Evaluating Hydrologic Model Forcings for Use in Reservoir Operations Planning

A current mandate requires all California State Water Project reservoirs to have 10 day
operations plans. This mandate includes three small (< 200 km² ), ungauged reservoirs in the
Upper Feather River Basin: Frenchman Lake, Lake Davis, and Antelope Lake. Hydrologic
forecasts for these basins are produced by the California Department of Water Resources
(DWR) using the HEC-HMS hydrologic model forced by the current operational medium-range
weather forecast model (the GFS). At the relatively coarse 13-km grid spacing of the GFS,
these small basins are represented by only 1 or 2 pixels. The current operational short-range
forecast model, the High Resolution Rapid Refresh (HRRR) is available at 3-km grid spacing,
which would increase the density of meteorological forecast forcing gridpoints for the basins of
interest to 12-24 pixels each and would likely provide improved representations of precipitation
in the complex terrain of the upper Feather River Basin. In this study, we evaluate the use of
both the operational deterministic HRRR, which provides 18h forecasts every hour and 36-h
forecasts four times daily, and the experimental 9-member HRRR ensemble that produces 36h
forecasts daily at 00 UTC as forcing inputs to the HEC-HMS hydrologic model. We assess
whether higher resolution precipitation forecasts provide improved forcing data, and therefore
better forecasts of reservoir behavior compared to those forced by the GFS. The level of
uncertainty in reservoir behavior given by the use of the ensemble HRRR will also be examined.
Potential applications to larger basins in regions of similarly complex terrain will also be
discussed.