POSTER Session 4

DEEP LEARNING MODEL TO ESTIMATE RIVERINE, ESTUARINE AND COASTAL WATER QUALITY USING HIGH-RESOLUTION SATELLITE IMAGERY

Mapping Water Quality (WQ) parameters such as Total Suspended Solids (TSS) and Dissolved Organic Carbon (DOC) through satellite remote sensing approaches is crucial for understanding the health of large-scale aquatic ecosystems. This study maps fine-resolution optical satellite-derived TSS and DOC from catchments across estuarine and coastal areas. We obtained field measurements, including absorption and backscattering coefficients of particulate and dissolved substances across three Southeast Australian coastal catchments for subsequent bio-optics modeling. Simulated reflectance was derived from these measurements using the Hydrolight radiative transfer model. This multispectral data and associated uncertainty in the measurements served to train a Dense Deep Learning Network (DDLN). The model performed well with Mean Absolute Error (MAE) of 7% TSS and 20% DOC in optically complex waters, thereby capturing the spatial variability of optical properties. The DDLN inversion performs better than existing physics-based analytical models with faster inversion execution time and sensitivity to high and low flow conditions. We then invert the DDLN on atmospherically corrected Landsat-8 imagery available between 2013 and 2022 for the three catchments. Time-series analysis revealed that the DDLN distinctly identified sediment and carbon concentration between contrasting land-cover catchments. Large spatial variations in maximum TSS and DOC plume extents were mapped across all three catchments. Significant correlations exist between the annual medians of rainfall intensity across the catchments and the TSS/DOC at the river mouths. The DDLN thus proved effective in nonlinearly mapping reflectance and WQ, facilitating climate-scale TSS and DOC estimates.

S.L. Kesav Unnithan, Commonwealth Scientific and Industrial Research Organisation (CSIRO), Australia, [email protected], https://orcid.org/0000-0002-8758-2554

Nagur Cherukuru, Commonwealth Scientific and Industrial Research Organisation (CSIRO), Australia, [email protected], https://orcid.org/0000-0002-3617-5852

Nathan Drayson, Commonwealth Scientific and Industrial Research Organisation (CSIRO), Australia, [email protected], https://orcid.org/0000-0003-1248-0593

Tim Ingleton, New South Wales Department of Climate Change, Energy, the Environment and Water (DCCEEW), Australia, [email protected], https://orcid.org/0000-0003-4322-6758

Gemma Kerrisk, Commonwealth Scientific and Industrial Research Organisation (CSIRO), Australia, [email protected], https://orcid.org/0000-0001-8980-8038

Code of Conduct
Privacy Policy