat)Bottom Depth (m) 7.6Secchi Depth (m) 0.4Cloud cover high (>85%)Weather conditions st sun coming & goingTSS (mg/L) 24.6Chl a (µg/L) 3.7substrate not visible

N

E

S

W

Kilometers0 5 10 20

0.0%

1.0%

2.0%

3.0%

4.0%

5.0%

6.0%

400 450 500 550 600 650 700 750wavelength [nm]

Subs

urfa

ce Ir

radi

ance

Ref

lect

ance

R

(0-)

0.0

0.5

1.0

1.5

2.0

2.5

3.0

Verti

cal a

ttenu

atio

n K

d [m

-1]

R(0-) Kd

0.01.02.03.04.05.0

3:00 6:00 9:00 12:00 15:00 18:00 21:00 0:00 3:00

Tide

[m]

∗

0.000

0.005

0.010

0.015

0.020

0.025

0.030

0.035

0.040

0.045

450 500 550 600 650 700 750

wavelength [nm]

a*ph

, a*t

r

0.00.10.20.30.40.50.60.70.80.91.0

a*cd

om ,

aw


0.000

0.001

0.001

0.002

0.002

0.003

0.003

0.004

0.004

0.005

450 500 550 600 650 700 750

wavelength [nm]

bb*p

h, b

b*tr

, bbw

bbw bb*phy bb* tr


Figure 36: Underwater light climate parameters at Fitzroy River at Rocky Point (~ 10 km upstream from the mouth) (site FE_ROCK), 17/01/2002 15:30(clockwise from top right): tidal stage at sample time, spectral absorption, spectral backscattering, subsurface irradiance reflectance, and vertical diffuse attenuation, and sample site characteristics.

Fitzroy River Estuary Underwater Light Climate Fitzroy River at Rocky Point (~ 10 km upstream the mouth)

Date 17 January 2002Time (AEST) 15:30Latitude 23 30.252' S (boat)Longitude 150 47.390' E (boat)Bottom Depth (m) 5.3Secchi Depth (m) 0.05 - 0.2Cloud cover cloudly sunny patchesWeather conditions 0TSS (mg/L) 74.6Chl a (µg/L) 4.4substrate not visible

N

E

S

W

Kilometers0 5 10 20

0.0%

0.1%

0.2%

0.3%

0.4%

0.5%

0.6%

0.7%

0.8%

0.9%

400 450 500 550 600 650 700 750wavelength [nm]

Subs

urfa

ce Ir

radi

ance

Ref

lect

ance

R

(0-)

0.0

1.0

2.0

3.0

4.0

5.0

6.0

7.0

8.0

Verti

cal a

ttenu

atio

n K

d [m

-1]

R(0-) Kd

0.01.02.03.04.05.0

3:00 6:00 9:00 12:00 15:00 18:00 21:00 0:00 3:00

Tide

[m]

∗

0.000

0.010

0.020

0.030

0.040

0.050

0.060

0.070

0.080

450 500 550 600 650 700 750

wavelength [nm]

a*ph

, a*t

r

0.00.10.20.30.40.50.60.70.80.91.0

a*cd

om ,

aw


0.000

0.001

0.002

0.003

0.004

0.005

0.006

450 500 550 600 650 700 750

wavelength [nm]

bb*p

h, b

b*tr

, bbw

bbw bb*phy bb* tr

Figure 37: Underwater light climate parameters at Northern edge of Facing Island at "the OAKES" (site PC_OAK), 23/06/2002 9:35 (clockwise from top right): tidal stage at sample time, spectral absorption, spectral backscattering, subsurface irradiance reflectance, and vertical diffuse attenuation, and sample site characteristics.

Port Curtis Underwater Light ClimateNorthern edge of Facing Island at "the OAKES"

Date 23 June 2002Time (AEST) 9:35Latitude 23.76136Longitude 151.33020Bottom Depth (m) 6.2Secchi Depth (m) 2Cloud cover 0%Weather conditions fineTSS (mg/L) 10.8Chl a (µg/L) 0.3substrate Sandy / not visible

N

E

S

W

Kilometers0 5 10

0.0%

5.0%

10.0%

15.0%

20.0%

25.0%

400 450 500 550 600 650 700 750wavelength [nm]

Subs

urfa

ce Ir

radi

ance

Ref

lect

ance

R

(0-)

0.0

0.5

1.0

1.5

2.0

2.5

Verti

cal a

ttenu

atio

n Kd

[m

-1]

R(0-) Kd

0.01.02.03.04.05.0

0:00 3:00 6:00 9:00 12:00 15:00 18:00 21:00 0:00

Tide

[m]

∗

0.0000.0050.0100.0150.0200.0250.0300.0350.0400.0450.050

450 500 550 600 650 700 750

wavelength [nm]

a*ph

, a*tr

0.00.10.20.30.40.50.60.70.80.91.0

a*cd

om ,

aw


0.000

0.002

0.004

0.006

0.008

0.010

0.012

0.014

450 500 550 600 650 700 750

wavelength [nm]

bb*p

h, b

b*tr

, bbw

bbw bb*phy bb* tr


Figure 38: Underwater light climate parameters at SW of Quoin Island (site PC_QUO), 23/06/2002 11:20 (clockwise from top right): tidal stage at sample time, spectral absorption, spectral backscattering, subsurface irradiance reflectance, and vertical diffuse attenuation, and sample site characteristics.

Port Curtis Underwater Light ClimateSW of Quoin Island

Date 23 June 2002Time (AEST) 11:20Latitude 23.81569Longitude 151.28447Bottom Depth (m) 5.5Secchi Depth (m) 1.1Cloud cover 5%Weather conditions fine/ slight hazeTSS (mg/L) 12.2Chl a (µg/L) 0.8substrate Muddy (gray/black)

N

E

S

W

Kilometers0 5 10

0.0%

5.0%

10.0%

15.0%

20.0%

25.0%

30.0%

400 450 500 550 600 650 700 750wavelength [nm]

Subs

urfa

ce Ir

radi

ance

Ref

lect

ance

R

(0-)

0.0

0.5

1.0

1.5

2.0

2.5

3.0

3.5

Verti

cal a

ttenu

atio

n Kd

[m

-1]

R(0-) Kd

0.01.02.03.04.05.0

0:00 3:00 6:00 9:00 12:00 15:00 18:00 21:00 0:00

Tide

[m]

∗

0.000

0.010

0.020

0.030

0.040

0.050

0.060

0.070

450 500 550 600 650 700 750

wavelength [nm]

a*ph

, a*tr

0.00.10.20.30.40.50.60.70.80.91.0

a*cd

om ,

aw


0.0000.0020.0040.0060.0080.0100.0120.0140.0160.0180.020

450 500 550 600 650 700 750

wavelength [nm]

bb*p

h, b

b*tr

, bbw

bbw bb*phy bb* tr Figure 39: Underwater light climate parameters at SW of Quoin Island (site PC_QIS), 23/06/2002 12:15 (clockwise from top right): tidal stage at sample time, spectral absorption, spectral backscattering, subsurface irradiance reflectance, and vertical diffuse attenuation, and sample site characteristics.

Port Curtis Underwater Light ClimateSW of Quoin Island

Date 23 June 2002Time (AEST) 12:15Latitude 23.81640Longitude 151.28339Bottom Depth (m) 2.6Secchi Depth (m) 0Cloud cover 20%Weather conditions sunny/windyTSS (mg/L) 10.5Chl a (µg/L) 0.9substrate Mud

N

E

S

W

Kilometers0 5 10

0.0%

5.0%

10.0%

15.0%

20.0%

25.0%

400 450 500 550 600 650 700 750wavelength [nm]

Subs

urfa

ce Ir

radi

ance

Ref

lect

ance

R

(0-)

0.0

0.5

1.0

1.5

2.0

2.5

Verti

cal a

ttenu

atio

n Kd

[m

-1]

R(0-) Kd

0.01.02.03.04.05.0

0:00 3:00 6:00 9:00 12:00 15:00 18:00 21:00 0:00

Tide

[m]

∗

0.0000.0100.0200.0300.0400.0500.0600.0700.0800.0900.100

450 500 550 600 650 700 750

wavelength [nm]

a*ph

, a*tr

0.00.10.20.30.40.50.60.70.80.91.0

a*cd

om ,

aw


0.000

0.005

0.010

0.015

0.020

0.025

450 500 550 600 650 700 750

wavelength [nm]

bb*p

h, b

b*tr

, bbw

bbw bb*phy bb* tr


Figure 40: Underwater light climate parameters at South Trees Inlet (upstream causeway of Queensland Alumina) (site PC_STI), 23/06/2002 13:30 (clockwise from top right): tidal stage at sample time, spectral absorption, spectral backscattering, subsurface irradiance reflectance, and vertical diffuse attenuation, and sample site characteristics.

Port Curtis Underwater Light ClimateSouth Trees Inlet (upstream causeway of QLD alumina)

Date 23 June 2002Time (AEST) 13:30Latitude 23.86066Longitude 151.30442Bottom Depth (m) 2.9Secchi Depth (m) 0.8Cloud cover 40-50%Weather conditions variableTSS (mg/L) 10.3Chl a (µg/L) 1.1substrate not visible

N

E

S

W

Kilometers0 5 10

0.0%

5.0%

10.0%

15.0%

20.0%

25.0%

400 450 500 550 600 650 700 750wavelength [nm]

Subs

urfa

ce Ir

radi

ance

Ref

lect

ance

R

(0-)

0.0

0.5

1.0

1.5

2.0

2.5

Verti

cal a

ttenu

atio

n Kd

[m

-1]

R(0-) Kd

0.01.02.03.04.05.0

0:00 3:00 6:00 9:00 12:00 15:00 18:00 21:00 0:00

Tide

[m]

∗

0.000

0.0100.020

0.030

0.0400.0500.060

0.0700.080

450 500 550 600 650 700 750

wavelength [nm]

a*ph

, a*tr

0.00.10.20.30.40.50.60.70.80.91.0

a*cd

om ,

aw


0.0000.0020.0040.0060.0080.0100.0120.0140.0160.0180.020

450 500 550 600 650 700 750

wavelength [nm]

bb*p

h, b

b*tr

, bbw

bbw bb*phy bb* tr Figure 41: Underwater light climate parameters at Wiggins Island (mouth of Calliope River) (site PC_WIG), 24/06/2002 8:50 (clockwise from top right): tidal stage at sample time, spectral absorption, spectral backscattering, subsurface irradiance reflectance, and vertical diffuse attenuation, and sample site characteristics.

Port Curtis Underwater Light ClimateWiggins Island (mouth of Calliope River)

Date 24 June 2002Time (AEST) 8:50Latitude 23.81445Longitude 151.19567Bottom Depth (m) 1.7Secchi Depth (m) 1Cloud cover 30-40%Weather conditions mainly sunnyTSS (mg/L) 12.4Chl a (µg/L) 0.9substrate Mud

N

E

S

W

Kilometers0 5 10

0.0%

5.0%

10.0%

15.0%

20.0%

25.0%

30.0%

35.0%

400 450 500 550 600 650 700 750wavelength [nm]

Subs

urfa

ce Ir

radi

ance

Ref

lect

ance

R

(0-)

0.0

0.5

1.0

1.5

2.0

2.5

3.0

3.5

Verti

cal a

ttenu

atio

n Kd

[m

-1]

R(0-) Kd

0.01.02.03.04.05.0

0:00 3:00 6:00 9:00 12:00 15:00 18:00 21:00 0:00

Tide

[m]

∗

0.000

0.010

0.020

0.030

0.040

0.050

0.060

450 500 550 600 650 700 750

wavelength [nm]

a*ph

, a*tr

0.00.10.20.30.40.50.60.70.80.91.0

a*cd

om ,

aw


0.0000.0020.0040.0060.0080.0100.0120.0140.0160.0180.020

450 500 550 600 650 700 750

wavelength [nm]

bb*p

h, b

b*tr

, bbw

bbw bb*phy bb* tr


2.2.4 Mapping Water Quality

The modelling and inversion approach used to produce maps of organic and inorganic constituents of the water column from a time series of three Landsat 5 TM images and one Landsat 7 ETM image of the Port Curtis and Fitzroy Estuary area was based on the processing chain described in Brando and Dekker (2003) and Phinn et al. (2005). This project applied the same methods to mapping select water quality properties in the Port Curtis and Fitzroy Estuary area. Data and Methodology- Modified from Moreton Bay Report (Phinn and Dekker, 2005)

Optical model There are two modelling approaches to determine the relationship between the inherent and apparent optical properties of the water column and the water constituents: analytical modelling and radiative transfer modelling. If an analytical approach is followed, there are two main categories of inverse methods for finding the concentrations of organic and inorganic materials (Mobley 1994). In the first category, implicit solutions are found whereas in the second category, direct or explicit estimates of the concentrations are obtained. With implicit methods the concentrations are varied to minimize the difference between calculated and measured reflectance. Implicit methods allow the use of both analytical and numerical radiative transfer models for calculating the reflectance. A more efficient method for retrieval of concentrations is a direct inversion of an analytical model using a linear Matrix Inversion Method (MIM) as presented in Hoogenboom et al. (1998) and Brando and Dekker (2003). Analytical model inversions have been applied to retrieve optical water quality variables from hyperspectral imagery (Brando and Dekker 2003, Lee et al. 2001, Hoogenboom et al. 1998). Hoogenboom et al. (1998) retrieved concentrations of chlorophyll and tripton of inland turbid waters using MIM with two spectral bands of an airborne hyperspectral R(0-) imagery, Brando and Dekker (2003) retrieved concentrations of chlorophyll, CDOM and tripton in coastal waters using MIM with three spectral bands of an spaceborne hyperspectral R(0-) imagery, and Lee et al. (2001) estimated five water column and bottom properties (chlorophyll, dissolved organic matter, suspended sediments, bottom depth and bottom albedo) by means of an optimization scheme using of all the useable channels of a remote sensing reflectance airborne hyperspectral data set. The Analytical model and MIM In the present study an analytical model for water quality retrieval which relates the subsurface irradiance reflectance to the water constituent concentrations was adopted as described in equations 2 and 3. Substitution of equation 3 in the reflectance model (2) gives:

Rf

b b C

a a C b b C

b b

N

N

b b

N=

+

+ + +

∗

=

∗

=

∗

=

∑

∑ ∑

01

0

10

1

j j

j

j j

j

j j

j

. (7)

Since the inherent and apparent optical properties are dependent on wavelength, but the concentrations are not, this equation is rewritten to a set of M equations (for M wavelengths), where each equation has the form:

j

N

1j i

iij

i

iij

i

ii0

i

ii0

)()(

1)()()(

)(

)()(

1)()()(

)(

CfR

bfR

a

fR

bfR

a

b

b

∑=

∗∗

−−

=

−+−

λλλ

λλλ

λλλ

λλλ

. (8)

A more concise format of this equation is obtained by using matrix notation: y Ax= , (9)


,

, )()(1)(

)()()(

N1,...,=j M1,...,=i , )()(1)(

)()()(

jj

i

ii0

i

ii0i

i

iij

i

iijij

CxfRb

fRay

fRb

fRaA

b

b

=

−+−=

−−= ∗∗

λλλ

λλλ

λλλ

λλλ

(10)

where A is a MxN matrix in which the M rows represent the number of wavelengths and the N columns represent the number of constituents. In order to distinguish vectors and matrices from other variables, they are shown in bold. The vector x contains the N unknown concentrations. If the number of wavelengths equals the number of constituents the matrix A is a NxN square matrix and the concentration vector x can be directly calculated as the analytical solution of the linear system. If the system is over-determined, i.e. when there are number of wavelengths is larger than the number of constituents to retrieve, the Singular Value Decomposition method can be used. This type of solution has the property of minimizing the residual error in a least squares sense (see for details of the method Press et al. 1992). MIM implementation for Port Curtis and Fitzroy Estuary To apply the MIM method to the four Landsat R(0-) images, we need to parameterize the analytical model that describes the relationship between the R(0-) for optically deep waters as a function of the IOPs with a set of spectral SIOPs. Twelve sets of spectral SIOPs were selected from the underwater light climate parameterization of Port Curtis and Fitzroy River Estuary fieldwork sites (see sections 2.3.1 and 2.3.2) to be used in the Landsat imagery inversion. As already stated, the specific absorption by CHL was fixed to the a*ph from another Australian estuarine site with a similar chlorophyll concentration range. Thus, bb*CHL is fixed for consistency to the average of the bb*CHL retrieved for all sites. The adopted bb*CHL spectrum is comparable with Bukata (1995) and Dekker (1993). Figures 42-46 show the spectral SIOPS sets for Port Curtis field sites convoluted to Landsat bands, namely:

• Port Curtis at the Northern edge of Facing Island, 23-06-02. • Port Curtis at the SW of Quoin Island, 23-06-02, 11:20 AM. • Port Curtis at the SW of Quoin Island, 23-06-02., 12:15 PM. • Port Curtis at the outlet of South Trees Inlet (upstream causeway of Queensland Alumina),

23-06-02. • Port Curtis at Wiggins Island (mouth of Calliope River), 24-06-02

Figures 47-53 present the spectral SIOPs sets for the Fitzroy River Estuary field sites convoluted to Landsat bands, namely:

• Fitzroy River Estuary at Mud Island, 16-01-02. • Fitzroy River Estuary east of Mud Island, 17-01-02. • In Keppel Bay (near Sea Hill), 17-01-02. • Fitzroy River Estuary at Rocky Point (~ 10 km upstream the mouth), 17-01-02 • In middle of Keppel Bay, 22-06-02 • Fitzroy River Estuary at the Fitzroy River mouth, 22-06-02 • Fitzroy River Estuary (FE_PIR site) at Pirate Point, 22-06-02

The anisotropy factor of the downwelling light field f is a function of µd and µu (see equation 2), the average cosines of the downwelling and upwelling light field, respectively, which will depend on the downwelling irradiance distribution and the shape factors (see Aas (1987) for a more detailed explanation). For the inversion of the three Landsat images it was assumed a diffuse upwelling light field ( µu =0.5) and µ d was approximated by µ0 (the cosine of sun zenith) according to the Walker (1994):

0211

µ+=f (11)


Figure 42: SIOPs absorption coefficients (from: phytoplankton, tripton, pure water and CDOM) and backscattering coefficients (from: water, phytoplankton and tripton) convoluted to Landsat bands for Port Curtis (PC_OAK site) at the Northern edge of Facing Island at "The Oakes", 23-06-02.

PC_OAK

0.000

0.005

0.010

0.015

0.020

0.025

450 500 550 600 650 700 750 800 850

wavelength [nm]

a*ph

, a*t

r

0.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

a*cd

om ,

aw


0.000

0.002

0.004

0.006

0.008

0.010

0.012

450 500 550 600 650 700 750 800 850

wavelength [nm]

bb*p

h, b

b*tr

, bbw

bbw bb*phy bb* tr

Figure 43: SIOPs absorption coefficients (from: phytoplankton, tripton, pure water and CDOM) and backscattering coefficients (from: water, phytoplankton and tripton) convoluted to Landsat bands for Port Curtis (PC_QUO site) at the SW of Quoin Island, 23-06-02, 11:20 AM.

PC_QUO

0.000

0.005

0.010

0.015

0.020

0.025

0.030

450 500 550 600 650 700 750 800 850

wavelength [nm]

a*ph

, a*t

r

0.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

a*cd

om ,

aw


0.000

0.002

0.004

0.006

0.008

0.010

0.012

0.014

0.016

450 500 550 600 650 700 750 800 850

wavelength [nm]

bb*p

h, b

b*tr

, bbw

bbw bb*phy bb* tr


Figure 44: SIOPs absorption coefficients (from: phytoplankton, tripton, pure water and CDOM) and backscattering coefficients (from: water, phytoplankton and tripton) convoluted to Landsat bands for Port Curtis (PC_QIS site) at the SW of Quoin Island, 23-06-02., 12:15 PM.

PC_QIS

0.000

0.005

0.010

0.015

0.020

0.025

0.030

0.035

0.040

0.045

450 500 550 600 650 700 750 800 850

wavelength [nm]

a*ph

, a*t

r

0.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

a*cd

om ,

aw


0.000

0.002

0.004

0.006

0.008

0.010

0.012

0.014

0.016

0.018

450 500 550 600 650 700 750 800 850

wavelength [nm]

bb*p

h, b

b*tr

, bbw

bbw bb*phy bb* tr

Figure 45: SIOPs absorption coefficients (from: phytoplankton, tripton, pure water and CDOM) and backscattering coefficients (from: water, phytoplankton and tripton) convoluted to Landsat bands for Port Curtis (PC_STI site) at the outlet of South Trees Inlet (upstream causeway of QLD alumina), 23-06-02.

PC_STI

0.000

0.005

0.010

0.015

0.020

0.025

0.030

0.035

450 500 550 600 650 700 750 800 850

wavelength [nm]

a*ph

, a*t

r

0.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

a*cd

om ,

aw


0.000

0.002

0.004

0.006

0.008

0.010

0.012

0.014

0.016

0.018

450 500 550 600 650 700 750 800 850

wavelength [nm]

bb*p

h, b

b*tr

, bbw

bbw bb*phy bb* tr


Figure 46: SIOPs convoluted to Landsat bands for Port Curtis (PC_WIG site) at Wiggins Island (mouth of Calliope River), 24-06-02.

PC_WIG

0.000

0.005

0.010

0.015

0.020

0.025

450 500 550 600 650 700 750 800 850

wavelength [nm]

a*ph

, a*t

r

0.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

a*cd

om ,

aw


0.000

0.002

0.004

0.006

0.008

0.010

0.012

0.014

0.016

450 500 550 600 650 700 750 800 850

wavelength [nm]

bb*p

h, b

b*tr

, bbw

bbw bb*phy bb* tr

Figure 47: SIOPs absorption coefficients (from: phytoplankton, tripton, pure water and CDOM) and backscattering coefficients (from: water, phytoplankton and tripton) convoluted to Landsat bands for Fitzroy River Estuary (FE_MUD site) at Mud Island, 16-01-02.

