Optical remote sensing of coloured dissolved organic matter (cdom) for better estimation of dissolved organic carbon (doc) in south-east Queensland water reservoirs

Optical remote sensing of coloured dissolved organic matter (cdom) for better estimation of dissolved organic carbon (doc) in south-east Queensland water reservoirs

Australian inland water-bodies play a great role in the carbon cycle at the regional level and contribute effectively through the carbon exchange between their surface and the atmosphere. This carbon enters the aquatic system from terrestrial sources such as soil or it is generated within the water body itself through the primary production of phytoplankton or from organic matter. It is converted into dissolved organic carbon (DOC) which constitutes about 90% of the dissolved carbon in the aquatic system. Increasing the DOC levels in the water affects the water quality and ecosystem functioning. Remote sensing methods allow DOC to be monitored with a minimum of time,effort and cost. DOC has no colour and remote sensing cannot determine it directly, however, the coloured portion of the dissolved organic matter (CDOM) can be used as a proxy to estimate its amount and its concentration in the aquatic environment. But, remote sensing measurements of DOC can only be done if there is a good relationship between CDOM absorption and DOC concentration. This good relationship does not necessarily exist over many of the water-bodies because CDOM concentration varies both spatially and temporally according to its sources. This weak relationship was observed in the study area South East Queensland (SEQ) as reviewed in this thesis. Therefore, the researcher investigated different ways to improve the characterisation of this relationship in the study area to get a better estimation of DOC based on using remotely sensed CDOM concentrations.

The researcher sampled 11 discrete water bodies in South East Queensland (SEQ), then examined if there was a correlation between CDOM absorption coefficient and DOC but, the overall results showed a weakly positive correlation relationship among all reservoirs. The reason for this poor relationship is attributed to the potential impacts of the allochthonous source on the inputs DOC to water derived from the surrounding areas. Different estimation enhancement approaches were investigated by including both CDOM absorption and its slope in multiple linear regressions which relatively improved the estimation of DOC concentration and assisted in obtaining a better understanding of this relationship.

Then, another investigation was done by fractionatingCDOM into its major groups of humic and non-humic substances and measuring their absorption spectra and DOC contentsseparately. Solid-phase extraction (SPE) technique was used as a chemical separation method. The investigation in the relationship between CDOM fractions and DOC confirmed different relationships such as not all DOC in CDOM can be chromophoric.

Furthermore, the CDOM absorption spectrum shape is a proxy of CDOM composition in water. Therefore, it is important that CDOM absorption spectrum curve be clear, clean and free from any errors if possible (systematic or random) to give better estimation results. But, the most important is using the correct fitting model that characterizes the CDOM absorption spectrum accurately. Some fitting models can lead to a loss in and not fully capture all the information provided by CDOM absorption curve. The CDOM spectral decomposition technique was used to provide better and additional information about CDOM pool and dynamics from the absorption spectrum curve that was done by using the algebraic method of the linear, nonlinear and Gaussian decomposition approaches. The final results of using CDOM spectral decomposition were useful and helpful forgiving a good explanation to the relationship between CDOM absorption and DOC concentration on the one hand and between CDOM absorption and its sources on the other. Also, the advantage of using the multi-exponential model is convenient for optical modelling and remote sensing applications.

Finally, the researcher examined how CDOM sources (allochthonous and autochthonous) in SEQ affect the optical properties and on the remote sensing reflectance by separating CDOM into its major groups of humic and non-humic and modelling them. ECOLIGHT® simulation used first to simulate subsurface irradiance reflectance(R(0−))curves under the conditions found in SEQ waterbodies. The aim to show the contribution of other water components(phytoplankton and tripton) ofdifferent concentrations as well asCDOM-SIOPs (specific inherent optical properties) on the water reflectance. ECOLIGHT® hasa unique ability to isolate changes in the reflectance due to SIOPs or spectral variability. On the other hand, R(0−)modelled using a developed semi-analytical bio-optical model. A multi-components bio-optical model developed in this work by using and inserting SIOPs of CDOM fractions. The retrieving results of CDOM from water reflectance much better when using CDOM fractions SIOPs which reflected positively on estimating DOC.

The study concluded that the estimation of DOC concentrations from water's colour is more complex and the accuracy factor is limited, due to the confounding effects of water components and sources in addition to the poor performance of the standard models and algorithms. Also, CDOM fractions are participating in CDOM absorption spectrumshapewhich can give us good information and can be used to estimate DOC. The results of the spectral decomposition showed that SEQ water bodies tend to be dominated by humic acid due to the high ratio of HA compared to FA. Another finding, the calculated slope values of the study area were superior to the calculated slope values of Kirk for the Australian inland waters when they used in CDOM decomposition model to characterize the relationship with DOC. While Suwannee River slope values were not applicable within Australian inland waters when it used in CDOM decomposition model.