Spatial verification of multi-model forecast of three landing typhoons induced rainstorm in East China in 2018

The Method for Object-based Diagnostic Evaluation (MODE) is applied to evaluate the SMS-WARMS, ECMWF, and GFS models for spatial verification of their performance of rainstorm forecast for three landing typhoons in East China in 2018. The results show that: (1) The MODE method is more advanced than the traditional method in typhoon precipitation verification. It can provide more diagnostic information to reflect the prediction error of the model through various verification indices such as centroid distance, area, and axis angle etc. (2) The results of MODE indicate that the SMS-WARMS prediction is superior to the ECMWF and GFS predictions in terms of spatial predictions of heavy rainfall over 50 mm for the three landing typhoons, particularly for the rainfall over 100 mm. (3) The predicted intensity of rainfall by the regional model is stronger than that by the global models, reflecting the advantages of regional model in predicting typhoon heavy rainfall. (4) The SMS-WARMS model tends to over-predict the area of typhoon heavy rainfall, while ECMWF and GFS tend to under-predict.