FE_MUD

0.000

0.005

0.010

0.015

0.020

0.025

0.030

450 500 550 600 650 700 750 800 850

wavelength [nm]

a*ph

, a*t

r

0.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

a*cd

om ,

aw


0.000

0.002

0.004

0.006

0.008

0.010

0.012

0.014

450 500 550 600 650 700 750 800 850

wavelength [nm]

bb*p

h, b

b*tr

, bbw

bbw bb*phy bb* tr


Figure 48: SIOPs absorption coefficients (from: phytoplankton, tripton, pure water and CDOM) and backscattering coefficients (from: water, phytoplankton and tripton) convoluted to Landsat bands for Fitzroy River Estuary (FE_MUDE site) east of Mud Island, 17-01-02.

FE_MUDE

0.000

0.005

0.010

0.015

0.020

0.025

450 500 550 600 650 700 750 800 850

wavelength [nm]

a*ph

, a*t

r

0.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

a*cd

om ,

aw


0.000

0.002

0.004

0.006

0.008

0.010

0.012

0.014

450 500 550 600 650 700 750 800 850

wavelength [nm]

bb*p

h, b

b*tr

, bbw

bbw bb*phy bb* tr

Figure 49: SIOPs absorption coefficients (from: phytoplankton, tripton, pure water and CDOM) and backscattering coefficients (from: water, phytoplankton and tripton) convoluted to Landsat bands for Fitzroy River Estuary (FE_SEA site) in Keppel Bay (near Sea Hill), 17-01-02.

FE_SEA

0.000

0.005

0.010

0.015

0.020

0.025

450 500 550 600 650 700 750 800 850

wavelength [nm]

a*ph

, a*t

r

0.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

a*cd

om ,

aw


0.000

0.001

0.002

0.003

0.004

0.005

0.006

0.007

450 500 550 600 650 700 750 800 850

wavelength [nm]

bb*p

h, b

b*tr

, bbw

bbw bb*phy bb* tr


Figure 50: SIOPs absorption coefficients (from: phytoplankton, tripton, pure water and CDOM) and backscattering coefficients (from: water, phytoplankton and tripton) convoluted to Landsat bands for Fitzroy River Estuary (FE_ROCK site) at Rocky Point (~ 10 km upstream the mouth), 17-01-02.

FE_ROCK

0.000

0.005

0.010

0.015

0.020

0.025

0.030

0.035

450 500 550 600 650 700 750 800 850

wavelength [nm]

a*ph

, a*t

r

0.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

a*cd

om ,

aw


0.000

0.001

0.001

0.002

0.002

0.003

0.003

0.004

0.004

0.005

450 500 550 600 650 700 750 800 850

wavelength [nm]

bb*p

h, b

b*tr

, bbw

bbw bb*phy bb* tr

Figure 51: SIOPs absorption coefficients (from: phytoplankton, tripton, pure water and CDOM) and backscattering coefficients (from: water, phytoplankton and tripton) convoluted to Landsat bands for Fitzroy River Estuary (FE_KEB site) at In middle of Keppel Bay, 22-06-02.

FE-KEB

0.000

0.005

0.010

0.015

0.020

0.025

0.030

450 500 550 600 650 700 750 800 850

wavelength [nm]

a*ph

, a*t

r

0.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

a*cd

om ,

aw


0.000

0.002

0.004

0.006

0.008

0.010

0.012

0.014

0.016

0.018

450 500 550 600 650 700 750 800 850

wavelength [nm]

bb*p

h, b

b*tr

, bbw

bbw bb*phy bb* tr


Figure 52: SIOPs absorption coefficients (from: phytoplankton, tripton, pure water and CDOM) and backscattering coefficients (from: water, phytoplankton and tripton) convoluted to Landsat bands for Fitzroy River Estuary (FE_RMO site) at the Fitzroy River mouth, 22-06-02.

FE-RMO

0.000

0.005

0.010

0.015

0.020

0.025

450 500 550 600 650 700 750 800 850

wavelength [nm]

a*ph

, a*t

r

0.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

a*cd

om ,

aw


0.000

0.005

0.010

0.015

0.020

0.025

0.030

0.035

0.040

450 500 550 600 650 700 750 800 850

wavelength [nm]

bb*p

h, b

b*tr

, bbw

bbw bb*phy bb* tr

Figure 53: SIOPs absorption coefficients (from: phytoplankton, tripton, pure water and CDOM) and backscattering coefficients (from: water, phytoplankton and tripton) convoluted to Landsat bands for Fitzroy River Estuary (FE_PIR site) at Pirate Point, 22-06-02.

FE-PIR

0.000

0.005

0.010

0.015

0.020

0.025

0.030

450 500 550 600 650 700 750 800 850

wavelength [nm]

a*ph

, a*t

r

0.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

a*cd

om ,

aw


0.000

0.020

0.040

0.060

0.080

0.100

0.120

450 500 550 600 650 700 750 800 850

wavelength [nm]

bb*p

h, b

b*tr

, bbw

bbw bb*phy bb* tr

Porting the MIM implementation from Hyperspectral to Multispectral Image Data Sets MIM was originally implemented in an IDL/ENVI® routine to produce chlorophyll (CHL), coloured dissolved organic matter (CDOM) and tripton (TR) maps by inverting three Hyperion R(0-) bands (two centred at 490 and 670 nm and the bin of 5 the Hyperion R(0-) bands in the 700-740 nm range). Cramer’s rule was used for the analytical solution of the 3x3 linear system (equation 10).


The following two paragraphs provide an explanation of band selection for use in the MIM from EO-1 Hyperion image data and provide useful information for applications in this context where there is a choice of spectral bands to use (e.g. airborne or satellite hyperspectral data). Initially bands were selected that were closest to a spectral feature typical for one of the variables of CHL, CDOM and TR: this would be a blue band around 447 nm (for absorption by CDOM and CHL); a band at 550 nm for the least influence by CHL and a low influence by CDOM and a band around 676 nm where CHL has a red absorption maximum. An analysis of possible causes for this failure was (i) that the shortest wavelength available at 447 nm was still too noisy; (ii) the band at 550 nm posed a problem as the substrate may contribute here (one of the reasons to include a bottom signal at 8 m depth in the numerical simulation); and (iii) the 676 nm band is spectrally distant from the 447 nm CDOM band causing small errors in the SIOPs (e.g. the representation of the absorption by CDOM and tripton by an exponential function with a fixed slope) creating relatively large errors for a MIM based solution.

Thus, through some iterations, it was found that for Hyperion imagery of Deception Bay the 490, 670 and 700-740 nm bands performed best because they avoid spectral areas where Hyperion data are still noisy. Probably the 670 nm band for CHL estimation and the bin of the 700-740 nm bands for TR estimation being close together positively influenced the results as the MIM inversion could more accurately compute a scattering and absorption contribution for each of the components. Unlike the Hyperion implementation where the three narrow bands are located in selected spectral ranges according to the R(0-) sensitivity model and to avoid substrate visibility, the implementation of MIM for Landsat TM 5 and ETM 7 involves the use of the only available three broad bands in the visible region of the spectrum. In these Landsat TM 5 and ETM 7 broad spectral bands absorption features interfere and eliminate each other as can be seen comparing the UWLC descriptions at 1 nm (Figures 24-41) with the respective SIOPs parameterization for TM (Figures 42-52) (Dekker and Peters 1993). An outline of the steps used in the bio-optical model implementation used to estimate concen-trations of water column constituents from the Landsat image data is presented in Figure 54. Sensitivity analysis demonstrates that it is not possible to retrieve CHL at the concentration level present in Port Curtis or in the Fitzroy River Estuary with the level of noise associate with the Landsat 7 ETM imagery. Thus, CHL was fixed for all the inversions and we are left with a 3x2 linear system (three wavelengths and two constituents to retrieve, TR and CDOM for each pixel). The Singular Value Decomposition method was used to solve this over-determined system. This type of solution has the property of minimizing the residual error in a least squares sense. The image inversion, i.e. the concentration retrieval in each pixel, was based on an optimisation technique that selected the best optical parameterization for each pixel of the image. In fact, the SVD inversion (equation 10), was applied iteratively to each pixel varying the optical parameterization within one of the SIOP sets based on the January 2002 and June 2002 field campaigns for the Fitzroy River Estuary images (Figures 47-53) and or one of SIOP sets based on the June 2002 field campaign for the Port Curtis images (Figures 42-46). For each set of SIOPs, an analytical model R(0-) can be simulated using the MIM retrieved concentrations for each pixel. d(R0-), the difference between the imagery R(0-) and the inverse-forward simulated R(0-) was used to calculate a measure of the optical closure in each pixel. The concentration values associated with the best optical closure are used for the maps and the value of d(R0-) calculated for each pixel can be presented as a map of the level of confidence of water quality parameter estimates. If the value of d(R0-) exceeds a threshold, or if the retrieved concentrations are negative, the pixel can be flagged as not mapped.

Analytical model and inversion to retrieve Secchi Disk transparency A simple measure of water column clarity is the Secchi Depth (SD): which is the depth at which the


signal from the white portion of the Secchi disk is not discernible from the black portion of the Secchi disk. In order to derive an equation for estimating Secchi Disk transparency it is first necessary to describe the subsurface irradiance reflectance over a water body with bottom visibility according to Dekker et al. (2001b) as:

])(exp[)(),0( d HKRARHR κ+−−+=− ∞∞ (12) where (R(0-, H) is the subsurface irradiance reflectance over a water body with bottom visibility at depth H; R∞ is the subsurface irradiance reflectance of an infinitely deep water column; Kd is the vertical downward attenuation of the downwelling light stream; A is the bottom irradiance reflectance and κ is the vertical upward attenuation of upwelling irradiance originating from infinitely deep water or from the substrate reflectance. Exactly at the Secchi depth (H=SD) the Right Hand Side of equation 8 for the white portion of the disk equals (A=1) equals the Right Hand Side of equation 12 for the black portion of the disk (A=0) for the wavelength of maximum penetration:

])(exp[)0(])(exp[)1( dd SDKRRSDKRR κκ +−−+=+−−+ ∞∞∞∞ (13) if we simplify the terms , (13) becomes :

0])(exp[ d =+− SDK κ . (14)

If we assume that uK=κ , then

ud

1KK

SD+−=

(15)

According to definition of dK and uK as given in Dekker et al. (2001b), table 5, and remembering the definition of f (equation 2):

fKbababaKK bbb d

du

ud

udud =+⋅+=+++≅+

µµµµ

µµ (16)

We can now obtain an expression for Secchi disk transparency as it will be located at that wavelength (in the human eye visibility domain) where:

−=

)(d λλ KfMAXSD

(17) Since the concentrations of the CDOM, CHL and Tripton are retrieved for each pixel of the imagery using the MIM algorithm, and we also know the specific inherent optical properties for each pixel, we can calculate the SD values from the L7/ETM or the L5/TM image.


Figure 54: Outline of the steps used in the bio-optical model to estimate concentrations of water column constituents from the Landsat image data.

NOT OKOK

Optical parameterization(SIOPS)

Fully correctedimage R(0 -)

Concentration retrieval throughoptimisation.

The concentrations are retrieved and thebest optical parameterization for each pixel

is selected.

Thematic ProductMaps of Kd PAR and Secchi

Depth

Evaluate the opticalclosure

(optical closure =the “level of

confidence” of waterquality parameter

estimates)

Quality Control ProductMaps of level of confidence and optical parameterization

Primary ProductMaps of CDOM, TSS

(CHL is fixed)

Pixel is flagged as“not mapped”.


Results and Discussion The following section discusses the water quality mapping results in Port Curtis and Fitzroy Estuary as separate study sites. The maps of water quality parameters were produced by integrating field measured optical properties with commercially available moderate spatial resolution, multi-spectral satellite image data using the same technique applied in the Moreton Bay Study. Significantly different inherent optical properties (SIOPs) were recorded in the Port Curtis and Fitzroy Estuary waters due to differences in sediment and organic constituents of the water bodies. Realistic maps of the dissolved organic matter and total suspended matter load were produced using these SIOPs providing further proof of concept for the “Inherent optical domain based mapping of aquatic coastal ecosystem properties“ outlined in Figures 5 and 53.

The fully corrected Landsat 5 TM and Landsat 7 ETM R(0-) images are presented in Figures 13-20. The full set of water quality products are presented at four image-maps per page to provide a multi-temporal dimension to each of the following mapped variables: CDOM, TSM, Secchi depth, model fit and SIOP distribution for Fitzroy Estuary in Figures 54-58 and for Port Curtis in Figures 59-63. Full page maps for each site and date are contained in Appendix 6. The August 21st 1986 and July 24th 2002 images (Figures 13-14 and 19-20) were captured close to the high tide point, during two relatively dry and low flow periods. The other two images from August 16th 1980 and March 7th 1995 (Figures 15 to 18) were captured at mid-point of an outgoing high tide and a incoming low tides. The true colour images reveal characteristic optical properties of Port Curtis and Fitzroy Estuary water bodies which will be referred to as (inherent) optical domains. The characteristic patterns of optical properties mapped from CDOM, total suspended matter (tripton) and estimated Secchi transparency depth are outlined below and demonstrate, as in Moreton Bay, that these patterns vary spatially in response to tidal effects, fluvial discharge and prevailing wind conditions. In Port Curtis five distinct water colourations were apparent from inspecting the four single date images:

(1) constantly turbid case 2, high TSM and low CDOM waters of The Narrows, north of Friend Point;

(2) mixed inner-harbour waters, stretching from the entrance to the Narrows to the mouth of the Boyne River, highly variable, switching from low TSM/CDOM to high TSM/CDOM depending on tidal stage and run-off;

(3) complex harbour mixing zone, between Gladstone Port and Facing Island, and the North and South Entrances, predominantly green waters, with low CDOM and TSM;

(4) outer harbour mixing zone, dominated by near case 1 oceanic waters on incoming tides and green/brown case 2 waters on outgoing tides; and

(5) shallow near shore areas with submerged aquatic vegetation or substrate visible at the time of image acquisition.

From the four Port Curtis images examined, the primary variations in the spatial distribution and nature of estimated CDOM, TSM and Secchi-depth appeared to be controlled by tidal stage and stream-flow of major rivers flowing into the harbour. Two other factors also impacted on the accuracy of the mapped variables: the level of fit of the model used to estimate concentrations and errors in the Landsat image unable to be removed in pre-processing. The level of fit of the model used to match SIOP’s to each pixel and estimate CDOM and TSM concentrations and Secchi depth was lowest in areas where substrate may have been visible or exposed when the image was acquired (Figures 59-63). This would explain erroneous CDOM and TSM levels at Pelican Banks, Wiggins and South-Trees Islands. Due to an artefact in the image acquisition process, Landsat TM/ETM data over coastal areas is subject to significant striping effects, resulting in up to 5% variations in image pixel values. As mapping of variations in water column constituents and submerged aquatic vegetation relies on signals that are 5-10% of the reflectance range, the striping artefact will be distinct in our water quality and substrate maps.


For the Fitzroy Estuary sequence of images four distinct zones of water colouration were observed from characteristic patterns in the CDOM, TSM and Secchi depth image products for each of the four image dates:

(1) riverine waters, with consistently high TSM and CDOM levels, extending from Pirate Point to the barrage in Rockhampton;

(2) tidal (Casuarina, Raglan and Connor) creeks and The Narrows area, which appear as predominantly case 2 waters, alternating between high and low TSM and CDOM levels dependent on tidal stage;

(3) highly variable green-brown waters in the Fitzroy Estuary-Middle Channel area, with dominant colours being controlled by tidal stage and Fitzroy River discharge levels; and

(4) oceanic-estuarine waters, near-case 1 predominantly blue-green waters with changes to case 2 under high discharge and with a TSM-CDOM plume extension into Keppel Bay.

Likely controls and errors in the magnitude and spatial variation in the distribution of dissolved organic matter and particulate matter depicted in the CDOM and TSM image were similar to those observed for the Port Curtis area. A comprehensive error assessment of these map products was significantly beyond the scope of this project as significant field data were required for initial model calibration. Accuracy assessment will be undertaken in Phase 2 of the Coastal CRC projects in this region. The major difference for the Fitzroy was the influence of the Fitzroy River and a significantly larger estuary, as opposed to the enclosed water bodies of Port Curtis. As a result stream discharge from the Fitzroy River is likely to be the predominant control of the concentration and distribution of constituents in the water column, with tidal induced resuspension also playing a large role affecting patterns on a smaller scale and daily basis. Similar problems were observed in areas where the model had a poor fit to the image data, mainly in shallow clear water areas with submerged aquatic vegetation and areas of surface algal blooms.


Figure 54: Image based maps showing the concentration of coloured dissolved organic matter (CDOM, from absorption at 440 nm). Note the lilac colour represents areas where the error between measured and modelled spectra was too high to find a sensible solution-this is mainly due to a lack of pure ocean water optical properties in the model parameterisation. – Fitzroy Estuary sub-scene : - Landsat 5 Thematic Mapper image captured at 09:30 on 21st August 1986 - Landsat 5 Thematic Mapper image captured at 09:30 on 16th August 1990 - Landsat 5 Thematic Mapper image captured at 09:30 on 7th March 1995 - Landsat 7 Enhanced Thematic Mapper image captured at 09:45 on 24th July 2002.


Figure 55: Image based maps showing the concentration of total suspended matter (tripton) – Fitzroy Estuary sub-scene : - Landsat 5 Thematic Mapper image captured at 09:30 on 21st August 1986 - Landsat 5 Thematic Mapper image captured at 09:30 on 16th August 1990 - Landsat 5 Thematic Mapper image captured at 09:30 on 7th March 1995 - Landsat 7 Enhanced Thematic Mapper image captured at 09:45 on 24th July 2002


Figure 56: Image based maps showing the estimated Secchi disk transparency (an estimate of water clarity) – Fitzroy Estuary sub-scene : - Landsat 5 Thematic Mapper image captured at 09:30 on 21st August 1986 - Landsat 5 Thematic Mapper image captured at 09:30 on 16th August 1990 - Landsat 5 Thematic Mapper image captured at 09:30 on 7th March 1995 - Landsat 7 Enhanced Thematic Mapper image captured at 09:45 on 24th July 2002


Figure 57: Images maps showing an estimate of the reliability of the model used to predict water quality variables – based on difference between measured and modelled reflectance – Fitzroy Estuary sub-scene: above the salmon colour ( 1% error in R(0-) the error is significant) - Landsat 5 Thematic Mapper image captured at 09:30 on 21st August 1986 - Landsat 5 Thematic Mapper image captured at 09:30 on 16th August 1990 - Landsat 5 Thematic Mapper image captured at 09:30 on 7th March 1995 - Landsat 7 Enhanced Thematic Mapper image captured at 09:45 on 24th July 2002


Figure 58: Image map showing the specific inherent optical properties used for modelling water column constituents in each image pixel – Fitzroy Estuary sub-scene: Each colour represents a water mass with a different optical constituents composition) - Landsat 5 Thematic Mapper image captured at 09:30 on 21st August 1986 - Landsat 5 Thematic Mapper image captured at 09:30 on 16th August 1990 - Landsat 5 Thematic Mapper image captured at 09:30 on 7th March 1995 - Landsat 7 Enhanced Thematic Mapper image captured at 09:45 on 24th July 2002


Figure 59: Image based maps showing the concentration of coloured dissolved organic matter (CDOM, from absorption at 440 nm) Port Curtis sub-scene : - Landsat 5 Thematic Mapper image captured at 09:30 on 21st August 1986 - Landsat 5 Thematic Mapper image captured at 09:30 on 16th August 1990 - Landsat 5 Thematic Mapper image captured at 09:30 on 7th March 1995 - Landsat 7 Enhanced Thematic Mapper image captured at 09:45 on 24th July 2002.


