This project will develop a long short-term memory model using near-daily cubesat data, ground observations, and lidar snow maps from the airborne snow observatories, INC., to generate an advanced…

This Cooperative Agreement award from the U.S. Department of the Interior's Bureau of Reclamation, under the Snow Water Supply Forecasting Program (CFDA 15.078), will develop an advanced snow estimation product using a Long Short-Term Memory (LSTM) model. The $1,270,676.70 project, with an award date of January 30, 2025 and a completion date of December 31, 2027, will leverage near-daily CubeSat data, ground observations, and LiDAR snow maps to enhance a process-based distributed hydrologic model. This collaborative effort between Arizona State University and the Salt River Project will improve snow estimation in forested environments to support water supply, flood hazard, and water resilience forecasting. The LSTM-based snow water equivalent product will be compared to an independent gridded product and then applied to initialize reservoir inflow forecasts for short-, medium-, and long-term timeframes. The outcomes will be incorporated into the Salt River Project's reservoir operations plan, providing enhanced water supply forecasting capabilities.

This project will develop a long short-term memory model using near-daily cubesat data, ground observations, and lidar snow maps from the airborne snow observatories, INC., to generate an advanced snow estimation product. Applied for the first time in arizona, airborne snow observatories, INC. Snow maps will be used to enhance a process-based distributed hydrologic model. The project will use the high-resolution hydrologic model to translate weather and snowpack conditions to forecasted inflows to a reservoir system. This collaborative effort between arizona state university and the salt river project will improve snow estimation in forested environments for water supply, flood hazard, and water resilience forecasting. As the largest water supplier for the phoenix metropolitan area, salt river project operates six reservoirs in the salt and verde river basin. This modeling approach will provide salt river project with new water supply forecasting capabilities. A snow water equivalent product will be developed from near-daily planetscope (3-meter, 8-band) and skysat (0.5-meter, 4-band) multispectral imagery merged with ground and aerial data (snow telemetry, snowtography, lidar surveys) and constrained by daily, high-resolution snow water equivalent from the hydrologic model through the long short-term memory approach. The long short-term memory snow water equivalent product will be compared to an independent gridded product and then applied to initialize the snow water supply forecasts that generate reservoir inflows for short-, medium-, and long-lead times. Outcomes will be incorporated into the salt river project reservoir operations plan.