Comparison of gridded precipitation datasets used in climate and drought monitoring in the contiguous United States

Abstract

This study examines several gridded datasets that are commonly, or could be, used in climate and drought monitoring in the contiguous United States (CONUS). The focus is on accumulations at durations from months to water years, which are most relevant to drought, and on the western CONUS where complex terrain complicates precipitation estimation. When compared to gauge observations, the PRISM dataset matches closely. Consistent biases from year to year are found in several of the datasets; for example, large overestimates in the Stage IV product in the Missouri River Basin and underestimates in mountainous areas for the products that rely heavily on radar data. ERA5 has insufficient resolution and does not necessarily capture observed events; several datasets do not represent the mountains well. These differences can create important discrepancies when calculating commonly used drought indicators like standardized precipitation index. The findings highlight the importance of knowing the strengths and limitations of different precipitation datasets, and for climate and drought monitoring, it is strongly recommended that the dataset being used be compared against historical reference baselines from the same dataset.