Advancing Characterization and Modeling of Space-Time Correlation Structure and Marginal Distribution of Short-Duration Precipitation

Abstract

The statistical characterization of precipitation (P) at short durations (≤ 24 h) is crucial for practical and scientific applications. Here, we advance the knowledge of and ability to model the space-time correlation structure (STCS) and marginal distribution of short-duration P using a network of rain gages in central Arizona with one of the largest densities and spatial coverages in the world. We separately analyze summer and winter P sampled at multiple durations, Δt, from 0.5 to 24 h. We first identify an analytical model and a three-parameter distribution that robustly capture the empirical STCS and marginal distribution of P, respectively, across Δt’s. We then conduct Monte Carlo experiments consisting of multisite stochastic simulations of P time series to explore the spatial and seasonal variability of these properties. Significant seasonal differences emerge, especially at low Δt. Summer (winter) P exhibits weak (strong) correlation structure and heavy- (light-)tailed distributions resulting from short-lived, isolated thunderstorms (widespread, long-lasting frontal systems). The STCS of P is most likely homogeneous and isotropic except for winter at Δt ≥ 3 h, where anisotropy could be introduced via the motion of frontal storms. The spatial variability of the marginal distribution is reproduced by a regional parameterization dependent on elevation in all cases except, again, for winter at Δt ≥ 3 h where additional factors are needed to explain the variability of the mean P intensity. This work provides insights to improve stochastic P models and validate convection-permitting models used to investigate the mechanisms driving changes in short-duration P.