Forecast accuracy and physics sensitivity in high-resolution simulations of precipitation events in summer 2022 by the Korean Integrated Model

Abstract

The precipitation prediction of the Korean Integrated Model (KIM) is evaluated over South Korea for the summer season of July–August 2022, and key factors for accurate predictions are examined using various approaches, including case studies under distinct synoptic patterns and physics sensitivity experiments. In this study, a five-day prediction experiment was conducted using the latest version of KIM in a near real-time full cycle configuration with 8-km grid spacing, while additional case simulations and prediction tests were conducted on low-resolution or cold-run testbeds. For verification, a newly designed synoptic pattern verification was introduced to assist to the conventional dichotomous verification for daily precipitation. It was found that heavy rainfall events over South Korea are determined by two dominant patterns: frontal and cyclonic. KIM can successfully discriminate between synoptic patterns with a detection rate of approximately 85% for these two types within a short-range prediction. However, it is evident that the precise prediction of precipitation requires an accurate location of the precipitation system within a specified timeframe, wherein KIM shows weakness in delaying the movement of extratropical cyclones with forecast lead times. The significance of moist physics is also highlighted by sensitivity experiments that control convective trigger conditions. This demonstrates that large-scale precipitation from a microphysics scheme must be enhanced to properly represent the strong development of inland rain systems over South Korea, which are highly sensitive to convective precipitation activity in the numerical model, especially in upwind ocean regions.