Modeling of poleward-moving tropical cyclones in the northern East China Sea: Intercomparison of WRF physical schemes

Abstract

Reliable tropical cyclone (TC) modeling and forecasting requires appropriate selection of physical parameterization schemes. This study focusses on the poleward-moving TCs in the northern East China Sea. Numerical experiments are carried out to examine the sensitivity of modeled TC parameters to different combinations of microphysics (MP), planetary boundary layer (PBL), and cumulus convection (CC) schemes. Model performance is evaluated against observations from China Meteorological Administration best-track dataset, satellite retrieved wind fields, and land-based meteorological stations. A composite evaluation metric is introduced to enable cross-variable model assessment and quantify the relative performance of each scheme combination. Results indicate that while MP schemes have a moderate impact on TC's minimum sea level pressure and maximum wind speed, they evidently influence surface wind field reproduction. Model performance across the PBL schemes is generally comparable, except for the University of Washington boundary layer scheme, which demonstrates strong forecasting skill for maximum wind speed but introduces substantial discrepancies in TC track, minimum sea level pressure, and surface wind fields. Moreover, model performance is highly sensitive to the choice of CC scheme. The Tiedtke scheme consistently yields the highest performance score across all datasets, particularly effective in reducing wind field discrepancies. This study underscores the necessity of considering multiple TC parameters and addressing spatial discrepancies in model calibration and performance evaluation. The findings provide insights into optimizing physical scheme combinations for TC modeling in the northern East China Sea, with broader implications for improving extreme weather forecasting. However, their applicability to other regions and TC types remains uncertain. Further investigation is needed to elucidate the physical mechanisms driving scheme-related differences in model performance.