An Evaluative Study Of The Capability Of The Multiple Models In The Persistent Heavy Rainfall Forecasting

The persistent heavy rainfall (PHR) over southern China usually leads to serious flooding. The present study is aimed at assessing the predictability capacity of the number of operational forecasting Models, either domestic or outside China (like T639from China, US (GFS), UKMO from UK, CMC from Canada) in the continuous severe precipitation events over different regions of southern China, during11-31July2012and2013. The main findings are as follws:(1) The forecasting skill of atmosphere circulation fields is getting increasingly better, with the ACC values greater than0.6for the models for the lower heights with a lead time of6days, and up to10days for the higher heights. It is found that the forecast skill for the wind field is just4days. All the models have higher predictability capacity for low temperature and specific humidity in their leading time. It has also been revealed that the predictability of the meridional transfer of water vapor is better than zonal transport. The EC model is better in terms of position field.(2) Among these models compared, as far as the filtered components are concerned, the UK model shows a better forcasting skill for the lower troposphere, while T639does better for the upper troposphere. Especially, as for the GFS alone, the filtered3-6w of the wind field show a better result. In the V-fields the long waves and extra-long waves are better than for the U-fields with the leading times able to be prolonged upto5d or so.(3) All the models under study are able to predict more accurately the intensity and position of the subtropical high as well as the tendency of the westward extending of its ridge line. However, the predictive outcomes for the intensity are weaker in whole, while the position outputs are in better agreement with the observations. For the first5d forcasts, the tendency of the blocking high intensity changes can be inlarge reproduced, implies that the forecasting capacity of the blocking high is better than the subtropical high with scores for the Ural blocking high being better than that of Okhotsk blocking high.(4) The models under consideration can forcast the precipitation area ahead by2d. For the specific region assigned, the models show a better forecasting skill with light rain, and the mederate rain and heavy rain subsequently. The TS for torrential rain are lowest. Among the different models, T639shows a better forecasting capability during the whole process of continuous severe precipitation event, especially, for torrential rain forecasts it is better than any other model. The results made a new approach proposed in the paper for verifying the different precipitation events show that the forecasts of precipitation intensity are better than those for precipitation process. The CMC model is relatively better in terms of the precipitation pattern, whereas GFS model is better in terms of the occurrence area of severe precipitation event.