Figure 60: Image based maps showing the concentration of total suspended matter (tripton) – Port Curtis sub-scene : - Landsat 5 Thematic Mapper image captured at 09:30 on 21st August 1986 - Landsat 5 Thematic Mapper image captured at 09:30 on 16th August 1990 - Landsat 5 Thematic Mapper image captured at 09:30 on 7th March 1995 - Landsat 7 Enhanced Thematic Mapper image captured at 09:45 on 24th July 2002


Figure 61: Image based maps showing the estimated Secchi disk transparency (an estimate of water clarity) – Port Curtis sub-scene : - Landsat 5 Thematic Mapper image captured at 09:30 on 21st August 1986 - Landsat 5 Thematic Mapper image captured at 09:30 on 16th August 1990 - Landsat 5 Thematic Mapper image captured at 09:30 on 7th March 1995 - Landsat 7 Enhanced Thematic Mapper image captured at 09:45 on 24th July 2002


Figure 62: Images maps showing an estimate of the reliability of the model used to predict water quality variables – based on difference between measured and modelled reflectance – Port Curtis sub-scene : - Landsat 5 Thematic Mapper image captured at 09:30 on 21st August 1986 - Landsat 5 Thematic Mapper image captured at 09:30 on 16th August 1990 - Landsat 5 Thematic Mapper image captured at 09:30 on 7th March 1995 - Landsat 7 Enhanced Thematic Mapper image captured at 09:45 on 24th July 2002


Figure 63: Image map showing the specific inherent optical properties used for modelling water column constituents in each image pixel – Port Curtis sub-scene : - Landsat 5 Thematic Mapper image captured at 09:30 on 21st August 1986 - Landsat 5 Thematic Mapper image captured at 09:30 on 16th August 1990 - Landsat 5 Thematic Mapper image captured at 09:30 on 7th March 1995 - Landsat 7 Enhanced Thematic Mapper image captured at 09:45 on 24th July 2002


2.3 Port Curtis and Fitzroy Estuary Substrate Mapping

2.3.1 Field and Image Data Pre-Processing

The four Landsat scenes used for the substrate mapping were pre-processed for image based mapping of submerged and exposed inter-tidal areas where seagrass or other aquatic vegetation were present. The input image data consisted of four images (Figures 13 to 20) which had been processed to at-surface reflectance for terrestrial areas and to remote sensing reflectance for water areas using the methods outlined in Section 2.2.2. The first stage in the processing was to georeference all of the images to ensure they were in the same projection, datum and coordinate system, and that the measure of spatial mis-registration, root-mean square error, was < 0.5 pixels (i.e. < 12.5m) prior to masking for separate image processing. These preprocessed scenes were then split into Fitzroy Estuary and Port Curtis sub-scenes for separate processing of each area. To separate the water bodies and the exposed inter-tidal areas in images at low tide (1990 and 1995 images), separate masks were developed and applied to each image data set in the Fitzroy Estuary and Port Curtis sub-scenes. The high tide mask was defined using the near-infrared band in the image that was collected at the highest tidal stage, July 24th 2002. The NIR band value used to create the mask was selected after ensuring that all water edge features (e.g. Gladstone Harbour) facilities would be retained as land pixels. The mask was then applied to all of the Landsat sub-scenes for Fitzroy Estuary and Port Curtis to separate land and water pixels. The water masked area of two scenes collected at low tide (1990 and 1995) contained exposed inter-tidal substrate and vegetation hence a second mask was developed to extract this area for the 1990 and 1995 sub-scenes for further analysis. The resulting images from the masking process, listed in Table 6, were ready for image based mapping of sub-tidal and inter-tidal aquatic vegetation. Table 6: Listing of masked-image sub-scenes used for substrate mapping. Image Date Fitzroy Estuary Port Curtis 21st August 1986 Water masked – high tide Water masked – high tide 16th August 1990 Water masked – high tide

Inter-tidal masked area Water masked – high tide Inter-tidal masked area

7th March 1995 Water masked – high tide Inter-tidal masked area

Water masked – high tide Inter-tidal masked area

24th July 2002 Water masked – high tide Water masked – high tide

2.3.2 Mapping Substrate Parameters

Mapping Methodology The method used to map submerged and inter-tidal substrate in the Fitzroy Estuary and Port Curtis sub-scenes is similar to that applied for Moreton Bay (Phinn and Dekker, 2005) outlined in Figure 5. This approach has three stages, initially mapping optical domains or regions of varying water clarity, next in the clear water optical domains substrate cover types are mapped, and finally in images with exposed inter-tidal areas, exposed substrate types were mapped. Supervised classification approaches were used with the Spectral Angle Mapping (SAM) algorithm (Research Systems, 2004) used for both the optical domain mapping and subsequent substrate cover type mapping. Figures 64 to 67 demonstrate the results from the optical domain mapping for each of the four image dates in the Fitzroy Estuary and water-masked sub-scenes. For the optical domain classification, field experience and ancillary information were used to select training sites over water bodies of varying degrees of clarity. The training sites, or Regions of Interest (ROI), were used to train the SAM classifier and to produce a map of optical domains. If the image contained a clear optical or slightly mixed optical domain, this area was


then subject to further classification to map substrate cover types. In the case of the Fitzroy Estuary images, the only regions with clear water were located around Keppel Islands, hence inshore and estuarine mapping of substrate cover types was not carried out. For Port Curtis, field knowledge and ancillary data, including the Queensland Department of Primary Industries latest seagrass maps (Figure 72) were used to select training sites for the SAM classifier. This process was also repeated for the exposed inter-tidal areas in the 1990 and 1995 sub-scenes to map seagrass cover. The final map for the Port Curtis sub-scenes, included inter- and sub-tidal seagrass in exposed and clear water areas, and a classification of optical domain of the other areas.

Figure 64: Optical domains (pixels assigned to a domain based on estimated water colour) – Fitzroy Estuary sub-scene from Landsat 5 Thematic Mapper image captured at 09:30 on 21st August 1986.


Figure 65: Optical domains (pixels assigned to a domain based on estimated water colour) – Fitzroy Estuary sub-scene from Landsat 5 Thematic Mapper image captured at 09:30 on 16th August 1990.


Figure 66: Optical domains (pixels assigned to a domain based on estimated water colour) – Fitzroy Estuary sub-scene from Landsat 5 Thematic Mapper image captured at 09:30 on 7th March 1995.


Figure 67: Optical domains (pixels assigned to a domain based on estimated water colour) – Fitzroy Estuary sub-scene from Landsat 7 Enhanced Thematic Mapper image captured at 09:45 on 24th July 2002.


Mapping Results: The variability in extent of sub-tidal and inter-tidal substrate cover type in the Landsat based maps for Port Curtis shown in Figures 68 to 71 illustrates the limitations of attempting to map submerged aquatic vegetation in a highly turbid and dynamic environment using optical remote sensing techniques. Considering all of the image-based maps, the tidal stage at image acquisition and stream discharge prior to acquisition, substrate cover may only be mapped under certain optically optimal conditions within the Port Curtis area. Those conditions appear to be: below mid-tidal levels, on ingoing or outgoing tides; and at a time of low stream discharge. These conditions occurred in the 16th August 1990 and 7th March 1995 Landsat 5 Thematic Mapper image acquisitions. In both of these images, (Figures 65 and 66), seagrass beds were mapped in shallow waters and inter-tidal areas at locations throughout Port Curtis, including: Pelican Banks; the North Entrance area; western shores of Facing and Curtis Islands; the western shores of Gladstone harbour, extending from Friend Point in the north, including Wiggins Island and Calliope River mouth in the central section, to the Boyne River mouth and south side of South Trees Island. In these images there are still pixels misclassified as seagrass, due to the similarity of dark-turbid waters with seagrass on similar substrate. Some of these misclassification effects are evident in Figures 69 and 70, where deep and turbid harbour waters and upstream sections of the Calliope and Boyne Rivers have been classified as seagrass. The two high tide images, 21st August 1986 and 24th July 2002 (Figures 64 and 67) also illustrate this effect on a larger scale, as in both cases the harbour is dominated by turbid waters and mixed green-brown waters. Due to the increased level of scattering and absorption of sun and sky light in the water bodies in these images, in contrast to the lower tide images, no distinct signal was able to be detected from seagrass beds. As a result of this, large areas of the 1986 and 2002 images were misclassified as seagrass beds due to the similarity in reflectance values of the target and mixed water bodies. Mapping of sub-tidal and inter-tidal seagrass may be possible from moderate spatial resolution, multi-spectral image data (e.g. Landsat TM/ETM) in the Port Curtis area if the substrate is exposed or if the water column is shallow and clear enough for substrate to be visible. In addition, the substrate cover must be at a sufficient density that its reflectance signature is significantly different to that of the surrounding substrate. Further work is necessary to determine the level of seagrass density able to be detected on inter- and sub-tidal areas in Port Curtis using Landsat TM/ETM image data. Optimal conditions for implementing this type of mapping would occur during the lower tidal stage of spring tides, within the drier months of the year (May – September) – similar conditions to those already used for acquisition of aerial photography by the Gladstone Port Authority. These constraints would ensure minimal water cover and maximum water clarity over seagrass beds. Other problems associated with sensitivity and striping effects of the Landsat ETM sensor increased misclassification of seagrass beds. The trial of higher spectral, spatial and/or radiometric resolution satellite imaging sensors (e.g. Ikonos/Quickbird and ALI) may increase the range of water column conditions and density levels that seagrass could be mapped from.


Figure 68: Sub-tidal substrate cover types – Port Curtis sub-scene from Landsat 5 Thematic Mapper image captured at 09:30 on 21st August 1986.


Figure 69: Sub-tidal and inter-tidal substrate cover types – Port Curtis sub-scene from Landsat 5 Thematic Mapper image captured at 09:30 on 16th August 1990.


Figure 70: Sub-tidal and inter-tidal substrate cover types – substrate cover types –Port Curtis sub-scene from Landsat 5 Thematic Mapper image captured at 09:30 on 7th March 1995.


Figure 71: Sub-tidal substrate cover types – Port Curtis sub-scene from Landsat 7 Enhanced Thematic Mapper image captured at 09:45 on 24th July 2002.


Figure 72: Seagrass map for Port Curtis sub-scene from Queensland Department of Primary Industries and Fisheries.


2.4 Terrestrial Coastal Environment Mapping and Monitoring

The aim was to present demonstration products for mapping changes, between two successive dates, and trends over multiple image dates, in vegetation cover within the near coastal areas of the Fitzroy Estuary and Port Curtis/Gladstone areas.

2.4.1 Terrestrial Vegetation-Cover Mapping Port Curtis – Fitzroy Estuary

Vegetation Cover Mapping Process

The first stage in this process was to produce image based maps of Port Curtis and Fitzroy Estuary surrounds (Figure 1) showing density of vegetation cover. The type of vegetation present in the region has been mapped from aerial photographs for terrestrial and coastal communities by the Queensland Environmental Protection Agency Herbarium regional ecosystem mapping process (Accad et al. 2003). Coastal vegetation specifically, intertidal mangroves and saltmarsh vegetation have been mapped from Landsat TM and aerial photographs by the Queensland Department of Primary Industries and Fisheries Resource Condition and Trend Unit (Danaher and De Vries 2003, Bruinsma 2000, Bruinsma and Duncan 2000, Bruinsma and Danaher 2000). Input data to the mapping process consisted of the red and near-infrared bands of the following atmospherically corrected Landsat TM and ETM subscenes for Port Curtis and Fitzroy Estuary which had water bodies masked out using the inverse procedure outlined in Section 2.3.1: 21st August 1986; 16th August 1990; 7th March 1995; and 24th July 2002. The normalised difference vegetation index (NDVI) is computed for each land-based image pixel using the following equation, where NIR and red represent the apparent reflectance values contained in the red and NIR bands in each pixel:

NDVI = NIR – red (13) NIR + red The NDVI was established in 1974 and since then has been applied and assessed in a wide range of vegetated environments around the world. Although it is subject to limitations of non-linear variation in areas of very sparse cover and dark soils, and saturation in high-biomass tropical forests, it has been consistently shown to represent an integrated measurement of green projected foliage cover and green biomass (Elvidge and Chen 1994; Bannari et al. 1995; Huete et al. 1999; Huete et al. 2002). Due to the ratio function used to produce the image it also corrects for multiplicative scattering effects, i.e. variations in the amount of incident sunlight due to topographic or cloud shadow.

Vegetation Cover Maps The main trends shown in the NDVI maps presented in Figures 73 to 80 were tied to differences in the spatial distribution of vegetation communities with different structural forms and the level of clearance associated with human activities in the region. Targets providing high NDVI values in the Port Curtis scene were associated with coastal eucalypt forest on elevated areas and mangroves, with similar responses also being observed in the Fitzroy Estuary scene, which also had agriculture, pasture and riparian vegetation exhibiting very high NDVI levels. In contrast, low NDVI levels were associated with cleared areas, all artificial surfaces, urban and industrial sites, and saltpans in both the Port Curtis and Fitzroy Estuary scenes. Surface cover types with sparse or very dry vegetation cover, such as floodplain grasses, salt marsh, dry pasture, coastal sedge and open woodland exhibited moderate NDVI levels.


Figure 73: Normalised–difference-vegetation-index (NDVI) images for Fitzroy River Estuary sub-scenes from Landsat 5 Thematic Mapper image captured at 09:30 on 21st August 1986. Bright (dark) NDVI values indicate high (low) amounts of green vegetation cover.


Figure 74: Normalised–difference-vegetation-index (NDVI) images for Port Curtis sub-scenes from Landsat 5 Thematic Mapper image captured at 09:30 on 21st August 1986. Bright (dark) NDVI values indicate high (low) amounts of green vegetation cover.

Final Report Project PC2 Port Curtis - Remote Sensing


Figure 75: Normalised–difference-vegetation-index (NDVI) images for Fitzroy River Estuary sub-scenes from Landsat 5 Thematic Mapper image captured at 09:30 on 16th August 1990. Bright (dark) NDVI values indicate high (low) amounts of green vegetation cover.


Figure 76: Normalised–difference-vegetation-index (NDVI) images for Port Curtis sub-scenes from Landsat 5 Thematic Mapper image captured at 09:30 on 16th August 1990. Bright (dark) NDVI values indicate high (low) amounts of green vegetation cover.



Figure 77: Normalised–difference-vegetation-index (NDVI) images for Fitzroy River Estuary sub-scenes from Landsat 5 Thematic Mapper image captured at 09:30 on 7th March 1995. Bright (dark) NDVI values indicate high (low) amounts of green vegetation cover.


Figure 78: Normalised–difference-vegetation-index (NDVI) images for Port Curtis sub-scenes from Landsat 5 Thematic Mapper image captured at 09:30 on 7th March 1995. Bright (dark) NDVI values indicate high (low) amounts of green vegetation cover.



Figure 79: Normalised–difference-vegetation-index (NDVI) images for Fitzroy River Estuary sub-scenes from Landsat 7 Enhanced Thematic Mapper image captured at 09:45 on 24th July 2002. Bright (dark) NDVI values indicate high (low) amounts of green vegetation cover.


Figure 80: Normalised–difference-vegetation-index (NDVI) images for Port Curtis sub-scenes from Landsat 7 Enhanced Thematic Mapper image captured at 09:45 on 24th July 2002. Bright (dark) NDVI values indicate high (low) amounts of green vegetation cover.



2.4.2 Coastal Vegetation-Cover Change and Trend Detection Port Curtis 1990 – 2002

Overview of Coastal Environment Change Detection – Published Chapter Based on Work Completed as part of the Coastal CRC Project. Source: Phinn, S., Joyce, K., Scarth, P. and Roelfsema, C. (2004) The Role of Integrated Information Acquisition and Management in the Analysis of Coastal Ecosystem Change. In: Le Drew, E. and Richardson, L. (Eds) Remote sensing of coastal aquatic ecosystem processes. Remote Sensing and Digital Image Processing Series - Kluwer Academic Publishers. Change and trend detection in coastal ecosystems requires rigorous image pre-processing to ensure that the variable of interest is detected with sufficient signal to noise ratio. At a minimum, multi-temporal analysis requires sub-pixel precision georeferencing, atmospheric correction, multi-date normalization and ground-truthing for accuracy assessment (Palandro et al. 2001). These processes need to be explicitly considered within the framework used to select the remotely sensed data for the specific monitoring requirement (Phinn 1998). There has been little research on the effects of image mis-registration in aquatic environments on change detection accuracy; although research in terrestrial regions has found that significant and serious classification errors can be induced by a mis-registration of only one pixel (Townsend et al. 1992; Phinn and Rowland 2001). These registration errors will become increasingly significant in the move towards higher spatial resolution imagery. Correction of atmospheric effects is dependant on the analytical methods used in the change analysis. In many cases involving classification and change detection, atmospheric correction is unnecessary, as long as the training data and the data to be classified are in the same relative scale. Atmospheric correction is often unnecessary when using atmospherically resistant indices developed for the application of interest (Jensen 1996). Often atmospheric correction alone will not be adequate in images of aquatic environments due to whitecaps and/or sun glint, with the corrected images requiring additional empirical adjustment. Therefore, there is often no substantial benefit in performing an atmospheric correction compared to an empirical correction alone (Collins and Woodcock 1996; Andrefouet et al. 2001). When the processing is to derive change using semi-analytical modelling, corrections to a common radiometric scale are essential. (Song et al. 2001). Multi-date normalisation is used to minimise radiometric differences among images caused by changes in acquisition conditions, and require the use of reference and subject image pairs along with selected sample points. Normalisation methods include image regression, pseudo-invariant features, histogram matching, radiometric control set and no-change set determined from scattergrams (Tokola et al. 1999). Yang and Lo (2000) found that normalisation methods that used a large number of samples exhibited a better overall performance, but reduced the dynamic range and coefficient of variation of the images and therefore reduced the accuracy of image classification. The information requirements of the project and the environment of interest guide the choice of change or trend detection technique. No single change detection technique is suitable for the myriad of monitoring applications, with the various methods often giving differing map accuracy (Rogan et al. 2002). Change detection methods include direct image differencing, spectral index differencing, linear change enhancement techniques (e.g. selective principal components analysis), direct multi-date unsupervised classification, post-classification change differencing and decision tree analysis (Coppin and Bauer 1996; Mas 1999). Typically, these techniques are applied to imagery collected at two dates, with the differencing and linear change enhancement techniques resulting in a continuous map product that is subsequently thresholded to provide change classes. The classification approaches are either applied individually to each image; where the change can then be classed as change from one cover type to another, or to the entire image stack. In this case, the output classification will need careful interpretation to develop reliable change classes


(Jensen 1996). Trend detection methods typically involve the analysis of absolute values of some variable such as a vegetation index or chlorophyll or TSM concentration, and rely on a form of per-pixel time-series analysis through fitting of polynomial functions, Fourier or wavelet analysis (Ruiz Luna and Berlanga Robles 1999; Li and Kafatos 2000; Coppin et al. 2004). These deterministic trend detection models are advantageous, since they can be applied in the same way to a variety of similar trend detection situations, resulting in standardised reporting of the trend in the indicator of interest in different regions. Accuracy assessment is an important feature of mapping, not only as a guide to map quality and reliability, but also in understanding thematic uncertainty and its likely implications to the end user (Czaplewski 2003). Prior to image classification, calibration data must be sampled from appropriate areas, at an appropriate support size (Stehman and Czaplewski 1998). However, sampling for change detection is more challenging than that found in single-date approaches (Biging et al. 1998). Typically, a first step in this process is to highlight areas of change vs. no-change. This can be accomplished using an optimal threshold value based on similar spectral band comparisons between dates, vegetation indices or texture measures (Lunetta et al. 1998). To ensure appropriate sampling of no-change areas, the stratified adaptive cluster sampling (SACS) approach has been recommended (Brown and Manly 1998). SACS has particular utility for sampling disturbed locations (changed land-cover and land-use) because they usually represent a minor portion of the target population (most of the land area has not changed) and are often clustered (Rogan et al. 2002). Following classification, the accuracy of the change maps must be assessed. The total error in a thematic map is the sum of the following: (i) reference data errors; (ii) sensitivity of the classification scheme to observer variability; (iii) inappropriateness of the mapping process or the technological interpolation method; and (iv) general mapping error. General (total) map error conveys map quality, or ‘fitness for use’ by end users (Chrisman 1991). The conventional method of communicating ‘fitness of use’ for map users is the confusion or error matrix (Richards 1996). The error matrix summarizes results by comparing a primary reference class label to the map land-cover or land-use class for the sampling unit and presents errors of inclusion (commission errors) and errors of exclusion (omission errors) in a classification.


Change Detection Methods and Results To identify areas in the Port Curtis and Fitzroy Estuary where coastal vegetation cover had undergone specific changes between two successive image dates, an image differencing approach was used for change detection. The date to date image differencing approach assumes the successive images are geometrically aligned so the same area of land is compared over time; the NDVI pixel value from the earlier image is then subtracted from the NDVI pixel of the later image (e.g. NDVI_2002 – NDVI_1995). Areas of high positive ∆NDVI values indicate an increase in green vegetation cover, e.g. revegetation or plant growth, in the image pixel between the first and second image. Areas of high negative ∆NDVI values indicate a decrease in green vegetation cover, e.g. canopy thinning or clearing, in the image pixel between the first and second image. Thresholds (Tables 7 and 8) were identified from visual inspection of the ∆NDVI images and original images to delimit areas with significantly high positive and negative ∆NDVI values. To examine clearing (high negative ∆NDVI change) and regrowth (high positive ∆NDVI change) thresholds were identified from visual inspection of the ∆NDVI images and original images. Results from this analysis are presented spatially as difference images and summary statistics for three period, 1986-1990, 1990-1995 and 1995 – 2002 (Figures 81 - 86 and Tables 7 – 10). Observed changes in vegetation cover can be explained by a combination of natural and anthropogenic factors, including tidal stage, rainfall and clearance of native or regrowth vegetation. The 1990 and 1995 images were captured at or close to low tide, while the 1986 and 2002 images were captured close to high tide. As a result of these varying tidal levels the 1986-1990 (1995-2002) ∆NDVI images represented a change from high to low (low to high) tidal conditions. The 1986-1990 ∆NDVI image contained areas of increase in vegetation cover around all coastal, estuarine and riverine boundaries due to exposure of inter-tidal areas in 1990, while the opposite (vegetation cover decrease) was observed in the 1995-2002 ∆NDVI image, due to inundation of inter-tidal areas in 2002. Effects of variations in seasonal rainfall are harder to separate in the images, although the overall levels of ∆NDVI for clearing, regrowth and stability provide an indication of rainfall levels. Both 1990 and 2002 appear as wetter periods preceding image acquisition, while 1986 and 1995 appear to be drier. Consistent patterns in the distribution and type of ∆NDVI were observed for each of the three change periods over both the Port Curtis and Fitzroy Estuary sub-scenes. The main difference being the gradual decrease in vegetation cover in the Fitzroy Estuary sub-scene and relative stability in the amount of vegetation gain and loss in Port Curtis (Tables 9 and 10). The type of vegetation change was identified using the ∆NDVI image as a guide to key locations, and then each location was checked in the input multi-spectral images (Figures 13-20) to determine the likely land-cover or land-use activity forcing the change. Areas of complete vegetation removal were evident in each of the three ∆NDVI images (Figures 82, 84 and 86) and corresponded to sites where industrial facilities, housing estates and associated infrastructure were being expanded in Port Curtis, particularly about Gladstone. Some of the projected loss in coastal wetland areas may have been due to hailstorm damage. Similar trends were observed in the Fitzroy Estuary ∆NDVI images, but the dominant vegetation loss was due clearing to expand agricultural and rangeland activities, and urban development in coastal areas such as Yeppoon. Areas of significant vegetation cover increase were not widespread in the Port Curtis ∆NDVI image, and often conformed to coastal boundaries and wetlands during lower tide periods. In the Fitzroy Estuary sub-scene, areas of vegetation cover increase were mainly apparent on floodplain wetland, saltmarsh or former saltpan areas mainly associated with increased rainfall and freshwater flows through the system. In the transition to wetter periods (1986-1990 and 1995-2002), larger areas of each image displayed significant increases in vegetation cover. The results obtained here, and the simplistic interpretation and validation, demonstrate the potential of NDVI for mapping and monitoring changes in coastal vegetation cover and condition. As the NDVI images used represent snapshots of the environment at a set point in time, the ∆NDVI images represent the difference between those two points in time and may not represent the actual change in vegetation cover that occurred in the time between the two images. For future


application of this work it is essential that the image data used are corrected to at-surface or apparent reflectance to enable comparison of NDVI values over time. Figure 81: Change in Normalised-difference-vegetation-index (NDVI) images for Fitzroy Estuary sub-scenes between 21st August 1986 and 16th August 1990. Red indicates clearing (high negative ∆NDVI) and Green indicates regrowth (high positive ∆NDVI).


