ReVap: Monitoring and Forecasting Daily Reservoir Evaporation in the Western U.S.

The western United States is increasingly experiencing the compound impacts of climate change, with more frequent and severe droughts threatening the sustainability of water resources. In this context, accurate estimation of reservoir evaporation—a key contributor to water loss—becomes critical for effective water resource management. Over the past decades, the quantification of reservoir evaporation loss has relied on pan evaporation observations, which are subject to large errors due to multiple factors. Additionally, there is a lack of evaporation forecasts to meet water management needs. Therefore, our objective is to develop long-term, near real-time, and sub-seasonal forecasts of daily evaporation estimates that are physically based, accurate, and easily retrievable via relational databases and Application Programming Interface (API).

This presentation introduces the reservoir evaporation monitoring and forecasting system (ReVap) for the western U.S., where water scarcity poses significant challenges. The Daily Lake Evaporation Model (DLEM) is applied to over 800 reservoirs in this arid and semi-arid regions region. The input meteorological data are collected from the Real-Time Mesoscale Analysis (RTMA) dataset and gridMET data, with the latter being bias-corrected using RTMA. The model has been validated using eddy covariance data collected from four reservoirs, achieving a root mean square error (RMSE) between 1.13 and 1.97 mm/day, while the R² ranged from 0.50 to 0.73, demonstrating the model's robustness in capturing the complexities of evaporation process. This long-term record starts in 1980 and is produced operationally to near real-time (lagged by 4 days). The forecast data have a lead time of 28 days, containing representative forecasts at 10^th, 25^th, 50^th, 75^th, and 90^th percentiles. The data are made available to the public via an operational API.

Outputs from DLEM provide valuable daily evaporation estimates that are essential for forecasting changes in reservoir storage, water surface elevation, and operational considerations, particularly in the face of climate variability. These estimates support the development of adaptive strategies for water resource management, helping to mitigate the uncertainties associated with hydrologic forecasting under variable climatic conditions. By integrating our model's outputs into decision-making processes, decision-makers can more effectively plan for and respond to the challenges posed by climate change. This research contributes to the broader goal of enhancing climate-resilient water resource management, offering a multiscale approach that bridges the gap between global climate models and localized hydrologic forecasting.