Dissertation Defense-Punwath Prum

Leveraging Satellite Remote Sensing and Machine Learning to Monitor Global Estuarine Systems Estuaries are a unique ecosystem where inland waters and the ocean interact. Their responses to climate change and human disturbance are understudied on a global scale. This thesis leverages multi-decadal satellite data and machine learning methods to study environmental changes in estuaries around the world. First, Punwath presents a spatial and temporal analysis of estuarine surface water temperature from 1985 to 2022 across 1,060 estuaries. 47% of Earth's estuaries are warming, with a global average warming rate of 0.070 ± 0.004 °C yr⁻¹ (median = 0.060 °C yr⁻¹). Second, he presents a global model for estimating suspended sediment concentrations (SSC) at the water surface from multiple Landsat 5, 7, and 8 satellite datasets. This model was robustly designed to estimate SSC in various optical watercolor types—such as blue-green, green, and brown water—in which each water type is dominated by different optically active constituents. Lastly, he analyzed lateral salt marsh changes in 356 estuaries and topographic position changes in 290 estuaries across different morphodynamic systems (river-, tide-, and wave-dominated) worldwide. The data showed a global pattern of net estuarine lateral marsh loss of −353.65 km² between 2000 and 2019, with 223 estuaries experiencing net lateral marsh loss. Net loss scaled with estuarine size at a rate of 0.0012 km² of marsh per 1 km² of estuarine area (OLS, p < 0.05). Marshes in 184 of 290 estuaries transgressed upslope (mean = 0.85 m, median = 0.43 m), whereas marshes in the remaining 106 estuaries transgressed downslope (mean = −0.65 m, median = −0.24 m). This thesis advances our understanding of environmental changes in Earth’s estuaries under climate change and anthropogenic processes.