Evaluation of multi-model precipitation forecast performance for Typhoon Khanun(2306)with different verification methods
YANG Ruiwen
QU Rongqiang
LI Yang
TAN Zhenghua
YU Fanyue
Abstract：Taking the heavy precipitation caused by the long-duration Typhoon Khanun(2306)affecting Liaoning and its surrounding regions as an example,this study uses the TS(threat score)verification method and various spatial verification methods to examine and analyze the deviation characteristics of the 24-h,48-h and 72-h daily precipitation forecasts from the CMA_MESO,CMA_GFS and EC_IFS models in three aspects:intensity,distribution pattern and location.(1)The TS verification method shows that within the 24-h forecast period,the forecast scores of the CMA_MESO,CMA_GFS and EC_IFS models for precipitation magnitudes below heavy rain are 0.64,0.66 and 0.69,respectively.As the precipitation threshold increases,the TS decreases to around 0.40.Among them,EC_IFS achieves the best TS,while CMA_MESO has the highest false alarm rate.(2)The SAL(structure,amplitude and location)spatial verification method indicates that the overall precipitation location forecasts of the three numerical models are basically consistent,but the precipitation intensity forecasts are all weaker.The forecasted precipitaion area structure from EC_IFS is the most similar to the actual situation,however,its predicted range for precipitation of rainstorm and above magnitudes is larger.(3)The comprehensive similarity score of MODE(method for object based diagnostic evaluation)indicates that CMA_GFS is unstable in forecasting the precipitation associated with this typhoon;CMA_MESO shows stable performance,but its forecasting effect for rainstorms is not ideal;EC_IFS has the highest similarity score for 24-h forecast and demonstrates the best reliability in forecasting the precipitation of this typhoon.Compared with the traditional TS,the spatial verification method can more accurately identify the models' structural deviations in terms of rainstorm magnitude and quantify the positional offset of rainbands.In practical operational work,different verification methods should be selected according to different needs.At the same time,different verification methods can also be combined to analyze the performance of numerical models from different perspectives,which is helpful to improve the model applicability for precipitation forecasts of similar typhoons.
Keywords：Typhoon Khanundeviationthreat score(TS)verificationspatial verification
Publication Date：2026-02-28
Online Publishing Date：2026-05-22(First online date of this platform, not the publication date of the document)
Pages：11( 75-85 )