Figure 82: Change in Normalised–difference-vegetation-index (NDVI) images for Port Curtis sub-scenes between 21st August 1986 and 16th August 1990. Red indicates clearing (high negative ∆NDVI) and Green indicates regrowth (high positive ∆NDVI).



Figure 83: Change in Normalised-difference-vegetation-index (NDVI) images for Fitzroy River Estuary sub-scenes between 16th August 1990 and 7th March 1995. Red indicates clearing (high negative ∆NDVI) and Green indicates regrowth (high positive ∆NDVI).


Figure 84: Change in Normalised-difference-vegetation-index (NDVI) images for Port Curtis sub-scenes between 16th August 1990 and 7th March 1995. Red indicates clearing (high negative ∆NDVI) and Green indicates regrowth (high positive ∆NDVI).



Figure 85: Change in Normalised-difference-vegetation-index (NDVI) images for Fitzroy River Estuary sub-scenes between 7th March 1995 and 24th July 2002. Red indicates clearing (high negative ∆NDVI) and Green indicates regrowth (high positive ∆NDVI).


Figure 86: Change in Normalised-difference-vegetation-index (NDVI) images for Port Curtis sub-scenes between 7th March 1995 and 24th July 2002. Red indicates clearing (high negative ∆NDVI) and Green indicates regrowth (high positive ∆NDVI).



Table 7: Fitzroy River Estuary NDVI difference images- threshold NDVI change levels used to identify pixels with vegetation gain, loss or no change.

Change Dates NDVI Change Vegetation Loss

NDVI Change No Change

NDVI Change Vegetation Increase

1986 – 1990 1990 image (General trend - more standing water and increas-ed vegetation cover in 1990, with lower tidal levels)

-20 (< -50 = inundation)

-19 -- 30 > 30

1990 – 1995 1995 image (General trend – less standing water)

-35 -34 --37 > 37

1995 – 2002 2002 image (General trend - more standing water, with higher tidal levels)

-29 -28 -- 41 > 41

Table 8: Port Curtis NDVI difference images- threshold NDVI change levels used to identify pixels with vegetation gain, loss or no change.

Change Dates NDVI Change Vegetation Loss

NDVI Change No Change

NDVI Change Vegetation Increase

1986 – 1990 1990 image (General trend - more standing water and increased vegetation cover in 1990, with lower tidal levels)

-18 -17 --25 > 25

1990 – 1995 1995 image (General trend – less standing water)

-36 -35 -- 20 > 20

1995 – 2002 2002 image (General trend - more standing water, with higher tidal levels)

-38 -37 -- 48 > 48


Table 9: Fitzroy Estuary areas of vegetation change (based on Table 7 thresholds) in km2

Change Dates Vegetation Loss No Change Vegetation Increase

1986 – 1990

102.59 5972.79 119.96

1990 – 1995

105.84 6046.94 42.52

1995 – 2002

279.14 5894.34 21.83

Table 10: Port Curtis areas of vegetation change (based on Table 8 thresholds) in km2

Change Dates Vegetation Loss No Change Vegetation Increase

1986 – 1990 1990 image

33.81 1807.06 39.46

1990 – 1995 1995 image

38.06 1801.48 35.45

1995 – 2002

12.63 1549.158 10.54


Trend Detection Methods and Results The ∆NDVI approach outlined in the section above enables change in vegetation cover to be detected between two successive image dates, however, it cannot be used to summarise change over multiple dates of images where the information of interest may be the longer term trends in NDVI over time. The ability to produce a single image based map depicting the temporal trajectory of a set environmental parameter, e.g. vegetation cover as represented by NDVI, would fit into environmental indicator monitoring requirements as the spatial and temporal tends in the value of an indicator could be assessed. A demonstration project was developed to test the trend detection approach on a subset of the Port Curtis Landsat scenes around Gladstone. The text below explains the necessary image corrections, trend detection algorithm and validation from aerial photography captured at the same time as the Landsat images. Application of the trend detection technique requires all images being used to be geometrically and atmospherically corrected. The geometric and atmospheric correction approaches were described in Section 2.2.2. An alternative atmospheric correction approach was applied to provide a relative correction between the five image subsets used this work. The empirical-line correction approach or pseudo-invariant feature approach works by selecting a base image (24 July 2002) and then normalising each image to the reflectance values in the base image (Research Systems 2004; Schott 1997). This is done by selecting high, medium and low reflectance ground targets in each spectral band, where their reflectance would not be expected to change over time, i.e. pseudo invariant features (PIF). The PIF reflectance values in each scene are then plotted against the corresponding band from the base image, and a linear regression equation derived to convert the scene reflectance values to match those of the base image (Figure 87). Figure 87: Pseudo invariant features and calibration equations for bands 1-6 (blue, green, red, near infrared, mid-infrared1 and mid-infrared2) for Gladstone Landsat TM/ETM sub-scenes.

Band 1

Band 2


Band 3

Band 4

Band 5

Band 6


Once all of the Landsat images were geometrically and atmospherically corrected the NDVI transformation (Equation 13) was applied to each image to produce vegetation cover surfaces for 1990, 1995, 1997, 2001 and 2002. The NDVI will be used here as a surrogate environmental indicator for coastal vegetation cover. To produce a trend-image for NDVI all five NDVI images were combined into a single image stack where Band 1 = NDVI-1990 and Band 5 = NDVI-2002. To assign each image pixel in the output 1990-2002 NDVI trend image to one of the following categories (linear increase, linear decrease, recovery or deterioration) a regression analysis was first performed on each pixel in the NDVI image stack, fitting a line to five NDVI values (1990, 1995, 1997, 2001 and 2002) (Figure 88). NDVI trend image pixels were then assigned a label based on the following conditions:

a) regions where the calculated slope is greater than 3σ from zero, and labelling them as areas of significant increase

b) regions where the calculated slope is less than -3σ from zero, and labelling them as areas of significant decrease

c) regions unclassified in a) and b) where the calculated curvature is greater than 3σ from zero, and labelling them as areas of significant recovery

d) regions unclassified in a) and b) where the calculated curvature is less than -3σ from zero, and labelling them as areas of significant deterioration

e) no significant change for the remaining areas Each image pixel in the NDVI trend image summarised the changes over five consecutive Landsat TM/ETM images. To remove spurious changes and provide a more spatially coherent image map a smoothing (modal) filter was applied to the output NDVI trend image prior to completing accuracy assessment.


Figure 88: Typical types of NDVI trends observed in the Gladstone area 1990-2002 and the resulting fitted trend lines for increasing, decreasing, recovery and deterioration. The accuracy of detected trends was determined by applying standard image processing error assessment procedures to a time series of aerial photographs covering two sample areas within the Gladstone subset (Foody 2003, Jensen 1996). Aerial photographs used in the assessment were obtained from the Queensland Land Information Centre (Department of Natural Resources, Mines and Energy):

- colour 1:50000 BPA Urangan-St Lawrence Run H7 Photo 130 0107 1996 1409 Film Q5402 at 1200dpi ;

- Band W 1:12000 BPA Urangan-St Lawrence Run 19A Photo 184 1510 1992 Film Q4942 at 300dpi;

- colour 1:12000 BPA Urangan-St Lawrence Run 19A Photo 058 2605 2001 1545 Film Q5915 at 300dpi;

- Band W 1:25000 Gladstone 9150 Run 10 Photo 040 2809 1989 Film Q4831

The scanned images were registered to the smallest scale photograph (1996) and then to the Landsat ETM image from 2002. Twelve random sample points were placed in each of the five NDVI trend classes. Each sample point was then checked on the corresponding locations on the sequence of aerial photographs to assign it a reference trend in vegetation cover as: increase; decrease; deterioration; recovery; or no change. A standard error matrix was then constructed to summarise results and define overall trend mapping accuracy and individual accuracies for each type of NDVI trend. Overall, the resultant NDVI trend image (Figure 91), its smoothed version (Figure 92) and adjusted overall accuracy (71.2%) indicate that the trend detection approach may be a viable means of summarising multi-date image data that can be used to monitor changes in environmental indicators over time. Environmental features exhibiting significant linear changes, i.e. significant clearing or regrowth/restoration (Figures 90 and 92), corresponded to areas of bi-temporal change in the ∆NDVI images of Port Curtis (Figures 82, 84 and 86), were able to be associated with known types of land cover change (Figure 93) and were accurately defined (Tables 11-14). Non-linear changes were not identified as accurately (Tables 11-14) due to the nature and magnitude of changes causing them, but still produced realistic depiction of multi-date environmental changes. In some cases, the labelling of this change as recovery may also be misleading, e.g. clearance of

NDVI Change

-0.1

0

0.1

0.2

0.3

0.4

1990 1992 1994 1996 1998 2000 2002Year

ND

VI

Mangrove Site North Clearing Mangrove Recovery Eastern Mangrove RegrowthShale Oil Clearing Urban Thicken then Clearing Caliopie Ck Bend Thinned

Linear (Shale Oil Clearing) Poly. (Mangrove Recovery) Linear (Eastern Mangrove Regrowth)Linear (Mangrove Site North Clearing) Poly. (Urban Thicken then Clearing) Poly. (Caliopie Ck Bend Thinned)


native vegetation by human activity or natural disturbance may be replaced by non-native invasive species. Further work is required on the detection and labelling of these types of trends. Figure 89: Linear trends in NDVI from five Landsat TM/ETM images of the Gladstone area collected in 1990, 1995, 1997, 2001 and 2002.

305000

305000

310000

310000

315000

315000

320000

320000

325000

325000

7355000 7355000

7360000 7360000

7365000 7365000

7370000 7370000

Significant Linear ChangeDecreaseIncrease

Figure 90: Linear trends in NDVI from five Landsat TM/ETM images of the Gladstone area collected in 1990, 1995, 1997, 2001 and 2002, after the image has been subjected to a modal smoothing filter.

315000

315000

320000

320000

7360000 7360000

7365000 7365000

Significant Linear ChangeDecreaseIncrease


Figure 91: Overall trends in NDVI from five Landsat TM/ETM images of the Gladstone area collected in 1990, 1995, 1997, 2001 and 2002.

305000

305000

310000

310000

315000

315000

320000

320000

325000

325000

7355000 7355000

7360000 7360000

7365000 7365000

7370000 7370000

Significant ChangeDecreaseDeteriorationRecoveryIncrease

Figure 92: Linear trends in NDVI from five Landsat TM/ETM images of the Gladstone area collected in 1990, 1995, 1997, 2001 and 2002, after the image has been subjected to a modal smoothing filter.

315000

315000

320000

320000

7360000 7360000

7365000 7365000

Significant ChangeDecreaseDeteriorationRecoveryIncrease

Figure 93: Interpretation of causes in observed NDVI changes and trends from five Landsat TM/ETM images of the Gladstone area collected in 1990, 1995, 1997, 2001 and 2002..


Shale Oil Clearing

Caliopie Ck Bend Thinned

Mangrove Regrowth

Urban thickening then Clearing

Mangrove Clearing/Damage

Mangrove Recovery

Table 11: Error matrix for NDVI trend detection using aerial photographs as a reference source, (a) sample point counts and (b) percentage point counts.

(a) Air Photo Interpretation Classification Decrease Deterioration Increase No Change Recovery Total Decrease 11 1 0 1 0 13 Deterioration 0 9 0 2 0 11 Increase 0 0 4 0 1 5 No Change 0 2 0 13 3 18 Recovery 0 0 2 1 8 11 Total 11 12 6 17 12 58 (b) Air Photo Interpretation Classification Decrease Deterioration Increase No Change Recovery Total Decrease 19.0 1.7 0.0 1.7 0.0 22.4 Deterioration 0.0 15.5 0.0 3.4 0.0 19.0 Increase 0.0 0.0 6.9 0.0 1.7 8.6 No Change 0.0 3.4 0.0 22.4 5.2 31.0 Recovery 0.0 0.0 3.4 1.7 13.8 19.0 Total 19.0 20.7 10.3 29.3 20.7 100.0


Table 12: Error matrix for NDVI trend detection using aerial photographs as a reference source – Users Accuracy.

Air Photo Interpretation Classification Decrease Deterioration Increase No Change Recovery Total Decrease 84.6 7.7 0.0 7.7 0.0 100.0 Deterioration 0.0 81.8 0.0 18.2 0.0 100.0 Increase 0.0 0.0 80.0 0.0 20.0 100.0 No Change 0.0 11.1 0.0 72.2 16.7 100.0 Recovery 0.0 0.0 18.2 9.1 72.7 100.0 Total 19.0 20.7 10.3 29.3 20.7 100.0

Table 13: Error matrix for NDVI trend detection using aerial photographs as a reference source -Producers Accuracy.

Air Photo Interpretation Classification Decrease Deterioration Increase No Change Recovery Total Decrease 100.0 8.3 0.0 5.9 0.0 22.4 Deterioration 0.0 75.0 0.0 11.8 0.0 19.0 Increase 0.0 0.0 66.7 0.0 8.3 8.6 No Change 0.0 16.7 0.0 76.5 25.0 31.0 Recovery 0.0 0.0 33.3 5.9 66.7 19.0 Total 100.0 100.0 100.0 100.0 100.0 100.0

Table 14: Summary of error and accuracy results for NDVI trend detection using aerial photographs as a reference source Overall Accuracy is 77.6% Kappa Statistic is 71.2% with a significance of 99.99%

Change Type Producers AccuracyDecrease 100%Deterioration 75%No Change 77%Recovery 67%Increase 67%

Change Type Users AccuracyDecrease 85%Deterioration 82%No Change 72%Recovery 73%Increase 80%


3. Conclusions and Recommendations Conclusions

(1) Commercially available, multi-spectral, moderate spatial resolution satellite image data can

be used to produce maps of aquatic and terrestrial environmental parameters relevant to coastal ecosystem health monitoring in the Fitzroy Estuary and Port Curtis coastal areas.

The project team was able to apply the methods developed in the Moreton Bay project to map the following water quality parameters from commercial satellite image data (Landsat 5 Thematic Mapper and Landsat 7 Enhanced Thematic Mapper) collected over Port Curtis and Fitzroy Estuary in 1986, 1990, 1995 and 2002: - coloured dissolved organic material (CDOM); - total suspended matter TSM measured as tripton; and - Secchi transparency depth Distinct optical domains, or zones of water colouration, were evident in the Fitzroy Estuary and Port Curtis water quality maps, and appeared to be controlled by tidal currents and stream flow, along with surface and sub-surface aquatic vegetation. Due to the turbid nature of both Port Curtis and Fitzroy Estuary waterways, substrate mapping was not able to be completed using the dates of Landsat TM/ETM image that we had access to. Substrate mapping may be possible in Port Curtis under conditions when the tide is low and stream-flow is low, resulting in exposed inter-tidal seagrass beds and visible sub-tidal beds in very shallow and clear water. Terrestrial environments were also examined as part of this work, focussing on the use of “vegetation indices” to map vegetation cover throughout the immediate coastal areas of Fitzroy Estuary and Port Curtis.

(2) The image processing approach developed for optically complex waters in Moreton Bay

can be applied successfully to other coastal environments with different specific inherent optical properties (Phinn et al. 2004).

The optimised matrix inversion model used to estimate the concentration of organic and inorganic water column constituents was the same model as was developed for Moreton Bay. In this case the model was driven by a set of SIOP collected during fieldwork in 2002 in Fitzroy Estuary and Port Curtis areas. Realistic maps of CDOM, TSM and Secchi depth were produced, with the model not operating accurately in areas where substrate was visible or exposed.

(3) Multi-date image analysis techniques can be used to produce maps depicting trends in the

state or condition of an environmental parameter over time.

To present the results from multiple dates of vegetation index images in one image that would summarise the change in a pixel’s vegetation cover over time, a trend detection approach was developed and applied to a five date series of vegetation index images over the Gladstone area. This approach was applied and validated successfully using multidate Landsat TM/ETM based NDVI images and aerial photography as a reference source. Areas of increase or decrease in vegetation cover were identified along with areas that experienced a recovery or deterioration in vegetation cover. The approach could also be applied to multi-date substrate cove type and water quality parameter image maps.


(4) Future work can be used to take the demonstration projects presented in this work to operational, accurate and cost-efficient environmental mapping and monitoring programs

Although the image data used here was the most commonly available commercial product when this project commenced, there are a range of different sensors now operating that could provide the same and possibly more information. For water quality monitoring there is a need to enable a high-frequency (e.g. daily) data collection to monitor changes in water quality distribution as its controlling processes act at this scale. Products from the MODIS and MERIS sensors should be examined as new data sources, due to their high temporal, spectral and radiometric resolutions, and design specifically for some aspects of ocean colour. The same comment applies to terrestrial vegetation monitoring, where MODIS products may also be useful. Potential improvement to seagrass mapping in this region requires assessment of the capabilities of high spatial and radiometric resolution satellite images, along with airborne and satellite hyperspectral imaging sensors. In each case these sensors may improve mapping ability in shallow waters and/or low density seagrass cover.

Considerations for Future Use of Remote Sensing (Modified from Moreton Bay Report): Often the focus on what remote sensing can directly detect, map and monitor, obscures another level of products that may be delivered. It is necessary to make a distinction between three time-scales in which the remote sensing derived information needs to be delivered to an end-user :

• Near real time delivery; • Within a week of data capture; and • Within a month of data capture.

Spatial information products based on remotely sensed data can be used to deliver information relevant to coastal ecosystem monitoring and management at the three time scales listed above. Considerations for collecting and delivering this remotely sensed information relevant to each of these timeframes include:

• Does data need to be averaged: on a daily/weekly/monthly basis? • Is trend detection required? • Is there an interest in integrating with catchment hydrology models, e.g. run-off models? • Is there an interest in integration with hydrodynamic models? e.g. to understand the

distribution of organic micro pollutants adsorbed to suspended matter or to dissolved organic matter?

Whatever suite of image-based products are finally chosen, for operational cost-effective remote sensing data based mapping, it is essential to create an integrated processing chain. The basis for such a processing chain was established in this project. Application of the processing chain to an image data set converts it to a format where it can be directly compared to previous images of the same area and used to derive calibrated maps of biophysical variables, e.g. supra/inter/sub-tidal products. The following list of issues needs to be considered before embarking on such an ambitious goal. These questions need addressing by the coastal and water management authorities as they will be the end-users and product uptakes will only take place if these issues are clear:

• Are we discussing and offering the correct set of products? • What do the coast and water management agencies with responsibility for the Fitzroy River

Estuary and Port Curtis, their waterways and coasts need in the form of spatially and temporally dense information?

Specific issues to be discussed and determined are:

• Does remote sensing based information add to, enhance or replace existing methods for measuring the coastal and water environment?


• Is remote sensing based information delivery cost-effective or can it can it be made cost-effective? As an example, current processing costs for one Landsat scene for the Fitzroy Estuary – Port Curtis area for water quality would cost about 8 hours work time plus AUD$1200 for remote sensing image acquisition (provided all thematic data is available). Is this competitive with sending out boats and field teams and performing laboratory analyses to reach the same level of information?

• Is retrospective processing relevant or not? Based on Landsat data availability analyses it is possible to process five to 10 images a year going back to 1988 easily (back to 1984 via US archives @30 m resolution is possible-there is even a possibility of going back 1973 @ 80 m resolution!).

• Is higher spatial resolution image data (pixels < 10m x 10m) useful for coastal monitoring

and management? In the case of the Fitzroy Estuary – Port Curtis area it will be necessary to discuss the issues outlined above amongst many of the players involved in order to get a consensus on immediate, mid-term and long-term requirements and orders of priority. Possible players in such a consortium approach to adopting remote sensing derived information for Fitzroy Estuary – Port Curtis areas could be:

• Environmental Protection Agency • Queensland Department of Primary Industries and Fisheries • Queensland Department of Natural Resources, Mines and Energy • Fitzroy Basin Association • Gladstone Port Authority • Coastal CRC • UQ-BRG and CMS • CSIRO Land W/MR/CMIS

These recommendations are intended to stimulate a dialogue between data providers, science and applications oriented groups and potential end users of remote sensing derived information.


