Abstract: In this study,the SAL(structure amplitude and location) spatial verification method is used to evaluate the 12~36 hours total precipitation forecast performance of five China Meteorological Administration (CMA) models for the strongest precipitation day during the remnants of Typhoon "Meihua" over Northeast China in September of 2022 (from 08:00 on September 16 to 08:00 on September 17,with the typhoon category as a tropical depression).The results can be summarized as follows: The CMA-GFS model exhibits the best forecast of precipitation structure,amplitude,and location in this case,while for the other models tend to underestimate extreme precipitation,except CMA-BJ.Due to its well-performance in predicting the location,intensity,and moving speed of the 850 hPa low-level jet,the CMA-GFS model has the best precipitation forecast.Meanwhile,the faster speed of the low-level jet predicted by the CMA-TYM model leads to a smaller area of significant rainfall exceeding 100 mm.The CMA-GFS model demonstrates better forecasting performance as the lead time shortens,with notable advantages in now-casting forecasts but limited skill for longer lead times.In contrast,although CMA-TYM model exhibits poorer structure and amplitude performance in approaching lead time,it is the earliest model to provide indicative signals for the general location and magnitude of heavy precipitation.The summary of CMA models forecast performance in typhoon remnant system precipitation aims to enhance the applicability of numerical weather models in China in similar scenarios of the future.

Key words: CMA models, SAL (structure amplitude and location) spatial verification, Typhoon remnant precipitation forecast, Bias analysis