Recently, there has been growing interest in using gridded snowpack datasets to improve water supply forecasts in the western united states

The University of Arizona, through a $996,526.93 Cooperative Agreement award from the Bureau of Reclamation under the Snow Water Supply Forecasting Program (CFDA 15.078), will use a 40-year snowpack dataset and newly developed regression modeling and machine learning techniques to enhance water supply forecasts across the Colorado River Basin. Specifically, the project will incorporate improved snowpack data and model optimizations into the National Resources Conservation Networks' next-generation multi-model machine-learning metasystem to provide real-time water supply forecasts for locations such as the Salt-Verde Basin in Arizona. The goal is to leverage advanced snow monitoring approaches and hydrologic modeling to improve the accuracy of water supply predictions in drought-prone regions of the western United States. The award has an ultimate completion date of September 30, 2026.

However, many of these gridded datasets do not have the long temporal record needed to train the models producing the water supply forecasts. This project will use the university of arizona snowpack dataset, which spans 40 years, to improve water supply forecasts across the colorado river basin. The project will improve these snowpack data using recently developed regression modeling and machine learning of lidar data from across the colorado river basin. The project will investigate how the snowpack data, along with model optimizations of the sacramento soil moisture accounting, snow-17 and noah-MP models, can be used to enhance streamflow modelling, and will incorporate the gridded snow data into the national resources conservation networks next-generation multi-model machine-learning metasystem model. The updated models will be used to provide real-time water supply forecasts for various locations across the colorado river basin, such as the salt-verde basin in arizona as well as more broadly across the colorado river basin using a web decision support tool.