4. References Aas, E. (1997) Two-stream irradiance model for deep waters, Applied Optics, 26:2095-2101. Accad, A., Neldner, V.J., Wilson, B.A. and Niehus, R.E. (2003). Remnant vegetation in Queensland: analysis of remnant vegetation 1997-1999-2000-2001, including regional ecosystem information. Queensland Herbarium, Environmental Protection Agency, Brisbane. Brando, V.E. and Dekker, A.G. (2003) Satellite hyperspectral remote sensing for estimating estuarine and coastal water quality. IEEE Transactions on Geosciences and Remote Sensing, 41(6):1378-1387. Bruinsma, C. (2000). Queensland Coastal Wetland Resources: Sand Bay to Keppel Bay. Information Series QI00100. Department of Primary Industries, Queensland, Brisbane, 94 pp. Bruinsma, C. and Duncan, S. (2000). Queensland Coastal Wetland Resources: the Northern Territory Border to the Flinders River. Information Series QI00099. Department of Primary Industries, Queensland, Brisbane, 72 pp. Bruinsma, C. and Danaher, K. (2000). Queensland Coastal Wetland Resources: Round Hill Head to Tin Can Inlet. Information Series QI99081. Department of Primary Industries, Queensland, Brisbane, 101 pp. Bukata, R., Jerome, J., Kondratyev, K., and Pozdnyakov, D. (1995) Optical properties and remote sensing of inland and coastal waters. CRC Press, Boca Raton, 362p. Campbell, N.A., Mitchell, L.G. and Reece, J.B. (1997), Biology, Concepts and Connections, 2nd Ed., The Benjamin/Cummings Publishing Company, USA, pp.89-106. Clementson, L. A., Parslow, J. S., Turnbull, A.R., McKenzie, D.C. and Rathbone, C.E. 2001 Optical properties of waters in the Australasian sector of the Southern Ocean, Journal of Geophysical Research 106, pp 31611-31625. Danaher, K. and de Vries, C. (2003). Queensland coastal wetland resources of the Curtis Coast region: Raglan Creek to Round Hill Head. Department of Primary Industries, Queensland, Brisbane, 62 pp. De Haan,J.F., Kokke, J. M. M., Hoogenboom, H.J. and Dekker, A.G. (1997) An integrated toolbox for processing and analysis of remote sensing data of inland and coastal waters-atmospheric correction. In: Proceedings of the Fourth International Conference: Remote Sensing for Marine and Coastal Environments, Michigan, USA, 1997. Dekker, A.G. (1993) Detection of optical water quality parameters for eutrophic waters by high resolution remote sensing, Faculty of Earth Sciences. 1993, Vrije Universiteit: Amsterdam. p. 1-240. Dekker, A.G. and Peters, S.W.M. (1993) The use of the Thematic Mapper for the analysis of eutrophic lakes: A case study in The Netherlands, International Journal of Remote Sensing, 14: 799-822. Dekker, A.G., Peters, S.W.M., Rijkeboer, M. and Berghuis, H. (1999) Analytical processing of multitemporal SPOT and Landsat images for estuarine management in Kalimantan, Indonesia. In: G. Nieuwenhuis, R. Vaughan and M. Molenaar Eds. Operational Remote Sensing for Sustainable Development. AA. Balkema, Rotterdam, pp. 315-323.


Dekker, A.G., Vos, R. and Peters S.W.M. (2001a) Comparison of remote sensing data, model results and in-situ data for total suspended matter in the southern Frisian Lakes. The Science of the Total Environment, 268: 197-214. Dekker, A.G., Brando, V.E., Anstee, J.M., Pinnel, N., Kutser, T., Hoogenboom, H.J.., Peters, S.W.M., Pasterkamp, R., Vos, R., Olbert, C. and Malthus, T.J. (2001b) Imaging spectrometry of water. (2001) In: F.D. van der Meer, S. and M. de Jong Eds. Remote Sensing and Digital Image Processing Volume 4: Imaging Spectrometry, Ch. 11, Kluwer Publishers, Dordrecht. Dekker, A. G., Anstee, J. M., and Brando, V. E. (2005) Retrospective seagrass change detection in a shallow coastal tidal Australian lake: Mapping seagrasses in Wallis Lake : Remote Sensing of Environment (in press). Evans, K.G., Stephens, A.W. and Shorten, G.G.,(1992), Quaternary sequence stratigraphy of the Brisbane River delta, Moreton Bay, Australia., Marine Geology., 107: 61 79. Gordon, H. and Morel, A. (1983) Remote assessment of ocean colour for interpretation of satellite visible imagery. A review. Springer, New York. Hoogenboom, H.J. Dekker, A.G. and De Haan, J.F. (1998) Retrieval of chlorophyll and suspended matter in inland waters from CASI data by matrix inversion, Canadian Journal of Remote Sensing, 24:144-152. IOCCG (2000) Remote sensing of ocean colour in coastal and other optically-complex waters, Sathyendrannath, S. (Ed.) Reports of the International Ocean Colour Coordinating Group, No.3, IOCCG, Dartmouth, Canada. Kirk, J. (1994) Light and Photosynthesis in aquatic environments. 2nd Edition. Cambridge University Press, Melbourne. Kirk, J. (1997) Point-source integrating-cavity absorption meter: theoretical principles and numerical modelling, Applied Optics, vol. 36, pp. 6123-6127. Lee Z., Kendall, L.C., Chen, R.F. and Peacock, T.G. (2001), Properties of the water column and bottom derived from Airborne Visible Imaging Spectrometer (AVIRIS) data, Journal of Geophysical Research, 106:11639 - 11652. McEwan, J., Gabric, A. and Bell, P.R.F. (1998) Water quality and phytoplankton dynamics in Moreton Bay, south-eastern Queensland. II. Mathematical modelling. Marine and Freshwater Research, 49:227-239. Malthus, T., and Mumby, P. (2003) Remote sensing of the coastal zone: an overview and priorities for future research. International Journal of Remote Sensing, 24(13):2805-2815. Maffione, R.A. and Dana, D.R. (1997) Recent measurements of the spectral backward-scattering coefficient in coastal waters, presented at Ocean optics XIII, Bellingham, USA. Mobley C. D. (1994), Light and water; Radiative transfer in natural waters. London: Academic Press. Morel, A. and Prieur, L. (1977) Analysis of variations in ocean colour. Limnology and Oceanography, 22:709-722. Phinn, S. Roelfsema, C. Scarth, P., Dekker, A.G., Brando, V.E., Anstee, J.M. and Marks, A. (2005) An integrated remote sensing approach for adaptive management of complex coastal waters. Final Report – Moreton Bay Remote Sensing Tasks (MR2). Phinn, S. and Dekker, A.G (eds), Published by the CRC for Coastal Zone, Estuary and Waterway Management, Indooroopilly, Qld, Australia.


Phinn, S., Dekker, A.G., Brando, V. and Roelfsema, C. (2005) Mapping Water Quality and Substrate Cover in Optically Complex Coastal and Reef Waters: An Integrated Approach. In: Hutchings, P.A., Haynes, D. (Eds.), Proceedings of Catchment to Reef: Water Quality Issues in the Great Barrier Reef Region Conference. Marine Pollution Bulletin, 51: 459-469. Press W. H., Flannery B. P., Teukolsky S. A., Vetterling W. T. (1992), Numerical Recipes in C: The Art of Scientific Computing 2nd Edition, Cambridge University Press, 994 pages. Research Systems Inc. (2004) ENVI 4.1 – Environment for Visualising Images. Research Systems Inc. Boulder, Colorado, U.S.A. Roelfsema, C.M., Dennison, W.C., Phinn S.R., Dekker A.D. and Brando V.,(2001), Remotely Sensing the extend of a cyano bacterial bloom L. majuscula in Moreton Bay, Australia In Proceedings of International Geosciences and Remote Sensing Symposium, Sydney, July 2001. Roelfsema, C.M., Phinn, S.R. and Dennison, W.C., (2002) Development and Implementation of Monitoring Program for L. majuscula (toxic cyanobacteria), In: Proceedings of the 11th Australasian Remote Sensing and Photogrammetry Conference, Brisbane, 2-6 September, CD-Rom Roesler, C. S., M. J. Perry, and K. L. Carder, Modeling in situ phytoplankton absorption from total absorption spectra in productive inland marine waters, 1989, Limnology and Oceanography, 34, 1510-1523. Walker, R.E., 1994. Marine light field statistics. Wiley series in pure and applied optics. Wiley, New York, 675 pp. Wettle, M., Brando, V. E., Dekker, A.G. (2004) A methodology for retrieval of environmental noise equivalent spectra applied to four Hyperion scenes of the same tropical coral reef. Remote Sensing of Environment (93): p 188 – 197. Zaneveld J. R. V., Kitchen, J.C. and Moore, C. The scattering error correction of reflecting-tube absorption meters, Proceedings of the SPIE, 2258: 44-55, 1994. Coastal Ecosystem Change References: Andrefouet, S., F. E. Muller Karger, E. J. Hochberg, C. Hu and K. L. Carder (2001). "Change detection in shallow coral

reef environments using Landsat 7 ETM+ data." Remote Sensing of Environment 78(1-2): 150-162. Bannari, A. D., D. Morin, F. Bonn and A. R. Huete (1995). "A Review of Vegetation Indices." Remote Sensing

Reviews 13: 95 - 120. Brown, J. A. and B. J. F. Manly (1998). "Restricted adaptive cluster sampling." Environmental and Ecological Statistics

5(1): 49-63. Chrisman, N. R. (1991). The error component in spatial data. Geographical information systems. et al. Longman/Wiley.

1: principles: 165-174. Collins, J. B. and C. E. Woodcock (1996). "An assessment of several linear change detection techniques for mapping

forest mortality using multitemporal Landsat TM data." Remote Sensing of Environment 56(1): 66-77. Coppin, P., Jonckheere, I., Nackaerts, K., Muys, B. and Lambin, E. (2004). "Digital change detection methods in

ecosystem monitoring: a review." International Journal of Remote Sensing 25(9): 1565-1596. Coppin, P. R. and M. E. Bauer (1996). "Digital change detection in forest ecosystems with remote sensing imagery."

Remote Sensing Reviews 13(3-4): 207-234. Czaplewski, R. L. (2003). "Can a sample of Landsat sensor scenes reliably estimate the global extent of tropical

deforestation?" International Journal of Remote Sensing 24(6): 1409-1412. Elvidge, C. D. and Z. Chen (1994). "Comparison of broad-band and narrow-band red versus near infra-red indices."

Remote Sensing of Environment 54(1): 38 - 48. Huete, A., K. Didan, T. Miura, E. P. Rodriguez, X. Gao and L. G. Ferreira (2002). "Overview of the radiometric and

biophysical performance of the MODIS vegetation indices." Remote Sensing of Environment 83: 195-213.


Huete, A., C. Justice and W. Leeuwen (1999). MODIS Vegetation Index (MOD 13). Algorithm Theoretical Basis Document. Version 3. Charlottesville; Tucson, Arizona, University of Virginia; University of Arizona.

Jensen, J. R. (1996). Introductory digital image processing: a remote sensing perspective. Second edition, Prentice Hall; Series in Geographic Information Science.

Li, Z. and M. Kafatos (2000). "Interannual variability of vegetation in the United States and its relation to El Nino/Southern Oscillation." Remote Sensing of Environment 71(3): 239-247.

Lunetta, R. S., J. G. Lyon, B. Guindon and C. D. Elvidge (1998). "North American landscape characterization dataset development and data fusion issues." Photogrammetric Engineering and Remote Sensing 64(8): 821-829.

Mas, J. F. (1999). "Monitoring land-cover changes: A comparison of change detection techniques." International Journal of Remote Sensing 20(1): 139-152.

Palandro, D., S. Andrefouet, F. E. Muller Karger and P. Dustan (2001). Coral reef change detection using Landsats 5 and 7: a case study using Carysfort Reef in the Florida Keys. IGARSS 2001. Scanning the Present and Resolving the Future. Proceedings. IEEE 2001 International Geoscience and Remote Sensing Symposium Cat. No.01CH37217. 2001, IEEE, Piscataway, NJ, USA: 625-7 vol.2.

Phinn, S. and T. Rowland (2001). "Geometric misregistration of Landsat TM image data and its effects on change detection accuracy." Asia-Pacific Remote Sensing Journal 14: 41-54.

Phinn, S. R. (1998). "A framework for selecting appropriate remotely sensed data dimensions for environmental monitoring and management." International Journal of Remote Sensing 19(17): 3457-3463.

Richards, J. A. (1996). "Classifier performance and map accuracy." Remote Sensing of Environment 57(3): 161-166. Rogan, J., J. Franklin and D. A. Roberts (2002). "A comparison of methods for monitoring multitemporal vegetation

change using thematic mapper imagery." Remote Sensing of Environment 80(1): 143-156. Ruiz Luna, A. and C. A. Berlanga Robles (1999). "Modifications in coverage patterns and land use around the

Huizache-Caimanero lagoon system, Sinaloa, Mexico: A multi-temporal analysis using LANDSAT images." Estuarine Coastal and Shelf Science 49(1): 37-44.

Song, C., C. E. Woodcock, K. C. Seto, M. P. Lenney and S. A. Macomber (2001). "Classification and change detection using Landsat TM data: When and how to correct atmospheric effects?" Remote Sensing of Environment 75(2): 230-244.

Stehman, S. V. and R. L. Czaplewski (1998). "Design and analysis for thematic map accuracy assessment: Fundamental principles." Remote Sensing of Environment 64(3): 331-344.

Tokola, T., S. Lofman and A. Erkkila (1999). "Relative calibration of multitemporal landsat data for forest cover change detection." Remote Sensing of Environment 68(1): 1-11.

Townshend, J. R. G., C. O. Justice, C. Gurney and J. McManus (1992). "The impact of misregistration on change detection." IEEE Transactions on Geoscience and Remote Sensing 30(5): 1054-60.

Yang, X. and C. P. Lo (2000). "Relative radiometric normalization performance for change detection from multi-date satellite images." Photogrammetric Engineering and Remote Sensing 66(8): 967-980.


5. Acknowledgements

Additional support for the CSIRO and UQ-BRG team’s instrumentation, provided by Dr Alistair Melzer and his research unit at Central Queensland University, along with Bob Noble and Bob Packett from Department of Natural Resources, Mines and Energy in Rockhampton The University of Queensland School of Geography, Planning and Architecture provided image processing and GIS software and support for the project. Phil Ford and Ian Webster from CSIRO and Peter Baddily and other support staff at the Brisbane Bureau of Meteorology for provision of rainfall, stream discharge and radiosonde profile data.


6. Appendices:

Appendix 1 Communication of Results by the Remote Sensing Team

2001 • Presentation of project overview at a number of public forum, including:

- Moreton Bay Research Station Open Weekend, November 18, 2000 - EHMP meeting at Pine Rivers, January 31, 2001. - Stakeholders meeting Gladstone 12 October 2000 - Workshop Stradbroke Island Theme 5

• Preparation of brochure for National Stakeholders Advisory Council on the application of Remote Sensing in the CRC.

• Interview for ABC/CSIRO Radio on project for Science Bytes • Short personal interview article in Australasian Science, Vol. 22(1):11 by S. Luntz on

“Mapping ecosystems by Radar.” • Invitation to Dr Dekker and Dr Phinn to prepare a special session on Mapping and

monitoring aquatic ecosystems as part of the largest remote sensing conference in the world, The International Geosciences and Remote Sensing Symposium, to be held in Sydney in July 2001. There will be four papers presented at this conference based on work we have started within the CRC.

• Presentation of a paper by Dr Dekker at the Toxic Cyanobacteria conference in Noosa in July.

2002 Dr Dekker coordinated an international workshop on Remote Sensing of Coastal and Inland Waters in October 2001(http://www.waterobserver.org/members/workshop-2001-10/): ” This three day workshop (22-24 October 2001) focused on the use of remote sensing in the field of resource management in nearshore and inland waters. The objective was to bring together representatives of various technologies applicable to aquatic resource management and discuss the synergy of these technologies in the presence of representatives of user groups, including government agencies, private sector companies and NGOs.” Dr Dekker and Dr Phinn have also been participating in a number of international workshops and project meetings on the potential role of remote sensing in detecting coral bleaching events. Dr Brando has set up a workshop on Remote Sensing of Coastal Waters to be run as part of the “Italo-Australian Exhibition and Conference on Innovation Technologies” to be held in Melbourne March 25-28. Dr’s Brando, Dekker and Phinn will be presenting work from their CRC projects at this meeting. Further details are provided below: Dr Phinn was conference director for the 12th Australasian Remote Sensing and Photogrammetry Conference (September 2-9) in Brisbane which had a an applied science theme of “Images to Information.”


Vittorio Brando: Attended the CRC LEADERSHIP AND CAREER DEVELOPMENT COURSE in Melbourne October 8 - 11, 2002 with Fiona Manson and Regina Counihan Attended the short course on “Atmospheric Correction of Ocean Color Sensors: Alpha to Psi” at Ocean Optics Conference in Santa Fe, U.S.A. in November 2002. Attended the International Working Group for Ocean Color Algorithms (IWGOCA) kickoff meeting, November 17, 2002 in Santa Fe, NM Arnold Dekker: Attended the International Working Group for Ocean Color Algorithms (IWGOCA) kickoff meeting, November 17, 2002 in Santa Fe, NM Stuart Phinn: Was director for the 11th Australasian Remote Sensing Conference held in Brisbane from September 2-6. This conference had over 380 registrants with the most number of sessions being presented in areas of coastal and marine remote sensing applications. Dr Phinn presented to a workshop on hyperspectral applications in coral reef and coastal environments. Chris Roelfsema: Coordinated (with one of Dr Phinn’s PhD students – Karen Joyce) a week long technical and field workshop on remote sensing of coral reefs and coastal environments in conjunction with the 11th Australasian Remote Sensing Conference . The workshop was attended by 12 scientists and resource managers from around the Pacific. Played a major role in the first implementation of the University of Queensland’s College at Sea Program, a two week course on a tall-ship sailing from Townsville to Yeppoon. While onboard Chris coordinated projects collecting field data for validation of water quality projects. Presentations at the 11th Australasian Remote Sensing Conference: Brando, V. Dekker, A. and Anstee, J. (2002) Estuarine hyperspectral remote sensing from space: Moreton Bay case study. In: Proceedings of the 11th Australasian Remote Sensing and Photogrammetry Conference, Brisbane, 2-6 September, CD-Rom, Causal Publications. Knight, J., Dale, P. and Phinn, S. (2002) Mapping mosquito breeding habitats (water) under mangrove canopy: contribution from airborne thermal scanning. In: Proceedings of the 11th Australasian Remote Sensing and Photogrammetry Conference, Brisbane, 2-6 September, CD-Rom, Causal Publications. Phinn, S., Held, A., Stanford M., Ticehurst, C. and C. Simpson (2002) Optimising State of Environment Monitoring at Multiple Scales Using Remotely Sensed Data. In: Proceedings of the 11th Australasian Remote Sensing and Photogrammetry Conference, Brisbane, 2-6 September, CD-Rom, Causal Publications.

Roelfsema, C., Phinn S., Dennison, W., Dekker, A and Brando, V (2002) Monitoring cyanobacterial blooms of Lyngbya Majuscula in Moreton Bay, Australia by combining field techniques with remote sensing. In: Proceedings of the 11th Australasian Remote Sensing and Photogrammetry Conference, Brisbane, 2-6 September, CD-Rom, Causal Publications. Presentation at the Ecological Society of Australia Conference: Manson, F., Lonergan, N., Skilleter, G. and Phinn, S. (2002) Establishing links between mangroves and fisheries in Australia: a conceptual framework Presentation to Community Groups: C.Roelfsema

- Presentations to Oyster lease owners on required input for Lyngbya monitoring in Moreton Bay - Presentation to Centre for Marine Studies/The Ecology Centre Research Seminar Series


Vittorio Brandos CRC related presentations: CSIRO LAND and WATER SEMINAR series, Canberra- 16 October 2002 Vittorio Brando. A clear view of turbid waters: spaceborne hyperspectral data as a methodological workbench i.e. : Environmental monitoring of coastal waters with remote sensing. NASA EO1 Science Validation Team workshop, Hawaii November 2002-12-19 Brando, V. Dekker, A. and Anstee, J. Satellite hyperspectral remote sensing for estimating estuarine and coastal water quality. Results of The Australia Science Validation Team for assessing Hyperion in for coastal remote sensing. (presented by Alex Held) CNR-ISDGM, Venice Italy and CNR-IREA- Milan, ITALY - July 2002 Brando, V. Estuarine hyperspectral remote sensing from space: Moreton Bay case study. Arnold Dekkers CRC related presentations: University of Edinburgh, Geography Department. Hosted by Dr . T. J. Malthus. 19 September 2002, Seminar on Remote sensing of Coastal Ecosystems in Australia. Marine Classification Workshop-National Oceans Office and FRDC, Melbourne; 23-24 September 2002:” Detection and monitoring of optical water column properties and substrates”. Joint UQ/CSIRO Presentation on Moreton Bay Waterways and Catchments Partnership; 9th Scientific Expert Panel Meeting, with S.Phinn 22nd October 2002, 9.00 am-3.30 pm; The University of Queensland :” Detection and monitoring of optical water column properties, substrates, harmful algal blooms and riparian condition and vegetation” “Noel Cressie” Spatial Statistics Workshop at CMIS Cleveland: 6 November 2002, “From Point to Pixel and from Image to Process - Scale and Validation issues in Detection and Monitoring of Optical Water Column Properties and Substrates” National Estuaries Network Meeting Stradbroke Island UQ Field station: 13 November 2002, From Point to Pixel and from Image to Process- Scale and Validation issues in Operationalisation of Detection and Monitoring of Optical Water Column Properties and Substrates Dr Dekker organised a visit and presentation at the CRC Annual General Meeting by Dr Hans van der Woerd from the Institute for Environmental Studies, Vrije Universiteit , The Netherlands . Dr van der Woerd provided a presentation on an operational program for monitoring chlorophyll and total suspended that had been set up using modern satellite systems for the North Sea involving Sweden, Denmark, Germany, Netherlands, Belgium and UK. The outcomes of this project is/will be used for monitoring the compliance of the member states to the OSPARCOM treaty (Oslo-Paris Committee) regarding the eutrophication status of the respective member states coastal sea part of the North Sea.


Appendix 2 Student Achievements Directly funded (Biophysical Remote Sensing Group, University of Queensland): Fiona Manson Doctoral Student, 2001-2004 Thesis title: Mangroves and fisheries: Are there links between coastal habitats and

fishery production? Status: Submitted November 2004 Synopsis: Fiona’s work has provided a systematic examination of a long held assertion

that “mangroves are critical habitat for coastal fisheries.” To conduct this assessment Fiona has established a new conceptual model to include all know links between mangroves and fisheries, and the examined mangrove change in Moreton Bay, Fitzroy Estuary and the entire Queensland coast in relation to fisheries catch data. Based on this information Fiona will be able to establish which factors of mangroves are critical for fisheries species at various life-cycle stages, how mangrove distributions have changed, and what type of work is need in future of this information is to be used for guiding management decisions.


