Improving quantitative precipitation estimates in mountainous regions by modelling low-level seeder-feeder interactions constrained by Global Precipitation Measurement Dual-frequency Precipitation Radar measurements

A physically-based framework to address the underestimation and missed detection errors in Quantitative Precipitation Estimates (QPE) from Global Precipitation Measurement (GPM) Precipitation Radar (PR) in regions of complex terrain is presented. The framework is demonstrated using GPM Ku-PR because of its wider swath. GPM Ku-PR precipitation estimates are evaluated against ground validation (GV) observations from the long-term ground-based rain-gauge network in the Southern Appalachian Mountains. The detection and estimation errors exhibit a diurnal cycle consistent with the diurnal cycle of low-level clouds and fog (LLCF), thus suggesting the importance of low-level orographic microphysical processes. Contamination of near-surface reflectivity profiles due to ground-clutter is the major source of error in the Ku-PR QPE with spatial features that mirror landform. In particular, GPM Ku-PR drop size distribution (DSD) retrieval algorithms systematically overestimate D_m (mass-weighted mean diameter), and underestimate N_w (normalized DSD intercept) and the precipitation-rate when low-level multilayer clouds and fog are present. Second, column simulations of rainfall dynamics constrained by reflectivity measurements show an emergent relationship in D_m-N_w phase-space that explains an increase in the frequency of D_m < 1 mm in disdrometer observations due to seeder-feeder interactions (SFI) not captured by current retrieval microphysical products. To resolve ambiguity in the detection and characterization of SFI regimes, we demonstrate a physically-based framework to improve GPM Ku-PR orographic QPE that relies on a coupled microphysics-radar rainfall forward model to estimate DSD parameters using initial and boundary conditions from Ku-PR DSD estimates (Method-1), Ku-PR corrected reflectivity measurements (Method-2), and LLCF microphysics from GV observations. Model simulations using Method-2 produce realistic surface DSDs confirming that representation of SFI processes is critical. Application of the framework to GPM overpasses shows potential for robust improvement in QPE and elucidates the physical basis for improved retrievals against ground observations corresponding to which N_w increases by 3–5 dBNw, D_m decreases by approximately 0.03 mm, and rain-rate increases up to ten-fold in the presence of SFI.