Sediment Dynamics in the Brazos and Colorado Rivers: Insights from Satellite Imagery and Machine Learning

Abstract Submission: Understanding the sediment dynamics of the Brazos and Colorado Rivers is vital for the effective management of sediment transport and deposition in the Gulf of Mexico. Satellite imagery combined with machine learning models such as Random Forest (RF) and Artificial Neural Networks (ANN) allows for the accurate estimation of suspended sediment concentration (SSC) to aid in understanding sediment dynamics in the Gulf of Mexico. In the Brazos River, a model developed to estimate SSC using Sentinel and Landsat mission satellite imagery achieved correlation coefficients (R²) of 0.83 or higher, demonstrating the model's accuracy. This model was used to understand sediment dynamics during high flow conditions experienced in the Brazos River during Hurricane Harvey. In the Colorado River, distinct models for low-flow and high-flow conditions, below and above 1100 cfs (31 m³/s), the 2-year return period flow, were developed using field-measured SSC data and satellite imagery. This differentiation was essential due to notable variations in sediment transport between flow regimes. Both models, with correlation coefficients exceeding 0.8, successfully captured sediment dynamics under varying flow conditions. Additionally, the effective discharge, representing the flow that transports the most sediment over time, was identified. These models provide critical insights into sediment behavior, particularly in the Colorado River, where upstream dam discharges and the absence of tidal gauges present operational challenges. This research advances SSC estimation methodologies and supports informed sediment management practices, aiding the protection of ecosystems and infrastructure along the Gulf Coast.

Learning Objectives/Expected Outcome (Optional) :