Summary: Precipitation Evaluation
-
Validation of model skill requires attention be placed on the quality and uncertainty contained in observational data sets
-
In regions of complex terrain, models should be verified on their native grids with minimal postprocessing
-
Objective skill measures should be consistent with subjective, visual evaluations. The ETS may be misleading in some instances, since overforecasting precipitation is rewarded while slight underforecasting is penalized.
-
Models that resolve the terrain of the western United States tend to have higher skill during winter for moderate-to-large precipitation totals. Less skill is evident during spring.
-
Skill for coarse resolution models improves if model precipitation is readjusted across grid squares to conform to monthly PRISM climatology
-
Forecast skill is higher along West Coast than over Intermountain region