Appendix 3 Image Processing Log Project: Coastal CRC MR2 Component: Processing of Landsat 7 ETM+ image 24-jul-2002 Product Scene Center Location (lat/long) : -23.083355 151.233431 Product Scene Center Date/Time (yyyy mm dd): 2002 7 24 23:41:20.2663 Date Input Image/Data Output image Processing Notes Image 2002 l7_09176_multi_240702.l

an Image purchased from Geoimage

in Erdas Lan format, 6 bands and map-corrected.

02.12.03 /RADCOR/PCFE_TM7_jul02_radcor_full.img

Applied Gain and Offset as in the report.txt (VEB)

3.12.03 PCFE_TM7_jul02_radcor_full.img

PCFE_TM7_jul02_radcor_full_convol.img

33x33 low pass convolution (JMA)

14.12.03 PCFE_TM7_ jul02_rad_geo_full_6bds.img PCFE_TM7_ jul02_rad_geo_full_convol_bil_6bds.img \ATCOR\PCFE_TM7_240702_12.wcd

PCFE_TM7_ jul02_rad_geo_full_6bds_R_app.img PCFE_TM7_ jul02_rad_geo_full_6bds_R_0.img

16.12.03 PCFE_ TM7_jul02_rad_geo_full_R_app.img PCFE_ TM7_jul02_rad_geo_full_R_0.img

PC_ jul02_R_0.img PC_ jul02_R_0.img FE_ jul02_R_app.img FE_ jul02_R_app.img

Subset size: PC=2000x1500 (3220-5719,6741-8240) FE=3250x3050 (861-4110,3831-6880)


17.12.03 PC_jul02_R_app.img PC_jul02_R_lm_hightide Land mask by checking occurrence of small land ridges such as jetties then adjusting the mask by masking out does areas in that have in band 4 a reflectance less then 0.0275, High water mask 0.0275 was reach through an iterative process.

18.12.03 PC_jul02_R_lm_hightide

and PC_jul02_R_app.img PC_jul02_R_app_lm_v1.img Land is masked out using the high

tide mask of July 2002

19.12.03 FE_jul02_R_app.img FE_jul02_R_lm_hightide Land mask by checking

occurrence of small land ridges such as jetties then adjusting the mask by masking out does areas in that have in band 4 a reflectance less then 0.04, High water mask. 0.04 was reach through an iterative process.


19.12.03 PC_jul02_R_lm_hightide and PC_jul02_R0-.img

PC_jul02_R0-_water Using hightide mask all land was masked out of image.

19.12.03 FE_ jul02_R_lm_hightide

and FE_jul02_R0-.img FE_ jul02_R0-_water Using hightide mask all land was

masked out of image.


19.12.03 FE_ jul02_R_lm_hightide and FE_jul02_app.img

FE_ jul02_app_land Using hightide mask all water was masked out of image.

19.12.03 PC_jul02_R_lm_hightide

and PC_jul02_app.img PC_jul02_R0-_land Using hightide mask all land was


27.01.04 PC_jul02_R0-_water PC_jul02_OD_spectra_lib_roi.roi Shallow, Turbid, Turbid high,

mixed, deep water, shallow veg and channels rois were selected.


27.01.04 PC_jul02_OD_spectra_lib_roi.roi and PC_jul02_R0-_water

PC_jul02_OD_spectra_lib ROIS were turned into spectra

27.01.04 PC_jul02_OD_spectra_li

b and PC_jul02_R0-_water

PC_jul02_R_OD_SAM Spectra/ROIS was used as input for SAM classification using maximum angle 0.25.

27.01.04 PC_jul02_R_OD_SAM PC_jul02_mixed water area.roi ROI of mixed area by digitising

over light green in input image


27.01.04 PC_jul02_mixed water area.roi and PC_jul02_R0-_water

PC_jul02_mixed_water ROI is used to mask out mixed water area.

27 Jan 2004

PC_jul02_mixed_water PC_04mixed_water.roi Submerged vegetation, Shallow, Turbid, Turbid high, mixed, deep water, shallow veg and channels rois were selected


27 Jan 2004

PC_jul02_mixed_water and PC_04mixed_water.roi

PC_mar04_mixedwater_sam Spectra/ROIS was used as input for SAM classification using maximum angle 0.25.

27 Jan 2004

fe_jul02_R0-_water fe_july04_od_.roi Deep water, mixed water, turbid water and river rois were selected


27 Jan 2004

fe_jul02_R0-_water and fe_july04_od_.roi

fe_july04_od_SAM Spectra/ROIS was used as input for SAM classification using maximum angle 0.25.

20.04.04 fe_jul02_R0-_water

pc_jul02_R0-_water FE_jul02_R0-_water_convol.img PC_jul02_R0-_water_convol.img

5x5 low pass convolution (RvE)

20.04.04 FE_jul02_R0-_water_convol.img PC_jul02_R0-_water_convol.img

FE_jul02_R0-_water_convol_BIL.img PC_jul02_R0-_water_convol_BIL.img

Conversion BSQ ! BIL (RvE)

27.04.04 pc_jul02_R0-_water

PC_NE_DELTA.txt NE_DELTA plot from deep water (RvE)

27.04.04 pc_jul02_R0-_water_convol.img

PC_convol_NE_DELTA.txt NE_DELTA plot from deep water in 5x5 convoluted image (RvE)

27.04.04 PC_NE_DELTA.txt PC_convol_NE_DELTA.txt

PC_NE_DELTA.xls Plots imported into Excel and graphs created (RvE)

29.04.04 PCFE_ TM7_jul02_rad_geo_full_R_0.img jul02_R_0_darkest_convol.img jul02_scroll.bmp

Darkest water was subsetted for noise estimation because previous noise estimation was in plume. The BMP shows the location (RvE)

29.04.04 jul02_R_0_darkest_convol.img

PCFE_NE_DELTA.xls New NEdR(0-) plot created (RvE)

30.04.04 jul02_R_0_darkest.img

jul02_R_0_darkest_PIF_plot.txt R_0_darkest_PIF_means_plot.txt

PIF_means of darkest deep sea (400 x 400 pixels) (RvE) turned out not to be darkest area


04.05.04 PCFE_TM7_jul02_radcor_full.img PCFE_TM7_jul02_radcor_full_band1.dsr PCFE_TM7_jul02_radcor_full_band2.dsr PCFE_TM7_jul02_radcor_full_band3.dsr PCFE_TM7_jul02_radcor_full_band4.dsr RadCor_lowest_values.xls

Lowest radiance density slices for each band, combined in excel sheet (RvE)

11.05.04 PCFE_TM7_jul02_radcor_full.img PCFE_TM7_jul02_radcor_full_bil_Rapp.img

Extreme_Radiance_per _Band.xls Coal_PIFs.xls PIFs.roi Coal_PIFs.roi PCFE_TM7_jul02_radcor_full_extremes.roi

Coordinates and ROIS of highest and lowest values in each band and spectra (RvE)

12.05.04 PCFE_TM7_jul02_radcor_full_bil_R_0.imgPCFE_TM7_jul02_radcor_full_extremes.roi

Jul02_Darkest_Ocean_5x5_lots_plot.txt Darkest_Ocean_5x5_lots_plots.xls

Spectra of 5x5 convoluted lowest ocean water for fraction of negative pixels (RvE)

12.05.04 PCFE_TM7_jul02_radcor_full_bil_R_0.imgCoastal_5x5_lots_plots_ROI.roi

Jul02_Coastal_5x5_lots_plot.txt Coastal_5x5_lots_plots.xls

Spectra of 5x5 convoluted coastal water for fraction of negative pixels (RvE)

12.05.04 PCFE_TM7_jul02_radcor_full_bil_R_0.imgCoal_PIFs.roi

Coal_GreenRoi_mean_plot.txt Coal_GreenRoi_min_plot.txt Coal_RedRoi_mean_plot.txt Coal_RedRoi_min_plot.txt Coal_PIFs.xls

Mean and synthetical minima spectra of coal piles (RvE)

12.05.04 to 14.05

PCFE_TM7_jul02_radcor_full.img

PCFE_TM7_jul02_radcor_full_33x33_square_std_conv.img

Ran ALCL algorithm, which took 3 days on the cluster. (RvE)

14.05.04 PCFE_TM7_jul02_radcor_full_33x33_square_std_conv.img PCFE_TM7_jul02_radcor_full_bil_R_0.img

Noise estimation from visually homogeneous location and with ALCL image. Conclusion: Noise estimated from ALCL homogeneous location is about 0.05% lower than from visually selected location. Three days of processing is not worth the lower noise figure (for Landsat imagery). (RvE)

31.05.04 FE_jul02_R0- FE_jul02_R0-_water_convol_BIL_MIM_SVDC_opt_phys_04__A_CHL_100.img Ran MIM on


_water_convol_BIL.img FE_jul02_R0-_water_convol_BIL_MIM_SVDC_opt_phys_04__BB_CHL_100.img FE_jul02_R0-_water_convol_BIL_MIM_SVDC_opt_phys_04__Kdn_CHL_100.img FE_jul02_R0-_water_convol_BIL_MIM_SVDC_opt_phys_04_CHL_100.img

FE R(0-) images. (RvE)

02.06.04 fe_jul02_hightide.img FE_Mask.img Inverted mask. (RvE)

02.05.04 FE_jul02_R0-_water_convol_BIL_MIM_SVDC_opt_phys_04_CHL_100.img FE_jul02_R0-_water_convol_BIL_MIM_SVDC_opt_phys_04__Kdn_CHL_100.img FE_Mask.img

FE_jul02_R0-_water_convol_BIL_MIM_SVDC_opt_phys_04_CHL_100_masked.img FE_jul02_R0-_water_convol_BIL_MIM_SVDC_opt_phys_04__Kdn_CHL_100_masked.img

Applied high water mask to products. (RvE)

03.06.04 FE_jul02_R0-_water_convol_BIL_MIM_SVDC_opt_phys_04_CHL_100_masked.img FE_jul02_R0-_water_convol_BIL_MIM_SVDC_opt_phys_04__Kdn_CHL_100_masked.img Mask_Whitish+Nodata.dsr MIM_SVDC_dR0_scale.dsr MIM_SVDC_SIOP_scale.dsr

FE_jul02_CDOM_0-06.tif FE_jul02_dR(0-).tif FE_jul02_Kd_0-3.tif FE_jul02_SD_0-2_m_inversed.tif FE_jul02_SD_0-3_m.tif FE_jul02_SIOPs.tif FE_jul02_Tripton_0-50_mg.tif

Applied density slices and stretch and exported to GEOTif. (RvE)

04.06.04 FE_jul02_CDOM_0-06.tif FE_jul02_dR(0-).tif FE_jul02_Kd_0-3.tif FE_jul02_SD_0-2_m_inversed.tif FE_jul02_SD_0-3_m.tif FE_jul02_SIOPs.tif FE_jul02_Tripton_0-50_mg.tif

FE_jul02_CDOM.cdr FE_jul02_dR0.cdr FE_jul02_Kd.cdr FE_jul02_SD.cdr FE_jul02_SD.cdr FE_jul02_SIOPs.cdr FE_jul02_Tripton.cdr

Imported to CorelDraw and added legend, scale, names, date. (RvE)

07.06.04 PC_jul02_R0-_water_convol_BIL.img

PC_jul02_R0-_water_convol_BIL_MIM_SVDC_opt_phys_04__A_CHL_100.img PC_jul02_R0-_water_convol_BIL_MIM_SVDC_opt_phys_04__BB_CHL_100.img PC_jul02_R0-_water_convol_BIL_MIM_SVDC_opt_phys_04__Kdn_CHL_100.img PC_jul02_R0-_water_convol_BIL_MIM_SVDC_opt_phys_04_CHL_100.img

Ran MIM on PC R(0-) images. (RvE)

07.06.04 PC_jul02_hightide.img PC_Mask.img Inverted mask. (RvE)

07.05.04 PC_jul02_R0-_water_convol_BIL_MIM_SVDC_opt_phys_04_CHL_100.img PC_jul02_R0- Applied high


PC_jul02_R0-_water_convol_BIL_MIM_SVDC_opt_phys_04__Kdn_CHL_100.img PC_Mask.img

_water_convol_BIL_MIM_SVDC_opt_phys_04_CHL_100_masked.img PC_jul02_R0-_water_convol_BIL_MIM_SVDC_opt_phys_04__Kdn_CHL_100_masked.img

water mask to products. (RvE)

07.06.04 PC_jul02_R0-_water_convol_BIL_MIM_SVDC_opt_phys_04_CHL_100_masked.img PC_jul02_R0-_water_convol_BIL_MIM_SVDC_opt_phys_04__Kdn_CHL_100_masked.img Mask_Whitish+Nodata.dsr MIM_SVDC_dR0_scale.dsr MIM_SVDC_SIOP_scale.dsr

PC_jul02_CDOM_0-06.tif PC_jul02_dR(0-).tif PC_jul02_Kd_0-3.tif PC_jul02_SD_0-2_m_inversed.tif PC_jul02_SD_0-3_m.tif PC_jul02_SIOPs.tif PC_jul02_Tripton_0-50_mg.tif


07.06.04 PC_jul02_CDOM_0-06.tif PC_jul02_dR(0-).tif PC_jul02_Kd_0-3.tif PC_jul02_SD_0-2_m_inversed.tif PC_jul02_SD_0-3_m.tif PC_jul02_SIOPs.tif PC_jul02_Tripton_0-50_mg.tif

PC_jul02_CDOM.cdr PC_jul02_dR0.cdr PC_jul02_Kd.cdr PC_jul02_SD.cdr PC_jul02_SD.cdr PC_jul02_SIOPs.cdr PC_jul02_Tripton.cdr


08.06.04 PC_jul02_CDOM_0-06.cdr PC_jul02_dR(0-).cdr PC_jul02_Kd_0-3.cdr PC_jul02_SD_0-2_m_inversed.cdr PC_jul02_SD_0-3_m.cdr PC_jul02_SIOPs.cdr PC_jul02_Tripton_0-50_mg.cdr

PC_jul02_CDOM.jpg PC_jul02_dR0.jpg PC_jul02_Kd.jpg PC_jul02_SD.jpg PC_jul02_SD.jpg PC_jul02_SIOPs.jpg PC_jul02_Tripton.jpg

Products exported to A3 sized JPGs. (RvE)

08.06.04 PC_jul02_CDOM.jpg PC_jul02_dR0.jpg PC_jul02_Kd.jpg PC_jul02_SD.jpg PC_jul02_SD.jpg PC_jul02_SIOPs.jpg PC_jul02_Tripton.jpg

PC_MIM_Products.doc Imported images from old to new in word document. (RvE)


08.06.04 FE_jul02_CDOM.cdr

FE_jul02_dR0.cdr FE_jul02_Kd.cdr FE_jul02_SD.cdr FE_jul02_SD.cdr FE_jul02_SIOPs.cdr FE_jul02_Tripton.cdr

FE_jul02_CDOM.jpg FE_jul02_dR0.jpg FE_jul02_Kd.jpg FE_jul02_SD.jpg FE_jul02_SD.jpg FE_jul02_SIOPs.jpg FE_jul02_Tripton.jpg


08.06.04 FE_jul02_CDOM.jpg FE_jul02_dR0.jpg FE_jul02_Kd.jpg FE_jul02_SD.jpg FE_jul02_SD.jpg FE_jul02_SIOPs.jpg FE_jul02_Tripton.jpg

FE_MIM_Products.doc Imported images from old to new in word document. (RvE)

09.06.04 fe_july04_opticaldomains_SAM.jpg PC_mar02_substrate_SAM.jpg

FE_02_optical_domains.cdr PC_02_Substrate.cdr


09.06.04 FE_02_optical_domains.cdr PC_02_Substrate.cdr

FE_02_optical_domains.jpg PC_02_Substrate.jpg


09.06.04 FE_02_optical_domains.jpg FE_Optical_Domain_Products Imported images from old to new in word document. (RvE)

09.06.04 PC_02_Substrate.jpg PC_Substrate_Products.doc Imported images from old to new in word document. (RvE)

09.06.04 FE_02_95_NDVI.cdr PC_02_95_NDVI.cdr

FE_02_95_NDVI.cdr PC_02_95_NDVI.cdr

Changed layout and added water mask. (RvE)


FE_02_95_NDVI.jpg PC_02_95_NDVI.jpg


09.06.04 FE_02_95_NDVI.jpg

FE_NDVI_Products.doc Imported images from old to new in word document. (RvE)

09.06.04 PC_02_95_NDVI.jpg

PC_NDVI_Products.doc Imported images from old to new in word document. (RvE)

Component: Processing of Landsat 5 TM image 21-aug-1986 Product Scene Center Location (lat/long): -23.120930 151.280375 Product Scene Center Date/Time (yyyy mm dd): 1986 8 21 23:14:25.7499


Date Input Image/Data Output image/data Processing Notes 20.11.03 TM09176_210886 /GEOREF/21081986_2_base.pts GCP file for geocorrection to base

file=l7_09176_multi)240702.lan (JMA) rms ≈ 0.3 pixels

26.11.03 /RADCOR/PCFE_TM5_aug86_radcor_full.img Imported the raw TM5 image [reprocessed by ACRES to CCRS CPF] (VEB) Applied Gain and Offset as in the report.txt (VEB)

1.12.03 PCFE_TM5_aug86_radcor_full.img PCFE_TM5_aug86_rad_geo_full.img Warped to base file= l7_09176_multi)240702.lan (JMA) Warp using 1st degree polynomial / nearest neighbour

1.12.03 PCFE_TM5_aug86_rad_geo_full.img PCFE_TM5_aug86_rad_geo_full_convol.img 33x33 low pass convolution (JMA) 14.12.03

PCFE_TM5_aug86_rad_geo_full_6bds.img PCFE_TM5_aug86_rad_geo_full_convol_bil_6bds.img \ATCOR\PCFE_TM5_210886_4.wcd

PCFE_TM5_aug86_rad_geo_full_6bds_R_app.img PCFE_TM5_aug86_rad_geo_full_6bds_R_0.img

16.12.03

PCFE_TM5_ aug86_rad_geo_full_R_app.img PCFE_TM5_ aug86_rad_geo_full_R_0.img

PC_ aug86_0.img PC_ aug86_0.img FE_ aug86_app.img FE_ aug86_app.img


19.12.03 PC_aug86_R_lm_hightide and PC_jul02_R0-.img

PC_ aug86_R0-_water Using hightide mask all land was masked out of image.


19.12.03 FE_aug86_R_lm_hightide and FE_jul02_R0-.img

FE_ aug86_R0-_water Using hightide mask all land was masked out of image.

19.12.03 FE_ jul02_R_lm_hightide and

FE_aug86_app.img FE_ aug86_app_land Using hightide mask all water

was masked out of image.


19-dec-03 PC_aug86_R_lm_hightide and PC_jul02_app.img

PC_ aug86_R0-_land Using hightide mask all land was masked out of image.

21.01.04 PC_ aug86_R0-_water Optical_domain.sig Shallow, Turbid, Turbid high,

mixed, deep water, shallow veg and river spectra were selected.

21.01.04 PC_ aug86_R0-_water and

Optical_domain.sig PC_ aug86_od SAM classification using

maximum angle 0.25.


22.01.04 PC_ aug86_od PC_1986_mixed_water.roi ROI of mixed area by digitising over light green in input image

22.01.04 PC_1986_mixed_water.roi

and PC_ aug86_R0-_water PC_1986_od_mixed_water PC_ aug86_R0-_water is

subseted using PC_1986_mixed_water.roi


22.01.04 PC_1986_od_mixed_water PC_1986_OD_MW_classes.roi With the ROI tool polygons were created of the main substrate types in the mixed waters.

27.01.04 PC_1986_OD_MW_classes.r

oi PC_aug86_od_mixed_water_SAM

SAM classification using maximum angle 0.25. and after colour assignment


27.01.04 fe_ aug86_R0-_water fe_aug86_od_.roi Deep water, mixed water, turbid water and river rois were selected

27.01.04 fe_ aug86_R0-_water and

fe_ aug86_od_.roi fe_ aug86_od_SAM Spectra/ROIS was used as input

for SAM classification using maximum angle 0.25.

20.04.04 fe_aug86_R0-_water

pc_aug86_R0-_water FE_aug86_R0-_water_convol.img PC_aug86_R0-_water_convol.img


20.04.04 FE_aug86_R0-_water_convol.img PC_aug86_R0-_water_convol.img

FE_aug86_R0-_water_convol_BIL.img PC_aug86_R0-_water_convol_BIL.img


27.04.04 pc_aug86_R0-_water


27.04.04 pc_aug86_R0-_water_convol.img





29.04.04 PCFE_TM5_ aug86_rad_geo_full_R_0.img aug86_R_0_darkest_convol.img aug86_scroll.bmp


29.04.04 aug86_R_0_darkest_convol.img


30.04.04 aug86_R_0_darkest.img

aug86_R_0_darkest_PIF_plot.txt R_0_darkest_PIF_means_plot.txt

PIF_means of darkest deep sea (400 x 400 pixels) (RvE) turned out not to be darkest area

04.05.04 PCFE_TM5_aug86_radcor_VNIR.img PCFE_TM5_aug86_radcor_VNIR_band1.dsr PCFE_TM5_aug86_radcor_VNIR_band2.dsr PCFE_TM5_aug86_radcor_VNIR_band3.dsr PCFE_TM5_aug86_radcor_VNIR_band4.dsr RadCor_lowest_values.xls


11.05.04 PCFE_TM5_aug86_rad_geo_full_6bds.img PCFE_TM5_aug86_rad_geo_full_6bds_Rapp.img

Extreme_Radiance_per _Band.xls Coal_PIFs.xls PIFs.roi Coal_PIFs.roi PCFE_TM5_aug86_rad_geo_full_6bds_extremes.roi


12.05.04 PCFE_TM5_aug86_rad_geo_full_6bds_R_0.img PCFE_TM5_aug86_rad_geo_full_6bds_extremes.roi

Aug86_Darkest_Ocean_5x5_lots_plot.txt Darkest_Ocean_5x5_lots_plots.xls


12.05.04 PCFE_TM5_aug86_rad_geo_full_6bds_R_0.img Coastal_5x5_lots_plots_ROI.roi

Aug86_Coastal_5x5_lots_plot.txt Coastal_5x5_lots_plots.xls


12.05.04 PCFE_TM5_aug86_rad_geo_full_6bds_Rapp.img Coal_PIFs.roi



31.05.04 FE_aug86_R0-_water_convol_BIL.img FE_aug86_R0-_water_convol_BIL_MIM_SVDC_opt_phys_04__A_CHL_100.img FE_aug86_R0-_water_convol_BIL_MIM_SVDC_opt_phys_04__BB_CHL_100.img FE_aug86_R0-_water_convol_BIL_MIM_SVDC_opt_phys_04__Kdn_CHL_100.img FE_aug86_R0-_water_convol_BIL_MIM_SVDC_opt_phys_04_CHL_100.img

Ran MIM on FE R(0-) images.


