Dirty water: remote sensing of water quality in tropical and sub-tropical freshwater impoundments

Campbell, Glenn ORCID: https://orcid.org/0000-0002-4249-2512 (2011) Dirty water: remote sensing of water quality in tropical and sub-tropical freshwater impoundments. [Thesis (PhD/Research)]


Download (19MB)


Tropical and sub-tropical lakes and reservoirs not only provide drinking water and locations for recreation, fisheries and aquaculture but they have also been shown recently to have a disproportionate effect on the on the global carbon cycle. The purpose of this study was to investigate how techniques developed for the remote sensing of water quality parameters(chlorophyll a, tripton and coloured dissolved organic matter (CDOM)) in inland waters in temperate northern hemisphere environments could be adapted or improved to allow them to be applied to tropical and sub-tropical water bodies. This aim was achieved by adapting existing image based atmospheric correction techniques, by measuring and modelling the specific inherent optical properties of water quality parameters in a selection of the Northern Australian water bodies and by modifying existing inversion algorithms. The final algorithms were applied to a tropical water body using the MERIS sensor to determine the monitoring accuracy and precision that could be expected for each water quality parameter.

The water quality parameter specific inherent optical properties (SIOPs) were measured in three large water storages in north-eastern Australia, Wivenhoe Dam (27° 21´ S, 152° 36´ E), Fairbairn Dam (23° 42´ S, 148° 02´ E) and Burdekin Falls Dam (20° 37´ S, 147° 0´ E). Three existing
MERIS atmospheric correction methods were applied and found to be unsuitable because they use inappropriate water leaving radiance assumptions. A site specific atmospheric correction method was developed that utilised the 6S atmospheric model and the reflectance of the dense dark vegetation that surrounded the study sites. The study used two inversion methods, the direct, Matrix Inversion Method (MIM) and the stochastic, Particle Swarm Optimisation (PSO.
Both methods were implemented on an over-determined system of reflectance equations with semi-analytic models of the anisotropy of the in-water light field. The MIM used differential weighting for each sensor band and the PSO used four different reflectance matching criteria. The SIOPs, the typical water quality constituent concentrations and the Hydrolight® radiative transfer model were used to simulate reflectance spectra that could be employed to parameterise the reflectance models and investigate how the inversion methods performed in the presence of noise. The methods were applied to two images of Burdekin Falls Dam and the results were validated against in situ measurements.

The results of the application of the MIM algorithm showed that the best weighting scheme had a mean chlorophyll a retrieval error of 1.0 μg/l, the conventional three band scheme had a mean error of 4.2 μg/l and the unweighted scheme had a mean error of 5.5 μg/l. For tripton, the best
performed weighting scheme had a mean error of 1.2 mg/l, the three band scheme had a mean error of 3.4 mgl-1 and the unweighted scheme had a mean error of 1.8 mgl-1. For the CDOM retrieval, the mean error was found to be 0.12 /m for the best performed weighting scheme,0.25 /m for the three band scheme and 0.52 /m for the unweighted scheme. In the case of the PSO the mean retrieval error of the best performed similarity measure was 2.0 μg/l, 2.45 mg/l and 0.3 /m for chlorophyll a, tripton and CDOM respectively.

The study concluded that the atmospheric correction methods for MERIS images of Northern Australian inland waters cannot rely on site independent a priori knowledge of the water leaving radiance that has been developed from other environments and that instead, images of inland water bodies can be corrected by taking advantage of dense dark vegetation surrounding the impoundment. It also concluded that there was sufficient intra-impoundment variation in the
specific absorption and specific scattering of phytoplankton and tripton to require a well distributed network of measurement stations when characterising a new water body and that some inland water bodies may need more than one SIOP set to characterise the optical domains present. Significant improvements in the accuracy and precision of retrieved water quality parameter values can be obtained by using semi-analytically estimated values for the anisotropy factor and that over-determined systems of equations can be used to mitigate the effect of unknown and inherent sources of error in the remote sensing system. After application of the modified retrieval algorithms it was found that optical closure can be used to identify the most appropriate SIOP set in water bodies that have multiple SIOP domains, the over-determined weighted MIM algorithm can be more accurate and precise than the conventional three band or unweighted approach and the PSO does not offer improvements in accuracy and precision sufficient enough to justify the increased computational overhead in the inversion.

Statistics for USQ ePrint 18710
Statistics for this ePrint Item
Item Type: Thesis (PhD/Research)
Item Status: Live Archive
Additional Information: Doctor of Philosophy )PhD), University of Queensland.
Faculty/School / Institute/Centre: Historic - Faculty of Engineering and Surveying - No Department (Up to 30 Jun 2013)
Faculty/School / Institute/Centre: Historic - Faculty of Engineering and Surveying - No Department (Up to 30 Jun 2013)
Supervisors: Phinn, Stuart; Dekker, Arnold; Brando, Vittorio
Date Deposited: 06 Jul 2011 02:20
Last Modified: 03 Jul 2013 00:34
Uncontrolled Keywords: remote sensing, inland water, absorption, backscattering, Matrix Inversion Method, Particle Swarm Optimisation, atmospheric correction, phytoplankton, tripton, coloured dissolved organic matter
Fields of Research (2008): 09 Engineering > 0909 Geomatic Engineering > 090905 Photogrammetry and Remote Sensing
05 Environmental Sciences > 0502 Environmental Science and Management > 050206 Environmental Monitoring
Fields of Research (2020): 40 ENGINEERING > 4013 Geomatic engineering > 401304 Photogrammetry and remote sensing
41 ENVIRONMENTAL SCIENCES > 4104 Environmental management > 410404 Environmental management
Socio-Economic Objectives (2008): D Environment > 96 Environment > 9611 Physical and Chemical Conditions of Water > 961103 Physical and Chemical Conditions of Water in Fresh, Ground and Surface Water Environments (excl. Urban and Industrial Use)
URI: http://eprints.usq.edu.au/id/eprint/18710

Actions (login required)

View Item Archive Repository Staff Only