GPM Ground Validation DSD Variability as Revealed from Empirical Orthogonal Function Analysis

Now that five major NASA ground validation (GV) field campaigns (LPVEx, MC3E, IFloodS, IPHEx, and OLYMPEx) have been completed, we have the opportunity to examine variability in drop-size distributions (DSD) across the different campaign regions. Understanding the variability of DSDs has important implications for remote sensing retrievals of rainfall and precipitation, from both ground and satellite-based radars. Analysis of the extent and frequency of DSD parameters in each region is intended to provide possible constraints on satellite-based retrievals of microphysics and rain. The five experiments represent a cross-section of meteorological regimes, including light precipitation, orographic, and deep, organized mid-latitude systems. Herein we apply empirical orthogonal function (EOF) analysis to global disdrometer datasets to understand the main modes of DSD variability, and put the GV field observations in a larger context. The first mode of variability, associated with convective and stratiform rain, accounts for 53% of the variability in DSD parameters in all regions. The second mode of variability accounts for 25% of global DSD variability, and is consistent with DSDs associated with warm rain vs. ice-based convection. These different DSD `regimes" have important implications for remote-sensing rainfall retrievals, including ground- and satellite-based rainfall estimation. Specifically, they occupy different regions of reflectivity (Zh)-rainfall space, indicating the need for different Z-R relationships in each regime. The projection of the DSD in Nw and D0 space provides one way of examining variability. In fact, as has been suggested in previous studies, one can devise a separation line that divides convective rain (C) from stratiform (S) rain for different regions (e.g. Bringi et al. 2009, Thompson et al. 2015). Through our EOF analysis, we find that a global C-S separation line can be formed by uniting these two lines. Additionally, we can infer warm-rain vs. ice-based convection from Nw-D0 pairs. Using these groupings, we calculate the contributions from warm rain, ice-based convection and stratiform to total rain volume, and examine the contributions to rainfall in the GV campaigns compared to fractions obtained from the global dataset.