Uploaded August 2023 | Updated September 2026, 2 days ago
Watch 2023 NESSI intern Isabela Suaza-Sierra's end-of-summer presentation, "Integrating Hydrologic and Machine Learning Models to Predict Lake Water Temperature Profiles and Release Temperatures throughout the Red River Basin, USA."
This study focuses on the Red River Basin in the southern United States, with important water resources to support fish populations, societal water needs, and streamflow in the region. We developed and calibrated a daily WEAP ("Water Evaluation And Planning" system) water systems model to enable the exploration of future water use and management scenarios under climate change conditions. This model estimates the basin's streamflows, reservoir storage volumes, and inflows, utilizing terrain, atmospheric forcing, and land cover parameters. The calibration process involved 25 streamflow USGS gauge stations, daily discharge data from 1996 to 2010, and manual soil parameters calibration. Additionally, we explored machine learning methods (Random forest) to predict vertical lake/reservoir temperature profiles and release temperatures, which are crucial for understanding and managing freshwater ecosystems. By incorporating these predictions as a predictive function in WEAP, we demonstrate the potential benefits of the integrated approach. Future work includes validation and calibration of additional streamflows (2010-2020), reservoir operations, and water temperature observations. The study's findings contribute to better water resource management and climate change adaptation strategies in the Red River Basin.
Watch 2023 NESSI intern Isabela Suaza-Sierra's end-of-summer presentation, "Integrating Hydrologic and Machine Learning Models to Predict Lake Water Temperature Profiles and Release Temperatures throughout the Red River Basin, USA."
This study focuses on the Red River Basin in the southern United States, with important water resources to support fish populations, societal water needs, and streamflow in the region. We developed and calibrated a daily WEAP ("Water Evaluation And Planning" system) water systems model to enable the exploration of future water use and management scenarios under climate change conditions. This model estimates the basin's streamflows, reservoir storage volumes, and inflows, utilizing terrain, atmospheric forcing, and land cover parameters. The calibration process involved 25 streamflow USGS gauge stations, daily discharge data from 1996 to 2010, and manual soil parameters calibration. Additionally, we explored machine learning methods (Random forest) to predict vertical lake/reservoir temperature profiles and release temperatures, which are crucial for understanding and managing freshwater ecosystems. By incorporating these predictions as a predictive function in WEAP, we demonstrate the potential benefits of the integrated approach. Future work includes validation and calibration of additional streamflows (2010-2020), reservoir operations, and water temperature observations. The study's findings contribute to better water resource management and climate change adaptation strategies in the Red River Basin.
