Skill of rain-snow level forecasts for landfalling atmospheric rivers: A multi-model model assessment using California’s network of vertically profiling radars

January 30, 2020

Water resources managers must deal with substantial uncertainty in the forecasting of rain-snow levels during atmospheric river (AR) events in California, a new CW3E study shows. Major winter AR storms bring a highly variable mix of rain and snow to the Sierra Nevada; warmer events have higher rain-snow levels and thus produce more rain, greater flood risk, and contribute less to seasonal snowpack and water supply. The study showed that while operational weather models can forecast rain-snow levels well on average, they frequently had errors in rain-snow levels of several hundred meters, large enough to introduce major uncertainty in flood and water resources planning. This study is a part of CW3E’s ongoing effort to understand and improve the predictions of ARs and their impacts on public safety and water management, supporting local water agencies, California Department of Water Resources, the U.S. Army Corps of Engineers, and other agencies.

In the study by Brian Henn, Rachel Weihs, Andrew C Martin, F. Martin Ralph, and Tashiana Osborne of CW3E, forecasts of atmospheric rain-snow levels from December 2016 to March 2017, a period of active AR landfalls, were evaluated using 19 profiling radars in California. Three forecast model products were assessed: a global forecast model downscaled to 3 km grid spacing, 4 km river forecast center operational forecasts, and 50 km global ensemble reforecasts. Model forecasts of the rain-snow level were compared with observations of rain-snow melting level brightband heights. Models produced median bias magnitudes of less than 200 m across a range of forecast lead times. However, error magnitudes increased with lead time and were similar between models, averaging 342 m for lead times of 24 hr or less and growing to 700-800 m for lead times of greater than 144 hr. Significantly for flood forecasting, observed extremes in the rain-snow level were underestimated, particularly for warmer events, and the magnitude of errors increased with rain-snow level. Storms with high rain-snow levels were correlated with larger observed precipitation rates in Sierra Nevada watersheds. Flood risk increases with rain-snow levels, not only because a greater fraction of the watershed receives rain, but also because warmer storms carry greater water vapor and thus can produce heavier precipitation. The uncertainty of flood forecasts was shown to non-linearly with the rain-snow level for these reasons as well, highlighting the importance of improving forecast accuracy for flood risk management.

To aid water resources managers and the public, CW3E provides visualizations of current operational rain-snow level forecasts and their uncertainty for the U.S. West Coast.

The effect of rain-snow level forecast uncertainty on flood forecasting: a) schematic showing how the rain-snow level controls the fraction of precipitation received as rain in mountain watershed.. Bottom panels shows how rain-snow level forecast errors contributes most to flood risk uncertainty for high rain-snow level (warm) AR events, using the Merced River watershed above New ExchequerDam as an example: b) rain-snow level forecast error ranges as a function of observed rain-snow levels, c) rain-snow fraction ranges, d) precipitation rate ranges, and e) watershed rain rate ranges. Forecast uncertainty produces the largest watershed rain rate uncertainty for the warmest ARs.

Henn, B., R. Weihs, A.C. Martin, F.M. Ralph, and T. Osborne (2020): Skill of rain-snow level forecasts for landfalling atmospheric rivers: A multi-model model assessment using California’s network of vertically profiling radars. J. Hydrometeor., 124, 0 https://doi.org/10.1175/JHM-D-18-0212.1