(RvE) 02.06.04 fe_jul02_hightide.img FE_Mask.img Inverted mask. (RvE) 02.05.04 FE_aug86_R0-

_water_convol_BIL_MIM_SVDC_opt_phys_04_CHL_100.img FE_aug86_R0-_water_convol_BIL_MIM_SVDC_opt_phys_04__Kdn_CHL_100.img FE_Mask.img

FE_aug86_R0-_water_convol_BIL_MIM_SVDC_opt_phys_04_CHL_100_masked.img FE_aug86_R0-_water_convol_BIL_MIM_SVDC_opt_phys_04__Kdn_CHL_100_masked.img


03.06.04 FE_aug86_R0-_water_convol_BIL_MIM_SVDC_opt_phys_04_CHL_100_masked.img FE_aug86_R0-_water_convol_BIL_MIM_SVDC_opt_phys_04__Kdn_CHL_100_masked.img Mask_Whitish+Nodata.dsr MIM_SVDC_dR0_scale.dsr MIM_SVDC_SIOP_scale.dsr

FE_aug86_CDOM_0-06.tif FE_aug86_dR(0-).tif FE_aug86_Kd_0-3.tif FE_aug86_SD_0-2_m_inversed.tif FE_aug86_SD_0-3_m.tif FE_aug86_SIOPs.tif FE_aug86_Tripton_0-50_mg.tif


07.06.04 PC_aug86_R0-_water_convol_BIL.img PC_aug86_R0-_water_convol_BIL_MIM_SVDC_opt_phys_04__A_CHL_100.img PC_aug86_R0-_water_convol_BIL_MIM_SVDC_opt_phys_04__BB_CHL_100.img PC_aug86_R0-_water_convol_BIL_MIM_SVDC_opt_phys_04__Kdn_CHL_100.img PC_aug86_R0-_water_convol_BIL_MIM_SVDC_opt_phys_04_CHL_100.img


07.06.04 PC_aug86_hightide.img PC_Mask.img Inverted mask. (RvE) 07.05.04 PC_aug86_R0-

_water_convol_BIL_MIM_SVDC_opt_phys_04_CHL_100.img PC_aug86_R0-_water_convol_BIL_MIM_SVDC_opt_phys_04__Kdn_CHL_100.img

PC_aug86_R0-_water_convol_BIL_MIM_SVDC_opt_phys_04_CHL_100_masked.img PC_aug86_R0-_water_convol_BIL_MIM_SVDC_opt_phys_04_



PC_Mask.img _Kdn_CHL_100_masked.img 07.06.04 PC_aug86_R0-

_water_convol_BIL_MIM_SVDC_opt_phys_04_CHL_100_masked.img PC_aug86_R0-_water_convol_BIL_MIM_SVDC_opt_phys_04__Kdn_CHL_100_masked.img Mask_Whitish+Nodata.dsr MIM_SVDC_dR0_scale.dsr MIM_SVDC_SIOP_scale.dsr

PC_aug86_CDOM_0-06.tif PC_aug86_dR(0-).tif PC_aug86_Kd_0-3.tif PC_aug86_SD_0-2_m_inversed.tif PC_aug86_SD_0-3_m.tif PC_aug86_SIOPs.tif PC_aug86_Tripton_0-50_mg.tif


07.06.04 PC_aug86_CDOM_0-06.tif PC_aug86_dR(0-).tif PC_aug86_Kd_0-3.tif PC_aug86_SD_0-2_m_inversed.tif PC_aug86_SD_0-3_m.tif PC_aug86_SIOPs.tif PC_aug86_Tripton_0-50_mg.tif

PC_aug86_CDOM.cdr PC_aug86_dR0.cdr PC_aug86_Kd.cdr PC_aug86_SD.cdr PC_aug86_SD.cdr PC_aug86_SIOPs.cdr PC_aug86_Tripton.cdr


08.06.04 PC_aug86_CDOM_0-06.cdr PC_aug86_dR(0-).cdr PC_aug86_Kd_0-3.cdr PC_aug86_SD_0-2_m_inversed.cdr PC_aug86_SD_0-3_m.cdr PC_aug86_SIOPs.cdr PC_aug86_Tripton_0-50_mg.cdr

PC_aug86_CDOM.jpg PC_aug86_dR0.jpg PC_aug86_Kd.jpg PC_aug86_SD.jpg PC_aug86_SD.jpg PC_aug86_SIOPs.jpg PC_aug86_Tripton.jpg


08.06.04 PC_aug86_CDOM.jpg PC_aug86_dR0.jpg PC_aug86_Kd.jpg PC_aug86_SD.jpg PC_aug86_SD.jpg PC_aug86_SIOPs.jpg PC_aug86_Tripton.jpg


08.06.04 FE_aug86_CDOM.cdr FE_aug86_dR0.cdr FE_aug86_Kd.cdr

FE_aug86_CDOM.jpg FE_aug86_dR0.jpg FE_aug86_Kd.jpg



FE_aug86_SD.cdr FE_aug86_SD.cdr FE_aug86_SIOPs.cdr FE_aug86_Tripton.cdr

FE_aug86_SD.jpg FE_aug86_SD.jpg FE_aug86_SIOPs.jpg FE_aug86_Tripton.jpg

08.06.04 FE_aug86_CDOM.jpg FE_aug86_dR0.jpg FE_aug86_Kd.jpg FE_aug86_SD.jpg FE_aug86_SD.jpg FE_aug86_SIOPs.jpg FE_aug86_Tripton.jpg


09.06.04 fe_aug86_opticaldomains_SAM.jpg PC_aug86_substrate_SAM.jpg














09.06.04 FE_90_86_NDVI.jpg


09.06.04 PC_90_86_NDVI.jpg


Component: Processing of Landsat 5 TM image 16-aug-1990 Product Scene Center Location (lat/long): -23.113107 151.225788 Product Scene Center Date/Time (yyyy mm dd): 1990 8 16 23:12:58.9310


Date Input Image/Data Output image/data Processing Notes 20.11.03 TM09176_160890 /GEOREF/16081990_2_base.pts GCP file for geocorrection to

base file=l7_09176_multi)240702.lan (JMA) rms ≈ 0.3 pixels

26.11.03 /RADCOR/PCFE_TM5_aug90_radcor_full.img Imported the raw TM5 image [reprocessed by ACRES to CCRS CPF] (VEB) Applied Gain and Offset as in the report.txt (VEB)

1.12.03 PCFE_TM5_aug90_radcor_full.img PCFE_TM5_aug90_rad_geo_full.img Warped to base file= l7_09176_multi)240702.lan (JMA) Warp using 1st degree polynomial / nearest neighbour

1.12.03 PCFE_TM5_aug90_rad_geo_full.img PCFE_TM5_aug90_rad_geo_full_convol.img 33x33 low pass convolution (JMA)

14.12.03 PCFE_TM5_aug90_rad_geo_full_6bds.img PCFE_TM5_aug90_rad_geo_full_convol_bil_6bds.img \ATCOR\PCFE_TM5_160890_2.wcd

PCFE_TM5_aug90_rad_geo_full_6bds_R_app.img PCFE_TM5_aug90_rad_geo_full_6bds_R_0.img

16.12.03 PCFE_TM5_ aug90_rad_geo_full_R_app.img PCFE_TM5_ aug90_rad_geo_full_R_0.img

PC_ aug90_R_0.img PC_ aug90_R_0.img FE_ aug90_R_app.img FE_ aug90_R_app.img



17.12.03 PC_aug90_R_app.img PC_ aug90_R_lm_lowtide Land mask by checking occurrence of small land ridges such as jetties then adjusting the mask by masking out does areas in that have in band 4 a reflectance less then 0.025, Low water mask

18.12.03 PC_aug90_R_lm_lowtide

and PC_jul02_R_lm_hightide

PC_aug90_R_exposed_mask.img

Mask made by sub stracting low from high tide mask using band math.

19-dec-03 PC_aug90_R_lm_hightid

e and PC_jul02_R0-.img PC_ aug90_R0-_water Using hightide mask all land was



19.12.03 FE_aug90_R_lm_hightide and FE_jul02_R0-.img

FE_ aug90_R0-_water Using hightide mask all land was masked out of image.


and FE_aug90_app.img FE_ aug90_app_land Using hightide mask all water was



19.12.03 PC_aug90_R_lm_hightide and PC_jul02_app.img

PC_ aug90_R0-_land Using hightide mask all land was masked out of image.

27.01.04 PC_aug90_R0-_water PC_aug90_R_tidal_mask Mask made by sub stracting low

from high tide mask using band math.

27.01.04 PC_aug90_R_tidal_mask

and PC_aug90_R0-_water

PC_aug90_R_exposed Band math PC_aug90_R_tidal_mask X PC_aug90_R0-_water = PC_aug90_R_exposed then screen looked back but after linear stretch areas occur which are exposed areas.


27.01.04 PC_aug90_R_exposed and PC_aug90_R_exposed_ROI.roi

PC_aug90_R_exposed_sam SAM classification using maximum angle 0.25. and after colour assignment

27.01.04 PC_aug90_R0-_water PC_1990_od_sam SAM classification using

maximum angle 0.25. and after colour assignment And With the ROI tool polygons were created of the main water types and substrate types in the mixed waters. HOW NOW TO CLASSIFY OR TAKE OUT MIXED WATER?

27.01.04 PC_1990_od_sam PC_1990_mixed_mask.roi ROI of mixed area by digitising

over light green in input image


27.01.04 PC_aug90_R0-_water PC_1990_mixed_mask subsetted

27.01.04 PC_1990_mixed_mask PC_aug90_DO_mixedwater_ROI

.roi Submerged veg, shallow banks, Shallow, Turbid, Turbid high, mixed, deep water, shallow veg and channels rois were selected.


27.01.04 PC_1990_mixed_mask and PC_aug90_DO_mixedwater_ROI.roi

PC_aug90_DO_mixedwater_sam SAM classification using maximum angle 0.25. and after colour assignment And With the ROI tool polygons were created of the main water types and substrate types in the mixed waters.

27.01.04 fe_aug90_r0-_water fe_aug90_od.roi Deep water, mixed water, turbid water and river ROIS were selected


27.01.04 fe_aug90_r0-_water and fe_aug90_od.roi

fe_aug90_od_SAM Spectra/ROIS was used as input for SAM classification using maximum angle 0.25.

20.04.04 fe_aug90_R0-_water

pc_aug90_R0-_water FE_aug90_R0-_water_convol.img PC_aug90_R0-_water_convol.img


20.04.04 FE_aug90_R0-_water_convol.img PC_aug90_R0-_water_convol.img

FE_aug90_R0-_water_convol_BIL.img PC_aug90_R0-_water_convol_BIL.img


27.04.04 pc_aug90_R0-_water


27.04.04 pc_aug90_R0-_water_convol.img




29.04.04 PCFE_TM5_ aug90_rad_geo_full_R_0.img aug90_R_0_darkest_convol.img aug90_scroll.bmp


29.04.04 aug90_R_0_darkest_convol.img


30.04.04 aug90_R_0_darkest.img

aug90_R_0_darkest_PIF_plot.txt R_0_darkest_PIF_means_plot.txt

PIF_means of darkest deep sea (400 x 400 pixels) (RvE)


turned out not to be darkest area

04.05.04 PCFE_TM5_aug90_radcor_VNIR.img PCFE_TM5_aug90_radcor_VNIR_band1.dsr PCFE_TM5_aug90_radcor_VNIR_band2.dsr PCFE_TM5_aug90_radcor_VNIR_band3.dsr PCFE_TM5_aug90_radcor_VNIR_band4.dsr RadCor_lowest_values.xls


11.05.04 PCFE_TM5_aug90_rad_geo_full_6bds.img PCFE_TM5_aug90_rad_geo_full_6bds_Rapp.img

Extreme_Radiance_per _Band.xls Coal_PIFs.xls PIFs.roi Coal_PIFs.roi PCFE_TM5_aug90_rad_geo_full_6bds_extremes.roi


12.05.04 PCFE_TM5_aug90_rad_geo_full_6bds_R_0.img PCFE_TM5_aug90_rad_geo_full_6bds_extremes.roi

Aug90_Darkest_Ocean_5x5_lots_plot.txt Darkest_Ocean_5x5_lots_plots.xls


12.05.04 PCFE_TM5_aug90_rad_geo_full_6bds_R_0.img Coastal_5x5_lots_plots_ROI.roi

Aug90_Coastal_5x5_lots_plot.txt Coastal_5x5_lots_plots.xls


12.05.04 PCFE_TM5_aug90_rad_geo_full_6bds_Rapp.img Coal_PIFs.roi



31.05.04 FE_aug90_R0-_water_convol_BIL.img FE_aug90_R0-_water_convol_BIL_MIM_SVDC_opt_phys_04__A_CHL_100.img FE_aug90_R0-_water_convol_BIL_MIM_SVDC_opt_phys_04__BB_CHL_100.img FE_aug90_R0-_water_convol_BIL_MIM_SVDC_opt_phys_04__Kdn_CHL_100.img FE_aug90_R0-_water_convol_BIL_MIM_SVDC_opt_phys_04_CHL_100.img

Ran MIM on FE R(0-) images. (RvE)

02.06.04 fe_jul02_hightide.img FE_Mask.img Inverted mask. (RvE) 02.05.04 FE_aug90_R0-

_water_convol_BIL_MIM_SVDC_opt_phys_04_CHL_100.img FE_aug90_R0-_water_convol_BIL_MIM_SVDC_opt_phys_04__Kdn_CHL_100.img

FE_aug90_R0-_water_convol_BIL_MIM_SVDC_opt_phys_04_CHL_100_masked.img FE_aug90_R0-_water_convol_BIL_MIM_SVDC_opt_phys_04__Kdn_CHL_100_masked.img



FE_Mask.img 03.06.04 FE_aug90_R0-

_water_convol_BIL_MIM_SVDC_opt_phys_04_CHL_100_masked.img FE_aug90_R0-_water_convol_BIL_MIM_SVDC_opt_phys_04__Kdn_CHL_100_masked.img Mask_Whitish+Nodata.dsr MIM_SVDC_dR0_scale.dsr MIM_SVDC_SIOP_scale.dsr

FE_aug90_CDOM_0-06.tif FE_aug90_dR(0-).tif FE_aug90_Kd_0-3.tif FE_aug90_SD_0-2_m_inversed.tif FE_aug90_SD_0-3_m.tif FE_aug90_SIOPs.tif FE_aug90_Tripton_0-50_mg.tif


07.06.04 PC_aug90_R0-_water_convol_BIL.img PC_aug90_R0-_water_convol_BIL_MIM_SVDC_opt_phys_04__A_CHL_100.img PC_aug90_R0-_water_convol_BIL_MIM_SVDC_opt_phys_04__BB_CHL_100.img PC_aug90_R0-_water_convol_BIL_MIM_SVDC_opt_phys_04__Kdn_CHL_100.img PC_aug90_R0-_water_convol_BIL_MIM_SVDC_opt_phys_04_CHL_100.img


07.06.04 PC_aug90_hightide.img PC_Mask.img Inverted mask. (RvE) 07.05.04 PC_aug90_R0-

_water_convol_BIL_MIM_SVDC_opt_phys_04_CHL_100.img PC_aug90_R0-_water_convol_BIL_MIM_SVDC_opt_phys_04__Kdn_CHL_100.img PC_Mask.img

PC_aug90_R0-_water_convol_BIL_MIM_SVDC_opt_phys_04_CHL_100_masked.img PC_aug90_R0-_water_convol_BIL_MIM_SVDC_opt_phys_04__Kdn_CHL_100_masked.img


07.06.04 PC_aug90_R0-_water_convol_BIL_MIM_SVDC_opt_phys_04_CHL_100_masked.img PC_aug90_R0-_water_convol_BIL_MIM_SVDC_opt_phys_04__Kdn_CHL_100_masked.img

PC_aug90_CDOM_0-06.tif PC_aug90_dR(0-).tif PC_aug90_Kd_0-3.tif PC_aug90_SD_0-2_m_inversed.tif PC_aug90_SD_0-3_m.tif PC_aug90_SIOPs.tif



Mask_Whitish+Nodata.dsr MIM_SVDC_dR0_scale.dsr MIM_SVDC_SIOP_scale.dsr

PC_aug90_Tripton_0-50_mg.tif

07.06.04 PC_aug90_CDOM_0-06.tif PC_aug90_dR(0-).tif PC_aug90_Kd_0-3.tif PC_aug90_SD_0-2_m_inversed.tif PC_aug90_SD_0-3_m.tif PC_aug90_SIOPs.tif PC_aug90_Tripton_0-50_mg.tif

PC_aug90_CDOM.cdr PC_aug90_dR0.cdr PC_aug90_Kd.cdr PC_aug90_SD.cdr PC_aug90_SD.cdr PC_aug90_SIOPs.cdr PC_aug90_Tripton.cdr


08.06.04 PC_aug90_CDOM_0-06.cdr PC_aug90_dR(0-).cdr PC_aug90_Kd_0-3.cdr PC_aug90_SD_0-2_m_inversed.cdr PC_aug90_SD_0-3_m.cdr PC_aug90_SIOPs.cdr PC_aug90_Tripton_0-50_mg.cdr

PC_aug90_CDOM.jpg PC_aug90_dR0.jpg PC_aug90_Kd.jpg PC_aug90_SD.jpg PC_aug90_SD.jpg PC_aug90_SIOPs.jpg PC_aug90_Tripton.jpg


08.06.04 PC_aug90_CDOM.jpg PC_aug90_dR0.jpg PC_aug90_Kd.jpg PC_aug90_SD.jpg PC_aug90_SD.jpg PC_aug90_SIOPs.jpg PC_aug90_Tripton.jpg


08.06.04 FE_aug90_CDOM.cdr FE_aug90_dR0.cdr FE_aug90_Kd.cdr FE_aug90_SD.cdr FE_aug90_SD.cdr FE_aug90_SIOPs.cdr FE_aug90_Tripton.cdr

FE_aug90_CDOM.jpg FE_aug90_dR0.jpg FE_aug90_Kd.jpg FE_aug90_SD.jpg FE_aug90_SD.jpg FE_aug90_SIOPs.jpg FE_aug90_Tripton.jpg


08.06.04 FE_aug90_CDOM.jpg FE_aug90_dR0.jpg



FE_aug90_Kd.jpg FE_aug90_SD.jpg FE_aug90_SD.jpg FE_aug90_SIOPs.jpg FE_aug90_Tripton.jpg

09.06.04 fe_aug90_opticaldomains_SAM.jpg PC_aug90_substrate_SAM.jpg














09.06.04 FE_95_90_NDVI.jpg


09.06.04 PC_95_90_NDVI.jpg








09.06.04 FE_90_86_NDVI.jpg


09.06.04 PC_90_86_NDVI.jpg


Component: Processing of Landsat 5 TM image 7-mar-1995 Product Scene Center Location (lat/long): -23.126182 151.316554 Product Scene Center Date/Time (yyyy mm dd): 1995 3 7 23:03:32.9129


Date Input Image/Data Output image/data Processing Notes 19.11.03 TM09176_070395 /GEOREF/07031995_2_base.pts GCP file for geocorrection to

base file=l7_09176_multi)240702.lan (JMA) rms ≈ 0.3 pixels

26.11.03 /RADCOR/PCFE_TM5_mar95_radcor_full.img Imported the raw TM5 image [reprocessed by ACRES to CCRS CPF] (VEB) Applied Gain and Offset as in the report.txt (VEB)

1.12.03 PCFE_TM5_mar95_radcor_full.img PCFE_TM5_mar95_rad_geo_full.img Warped to base file= l7_09176_multi)240702.lan (JMA) Warp using 1st degree polynomial / nearest neighbour

11.12.03 PCFE_TM5_mar95_rad_geo_full_6bds.img PCFE_TM5_mar95_rad_geo_full_convol_bil_6bds.img \ATCOR\PCFE_TM5_070395_11.wcd

PCFE_TM5_mar95_rad_geo_full_6bds_R_app.img PCFE_TM5_mar95_rad_geo_full_6bds_R_0.img

jma

16.12.03 PCFE_TM5_mar95_rad_geo_full_R_app.img PCFE_TM5_mar95_rad_geo_full_R_0.img

PC_mar95_R_0.img PC_mar95_R_0.img FE_mar95_R_app.img FE_mar95_R_app.img

Subset size: PC=2000x1500 (3323-5322,6705-8205) FE=3250x3050 (964-4213, 3796-6845)

19.12.03 PC_mar95_R_lm_hightide and PC_jul02_R0-.img

PC_mar95_R0-_water Using hightide mask all land was masked out of image.


