UCLA

There is a growing need for high-resolution, spatially complete meteorological data. These data are utilized within weather, climate, ecological, and environmental research. With so many different gridded datasets available, it is unclear which is best for a given application. There are few comprehensive studies that have examined the strengths and weaknesses of gridded datasets, and they have mainly identified flaws or differences without understanding how these differences arise from the methodologies or suggesting ways to improve. Precipitation extremes are especially difficult to capture in gridded data. Here we assess precipitation extremes of five high-resolution gridded datasets over California to by comparing with daily station data and interpret the results in the context of each dataset’s methodology. Multiple statistics of extreme precipitation were considered, reflecting intensity, frequency, and duration. Large differences are found both between the gridded datasets and relative to the station data. Maximum single day precipitation is underestimated in nearly all datasets, and precipitation frequency is severely overestimated in many cases. Datasets differ most notably in magnitude and precipitation occurrence in mountainous and coastal regions. The errors of these gridded datasets can vary significantly when compared to station data; one dataset within this study gives a 54% error for consecutive precipitation frequency. The results of this assessment are likely to be useful for users of gridded datasets looking to select a dataset appropriate for their research. They could also aid gridded dataset creators in improving existing products or building new ones.