An Analysis of Precipitation Variability, Persistence, and Observational Data Uncertainty in the Western United States

Abstract

This paper presents an intercomparison of precipitation observations for the western United States. Using nine datasets, the authors provide a comparative climatology and season- and location-specific evaluations of precipitation uncertainty for the western United States and for five subregions that have distinct precipitation climates. All data are shown to represent the general climate features but with high bias among datasets. Interannual variability is similar among datasets with respect to the timing of precipitation excesses and deficits, but important differences occur in the spatial distribution of specific anomalous events. Dataset distribution differences, as represented by their cumulative density functions (CDFs), are statistically significant for 80% of data combinations stratified by subregion and season. The CDFs of anomaly fields are more similar but uncertainty remains, as data differences are significant for 40% of dataset comparisons. Observational uncertainty is low for persistence studies because the data are found to be similar with respect to (i) grid cell estimates of a characteristic persistence time scale and (ii) distributions of anomaly length scales. Spatially, the greatest uncertainty in magnitude differences occurs along the Rocky Mountains in winter, spring, and fall, and along the California coastline in summer. In linear (phase) association, the greatest differences occur in northern Mexico during all seasons; along the Rocky Mountains in winter, spring, and fall; and in California, Nevada, and the intermountain region in summer. Overall, data similarity is lowest in summer as a result of a reduction in phase association and an increase in amplitude differences.