The Effects of Increased Accretion Efficiency on the Surface Precipitation Simulation of Air–Sea Interaction Events over the Korean Peninsula

This study presents a comprehensive evaluation of the recently developed Weather Research and Forecasting (WRF) double-moment 6-class (WDM6) microphysics parameterization scheme with prognostic small ice crystal number concentration (WDM6_NI), using three wintertime air–sea interaction cases over the northeastern Korean Peninsula. Compared to the default WDM6, WDM6_NI produces reduced surface precipitation, which is inconsistent with automatic weather station (AWS) observations. Sensitivity experiments (hereafter EFF), in which the accretion efficiency is set to 1 without the reduction factor, confirm that the reduced surface precipitation in WDM6_NI results from its decreased accretion efficiency, causing a reduction in mixing ratios of solid-phase hydrometeors in the atmosphere. EFF increases the mixing ratios of solid-phase hydrometeors and results in greater surface snow and graupel along the coast, improving agreement with precipitation-type classifications from radar/AWS observations. Additionally, EFF enhances the probability of detection and pattern correlation scores for surface precipitation, demonstrating improved predictive performance for wintertime air–sea interaction events over the Korean Peninsula. These results support improving the operational configuration of WDM6 in numerical weather prediction (NWP). This study also compares implicit and explicit approaches for classifying surface precipitation types using EFF simulations. Two implicit methods-the revised Matsuo (RM) and the 1000–700-hPa thickness (TK700)-are assessed against precipitation types from radar/AWS observations. The explicit method, which classifies precipitation types based on simulated surface hydrometeors from the microphysics scheme, outperforms TK700 and yields performance comparable to RM. These findings suggest that, when used with refined microphysics schemes, the explicit method may provide more accurate precipitation-type classifications than commonly used implicit methods.