19.12.03 FE_mar95_R_lm_hightide and FE_jul02_R0-.img

FE_ mar95_R0-_water Using hightide mask all land was masked out of image.


and FE_mar95_app.img FE_ mar95_app_land Using hightide mask all water was



19.12.03 PC_mar95_R_lm_hightide and PC_jul02_app.img

PC_mar95_R0-_land Using hightide mask all land was masked out of image.

27.01.04 PC_mar95_R0-_water PC_mar95_Combined spect.roi Exposed vegetation,

Trichodesmium, Shallow, Turbid, Turbid high, mixed, deep water, shallow veg and channels ROIS were select

27.01.04 PC_mar95_Combined

spect.roi Spectra of rois


27.01.04 PC_mar95_R0-_water and PC_mar95_Combined spect.roi

PC_mar95_R0-_combined_SAM Spectral an SAM classification using maximum angle 0.25. and after colour assignment And With the ROI tool polygons were created of the main water types and substrate types in the mixed waters.

27.01.04 fe_mar95_r0-_water fe_mar95_od.roi Substrate, Deep water, mixed water, turbid water and river ROIS were selected


27.01.04 fe_mar95_r0-_water and fe_mar95_od.roi

fe_mar95_od_SAM Spectra/ROIS was used as input for SAM classification using maximum angle 0.25.

20.04.04 fe_mar95_R0-_water

pc_mar95_R0-_water FE_mar95_R0-_water_convol.img PC_mar95_R0-_water_convol.img


20.04.04 FE_mar95_R0-_water_convol.img PC_mar95_R0-_water_convol.img

FE_mar95_R0-_water_convol_BIL.img PC_mar95_R0-_water_convol_BIL.img


27.04.04 pc_mar95_R0-_water


27.04.04 pc_mar95_R0-_water_convol.img




29.04.04 PCFE_TM5_ mar95_rad_geo_full_R_0.img mar95_R_0_darkest_convol.img mar95_scroll.bmp


29.04.04 mar95_R_0_darkest_convol.img


30.04.04 mar95_R_0_darkest.img

mar95_R_0_darkest_PIF_plot.txt R_0_darkest_PIF_means_plot.txt

PIF_means of darkest deep sea (400 x 400 pixels) (RvE)


04.05.04 PCFE_TM5_mar95_radcor_VNIR.img PCFE_TM5_mar95_radcor_VNIR_band1.dsr

PCFE_TM5_mar95_radcor_VNIR_band2.dsr PCFE_TM5_mar95_radcor_VNIR_band3.dsr PCFE_TM5_mar95_radcor_VNIR_band4.dsr RadCor_lowest_values.xls

Lowest radiance density slices for each band, combined in excel sheet (RvE) turned out not to be darkest area

11.05.04 PCFE_TM5_mar95_rad_geo_full_6bds.img PCFE_TM5_mar95_rad_geo_full_6bds_Rapp.img

Extreme_Radiance_per _Band.xls Coal_PIFs.xls PIFs.roi Coal_PIFs.roi PCFE_TM5_mar95_rad_geo_full_6bds_extremes.roi


12.05.04 PCFE_TM5_mar95_rad_geo_full_6bds_R_0.img PCFE_TM5_mar95_rad_geo_full_6bds_extremes.roi

Mar95_Darkest_Ocean_5x5_lots_plot.txt Darkest_Ocean_5x5_lots_plots.xls


12.05.04 PCFE_TM5_mar95_rad_geo_full_6bds_R_0.img Coastal_5x5_lots_plots_ROI.roi

Mar95_Coastal_5x5_lots_plot.txt Coastal_5x5_lots_plots.xls


12.05.04 PCFE_TM5_mar95_rad_geo_full_6bds_Rapp.img Coal_PIFs.roi



31.05.04 FE_mar95_R0-_water_convol_BIL.img FE_mar95_R0-_water_convol_BIL_MIM_SVDC_opt_phys_04__A_CHL_100.img FE_mar95_R0-_water_convol_BIL_MIM_SVDC_opt_phys_04__BB_CHL_100.img FE_mar95_R0-_water_convol_BIL_MIM_SVDC_opt_phys_04__Kdn_CHL_100.imgFE_mar95_R0-_water_convol_BIL_MIM_SVDC_opt_phys_04_CHL_100.img

Ran MIM on FE R(0-) images. (RvE)

02.06.04 fe_jul02_hightide.img FE_Mask.img Inverted mask. (RvE) 02.05.04 FE_mar95_R0-

_water_convol_BIL_MIM_SVDC_opt_phys_04_CHL_100.img FE_mar95_R0-_water_convol_BIL_MIM_SVDC_opt_phys_04__Kdn_CHL_100.img FE_Mask.img

FE_mar95_R0-_water_convol_BIL_MIM_SVDC_opt_phys_04_CHL_100_masked.img FE_mar95_R0-_water_convol_BIL_MIM_SVDC_opt_phys_04__Kdn_CHL_100_masked.img


03.06.04 FE_mar95_R0- FE_mar95_CDOM_0-06.tif Applied density slices and


_water_convol_BIL_MIM_SVDC_opt_phys_04_CHL_100_masked.img FE_mar95_R0-_water_convol_BIL_MIM_SVDC_opt_phys_04__Kdn_CHL_100_masked.img Mask_Whitish+Nodata.dsr MIM_SVDC_dR0_scale.dsr MIM_SVDC_SIOP_scale.dsr

FE_mar95_dR(0-).tif FE_mar95_Kd_0-3.tif FE_mar95_SD_0-2_m_inversed.tif FE_mar95_SD_0-3_m.tif FE_mar95_SIOPs.tif FE_mar95_Tripton_0-50_mg.tif

stretch and exported to GEOTif. (RvE)

07.06.04 PC_mar95_R0-_water_convol_BIL.img PC_mar95_R0-_water_convol_BIL_MIM_SVDC_opt_phys_04__A_CHL_100.img PC_mar95_R0-_water_convol_BIL_MIM_SVDC_opt_phys_04__BB_CHL_100.img PC_mar95_R0-_water_convol_BIL_MIM_SVDC_opt_phys_04__Kdn_CHL_100.img PC_mar95_R0-_water_convol_BIL_MIM_SVDC_opt_phys_04_CHL_100.img


07.06.04 PC_mar95_hightide.img PC_Mask.img Inverted mask. (RvE) 07.05.04 PC_mar95_R0-

_water_convol_BIL_MIM_SVDC_opt_phys_04_CHL_100.img PC_mar95_R0-_water_convol_BIL_MIM_SVDC_opt_phys_04__Kdn_CHL_100.img PC_Mask.img

PC_mar95_R0-_water_convol_BIL_MIM_SVDC_opt_phys_04_CHL_100_masked.img PC_mar95_R0-_water_convol_BIL_MIM_SVDC_opt_phys_04__Kdn_CHL_100_masked.img


07.06.04 PC_mar95_R0-_water_convol_BIL_MIM_SVDC_opt_phys_04_CHL_100_masked.img PC_mar95_R0-_water_convol_BIL_MIM_SVDC_opt_phys_04__Kdn_CHL_100_masked.img

PC_mar95_CDOM_0-06.tif PC_mar95_dR(0-).tif PC_mar95_Kd_0-3.tif PC_mar95_SD_0-2_m_inversed.tif PC_mar95_SD_0-3_m.tif



Mask_Whitish+Nodata.dsr MIM_SVDC_dR0_scale.dsr MIM_SVDC_SIOP_scale.dsr

PC_mar95_SIOPs.tif PC_mar95_Tripton_0-50_mg.tif

07.06.04 PC_mar95_CDOM_0-06.tif PC_mar95_dR(0-).tif PC_mar95_Kd_0-3.tif PC_mar95_SD_0-2_m_inversed.tif PC_mar95_SD_0-3_m.tif PC_mar95_SIOPs.tif PC_mar95_Tripton_0-50_mg.tif

PC_mar95_CDOM.cdr PC_mar95_dR0.cdr PC_mar95_Kd.cdr PC_mar95_SD.cdr PC_mar95_SD.cdr PC_mar95_SIOPs.cdr PC_mar95_Tripton.cdr


08.06.04 PC_mar95_CDOM_0-06.cdr PC_mar95_dR(0-).cdr PC_mar95_Kd_0-3.cdr PC_mar95_SD_0-2_m_inversed.cdr PC_mar95_SD_0-3_m.cdr PC_mar95_SIOPs.cdr PC_mar95_Tripton_0-50_mg.cdr

PC_mar95_CDOM.jpg PC_mar95_dR0.jpg PC_mar95_Kd.jpg PC_mar95_SD.jpg PC_mar95_SD.jpg PC_mar95_SIOPs.jpg PC_mar95_Tripton.jpg


08.06.04 PC_mar95_CDOM.jpg PC_mar95_dR0.jpg PC_mar95_Kd.jpg PC_mar95_SD.jpg PC_mar95_SD.jpg PC_mar95_SIOPs.jpg PC_mar95_Tripton.jpg


08.06.04 FE_mar95_CDOM.cdr FE_mar95_dR0.cdr FE_mar95_Kd.cdr FE_mar95_SD.cdr FE_mar95_SD.cdr FE_mar95_SIOPs.cdr FE_mar95_Tripton.cdr

FE_mar95_CDOM.jpg FE_mar95_dR0.jpg FE_mar95_Kd.jpg FE_mar95_SD.jpg FE_mar95_SD.jpg FE_mar95_SIOPs.jpg FE_mar95_Tripton.jpg


08.06.04 FE_mar95_CDOM.jpg FE_mar95_dR0.jpg



FE_mar95_Kd.jpg FE_mar95_SD.jpg FE_mar95_SD.jpg FE_mar95_SIOPs.jpg FE_mar95_Tripton.jpg

09.06.04 fe_mar95_opticaldomains_SAM.jpg PC_mar95_substrate_SAM.jpg














09.06.04 FE_02_95_NDVI.jpg


09.06.04 PC_02_95_NDVI.jpg








09.06.04 FE_95_90_NDVI.jpg


09.06.04 PC_95_90_NDVI.jpg


Radiosonde Data Date Input Image/Data Output image/data Processing Notes 3.12.03 UA01D_Data_039083_1990_12267254998.txt

UA01D_Data_039083_1995_12267254999.txt|ATCOR_ancillary\PCFE_radiosonde.xls Used measured radiosonde data where

available and Maher and Lee averages


UA01D_Data_039083_2002_12267255003.txt (Brisbane and Townsville) for August to estimate precipitable water (JMA)

Tape files for c-WOMBAT-c Date Input Image/Data Output image/data Processing Notes 11.12.03 PCFE_02_Spectral_big.lib/

PCFE_02_Spectral_big.roi PCFE_95_Spectral_bigger.lib/ PCFE_95_Spectral_roi_bigger.roi PCFE_90_Spectral_roi_bigger.lib/ PCFE_90_Spectral_roi_bigger.roi PCFE_86_Spectral_roi_bigger.lib/ PCFE_86_Spectral_roi_bigger.roi

\ATCOR\PCFE_TM7_240702_12.7sc \ATCOR\PCFE_TM5_070395_11.7sc \ATCOR\PCFE_TM5_160890_2.7sc \ATCOR\PCFE_TM5_210886_4.7sc

Used PIFs data, and coincident ROIS over deep water to ‘fine tune’ the parameterisation. Maintained tropical atmosphere for all 4 dates and varied the horizontal visibility (‘02=100km, ‘95=75km, ‘90=80km, ‘86=45km). Measured and modelled ozone and water vapour used. (JMA)

QC 16.12.03 PCFE_jul02_rad_geo_full_6bds_R_0.img

PCFE_mar95_rad_geo_full_6bds_R_0.img PCFE_aug90_rad_geo_full_6bds_R_0.img PCFE_aug86_rad_geo_full_6bds_R_0.img

\ATCOR\PCFE_jul02_R_0_negatives.jpg \ATCOR\PCFE_mar95_R_0_negatives.jpg\ATCOR\PCFE_aug90_R_0_negatives.jpg\ATCOR\PCFE_aug86_R_0_negatives.jpg

Density slice applied to R(0-) images Red < -0.005>Green<0 (JMA)


Appendix 4 - Publications produced in association with the PC2 and FE2 Tasks

Book Chapters Dekker, A., Brando, V., Anstee, J., Pinnel, N., Kutsre, T., Hoogenboom, E., Peters, S., Pasterkamp, R., Vos, R., Olbert, C. and Mathus, T. (2001) Imaging spectrometry of water. In: F.D. van der Meer, S. and M. de Jong Eds. Remote Sensing and Digital Image Processing Volume 4: Imaging Spectrometry, Ch. 11, Kluwer. Dekker, A. and Mertes, L. (2004) Remote sensing of coastal and inland waters. In Ustin, S.(Ed.) Manual of Remote Sensing, Volume 4, Remote Sensing for Natural Resource Assessment. American Society for Photogrammetry and Remote Sensing, Bethesda, Maryland. In press 05/04. Phinn, S., Scarth, P., Roelfsema, C. and Joyce, K. (2003) The role of integrated information acquisition and management in the analysis of coastal ecosystem change. In Le Drew, E. and Richardson, L. (Eds.) Remote sensing of coastal aquatic ecosystem processes. Remote Sensing and Digital Image Processing Series - Kluwer Academic Publishers. In preparation 06/03. Refereed Papers Brando, V. and Dekker, A. (2003) Satellite hyperspectral remote sensing for estimating estuarine and coastal water quality. IEEE Transactions on Geosciences and Remote Sensing, In press 12/02. Manson, F.J. , Lonergan, N.R. and Phinn, S.R. (2003) Spatial and temporal variation in distribution of mangroves in Moreton Bay, subtropical Australia: a comparison of pattern metrics and change detection analyses based on aerial photographs. Estuarine, Coastal and Shelf Science, 57(4), 657-670. Phinn, S., Stow, D., Franklin, J., Mertes, L. and Michaelsen, J. (2003) Remotely sensed data for ecosystem analyses: Combining hierarchy and scene models, Environmental Management, 31(3):429-441. Phinn, S., Nightingale, J. and Stanford, M. (2002) The Status of Remote Sensing for Environmental Monitoring and Management Applications in Australia. GIS User, April-May Issue, 51:26-27 Phinn, S. and Stanford, M. (2001) Monitoring land-cover and land-use change in a rapidly urbanising coastal environment, the Maroochy and Mooloolah Rivers catchment, 1988-1997, Australian Geographical Studies, 39(2):217-232. Phinn, S.R. , Menges , C. , Hill, G.J.E. and Stanford, M. (2000) "Optimising remotely sensed solutions for monitoring, modelling and managing coastal environments," Remote Sensing of Environment, 73(2):117-132. Roelfsema, C., Dennison, W.C. and Phinn S.R. (2002) Spatial distribution of benthic microalgae on coral reefs as determined by remote sensing. Coral Reefs, 21(3), 264-274. Conference Papers Phinn, S., Held, A., Lucas, R., Knight J. and Dale, P. (2000) Monitoring and Mapping The Composition, Structure And Dynamics Of Tropical Mangrove Environments In Australia: A Multi-


Sensor Approach, Invited paper in: Proceedings of Intecol Wetlands 2000 Conference, Quebec, August. Knight, J., Phinn, S.R. and Dale, P. (2000) Integration of optical, thermal and synthetic aperture radar systems for mapping inundation in mangrove environments. Invited paper In: Proceedings of the 10th Australasian Remote Sensing and Photogrammetry Conference, Adelaide, August. Phinn, S., Scarth, P., Stanford, M. Hill, G., Smith, L. and Tomlinson, P. (2000) Applications development for an integrated airborne and field remote sensing system in tropical wetland, rainforest and coral reef environments In: Proceedings of the 10th Australasian Remote Sensing and Photogrammetry Conference, Adelaide, August. Phinn, S., Roelfsema, C., Dekker, A. and Brando, V. (2001) Approaches for monitoring benthic and water surface biophysical properties in Australian coastal environments. In: Proceedings of the International Geosciences and Remote Sensing Symposium 2001, Sydney, CD-ROM Proceedings, IEEE-Piscataway NY, USA. Knight, J., Scarth, P., Roelfsema, C., Phinn, S. and Dale, P. (2001) Integration of Airborne Remote Sensing Data for Mapping Mangrove Ecosystems in Moreton Bay, South East Queensland. In: In: Proceedings of the International Geosciences and Remote Sensing Symposium 2001, Sydney, CD-ROM Proceedings, IEEE-Piscataway NY, USA. Held, A., Phinn, S., Scarth, P., Ticehurst, C. and Lymburnber, L. (2001) Hyperspectral Mapping of Rainforests and Mangroves. In: Proceedings of the International Geosciences and Remote Sensing Symposium 2001, Sydney, CD-ROM Proceedings, IEEE-Piscataway NY, USA. Dekker, A., Brando, V., Anstee, J., Roelfsema, C., Phinn, S. and Dennison W. (2001) Hyperspectral airborne mapping of cyanobacteria. In Proceedings of the 5th International Conference on Toxic Cyanobacteria, Noosa Heads, July, In Press. Brando, V. (2002) “Calibration and validation of hyperspectral data in Australian coastal waters” Remote Sensing of Coastal Waters. In: Proceedings of The Remote Sensing of Coastal Waters Workshop - Italian-Australian Exhibition and Conference on Innovation Technologies, Rialto Hotel, Melbourne, March 27, 2002. Dekker, A. (2002) “Algorithm development in Australian Optically shallow waters” Remote Sensing of Coastal Waters. In: Proceedings of The Remote Sensing of Coastal Waters Workshop - Italian-Australian Exhibition and Conference on Innovation Technologies, Rialto Hotel, Melbourne, March 27, 2002. Phinn, S., Joyce, K., Nightingale, J. and Stanford, M. (2002) “Coastal Monitoring and Management Needs – Remotely Sensed Data and Solutions?” Remote Sensing of Coastal Waters. In: Proceedings of The Remote Sensing of Coastal Waters Workshop - Italian-Australian Exhibition and Conference on Innovation Technologies, Rialto Hotel, Melbourne, March 27, 2002. Brando, V. Dekker, A. and Anstee, J. (2002) Estuarine hyperspectral remote sensing from space: Moreton Bay case study. In: Proceedings of the 11th Australasian Remote Sensing and Photogrammetry Conference, Brisbane, 2-6 September, CD-Rom, Causal Publications. Dekker, A.G., V.E. Brando, P.J. Daniel, J.M. Anstee, L Clementson. Psicam - from myth to reality. Ocean Optics Xvi - Conference November 18 – 22, 2002 - Santa Fe, New Mexico. Brando, V.E. and A G. Dekker (2002) The fluorescence term on the observed 690-710 nm reflectance peak in eutrophic turbid (inland) waters: myth or reality? Ocean Optics Xvi - Conference November 18 – 22, 2002 - Santa Fe, New Mexico.


Knight, J., Dale, P. and Phinn, S. (2002) Mapping mosquito breeding habitats (water) under mangrove canopy: contribution from airborne thermal scanning. In: Proceedings of the 11th Australasian Remote Sensing and Photogrammetry Conference, Brisbane, 2-6 September, CD-Rom, Causal Publications. Manson, F., Lonergan, N., Skilleter, G. and Phinn, S. (2002) Establishing links between mangroves and fisheries in Australia: a conceptual framework, In: Proceedings of the Ecological Society of Australia, Cairns, December, 2002. Phinn, S., Held, A., Stanford M., Ticehurst, C. and C. Simpson (2002) Optimising State of Environment Monitoring at Multiple Scales Using Remotely Sensed Data. In: Proceedings of the 11th Australasian Remote Sensing and Photogrammetry Conference, Brisbane, 2-6 September, CD-Rom, Causal Publications.

Roelfsema, C., Phinn S., Dennison, W., Dekker, A and Brando, V (2002) Monitoring cyanobacterial blooms of Lyngbya Majuscula in Moreton Bay, Australia by combining field techniques with remote sensing. In: Proceedings of the 11th Australasian Remote Sensing and Photogrammetry Conference, Brisbane, 2-6 September, CD-Rom, Causal Publications. Roelfsema, C., Phinn, S., Scarth, P. Dekker, A. and Brando, V. (2003) Mapping a cyano-bacterial bloom (Lyngbya majuscula) in Moreton Bay, Australia : A comparison of hyperspectral and multispectral approaches. In: Proceedings of the 30th International Symposium on Remote Sensing of Environment, Honolulu Sheraton Hawaii, November 10-14, In press. Knight, J., Phinn, S. and Dale, P. (2003) Mapping mosquito breeding habitats in mangrove forests wetlands: Defining an appropriate spatial metric. In: Proceedings of the 30th International Symposium on Remote Sensing of Environment, Honolulu Sheraton Hawaii, November 10-14, In press.

Reports Phinn, S., Nightingale, J. and Stanford, M. (2001) Review of the Status of Remote Sensing for Environmental Monitoring and Management Applications in Australia. Research Monograph, The University of Queensland Foundation Research Excellence Awards for Early Career Researchers, 2000. Roelfsema, C. and Phinn, S. (2003) Development and Implementation of a Lyngbya majuscula Monitoring Program for the Eastern Banks, Moreton Bay. Report prepared for the Lyngbya Monitoring Project - Moreton Bay Catchment and Waterways Partnership, Brisbane, 54p. Roelfsema, C. and Phinn, S. (2002) Evaluating the Feasibility of Remote Sensing for Monitoring Seagrass Beds in the Fisherman’s Island Region, Moreton Bay. Report prepared for the Port of Brisbane Authority and WBM Consulting Engineers, 21p. Software c-Wombat-c IDL (Research Systems Inc) radiative transfer code for correction of atmospheric and air-water interface effects; converts at-sensor image pixel values to subsurface irradiance reflectance R(0-) or apparent reflectance.

Final Report PC2-FE2 - June 2005 - Land and Water - CSIRO · water. Beam Attenuation Coefficient...

Documents

Transcript of Final Report PC2-FE2 - June 2005 - Land and Water - CSIRO · water. Beam Attenuation Coefficient...