Multimodel Forecast Evaluation of the Track and Intensity of Supertyphoon Doksuri in 2023

Abstract Using ERA5 reanalysis and numerical forecast productions from the China Meteorological Administration Typhoon Model (CMA-TYM), CMA Global Forecast System (CMA-GFS), ECMWF High-Resolution Forecast (ECMWF-HRES), and NCEP-GFS, this study investigated the performance of multiple models on the track and intensity forecasts of Doksuri. The results indicate that ECMWF-HRES has the highest forecast skill in the track forecast, followed by CMA-GFS. Meanwhile, CMA-TYM and NCEP-GFS have significant errors in longer forecast lead times with high similarity. Except for ECMWF-HRES, the track forecasts from other models have large errors in the forecast of the early stages of typhoon generation and development, which is mainly due to the forecasts of the eastward receding trend of the western Pacific subtropical high at 500 hPa being faster, and steering flows are more northward compared with the real situation. Thus, the forecasts of the moving direction are biased to the right side of the real-time track. In the landfall position forecast, there are large uncertainties in the initial forecast, and that of CMA-TYM tends to be stable after 24 July. As for the landing time, CMA-GFS exhibits a high degree of reliability, while other models forecast a landfall earlier than 0200 UTC 28 July in terms of fast-moving speed in the initial forecast. In intensity forecast, CMA-TYM has the best forecast capability and also indicates a high indication of rapid intensification and its peak intensity. The other global models are significantly weaker in the forecast of rapid intensification and have limited ability to forecast the peak intensity of Doksuri. Significance Statement Learning about model forecast errors and their causes is important for a deeper understanding of the performance of the numerical weather prediction model for tropical cyclone forecasting and provides clues for model improvement. In this work, we focus on the forecast performance of international major operational numerical weather prediction systems for Supertyphoon Doksuri (2305). It is found that the track forecast from CMA-GFS is comparable to ECMWF-HRES, and the forecasts from the two models are highly referable; in addition, CMA-GFS has the best forecast skill in the landfall time of Doksuri. The capability of the regional model to forecast typhoon intensity, i.e., CMA-TYM, is significantly better than that of the global model, for both rapid intensification and its peak intensity.