61A Performance of Convection-Allowing WRF Simulations of Warm-Sector Heavy Rainfalls in South China

Wednesday, 26 July 2017
Kona Coast Ballroom (Crowne Plaza San Diego)
Murong Zhang, Peking University, Beijing, China; and Z. Meng and N. Wu

Warm-sector heavy rainfall in South China has long been a forecast headache. Forecasters usually have no clue in forecasting either its initiation, intensity or duration, likely due to the very limited capability of numerical models. Warm-sector heavy rainfall in South China refers to the heavy rainfall that occurs near the southeast coastlines from May to June within the weakly forced synoptic environment far from the front system or in the uniform southwesterly without fronts.

This work aims at statistically and quantitatively examining the capability of convection-allowing numerical simulations in predicting warm-sector heavy rainfall based on 20 typical warm-sector heavy rainfall cases in South China from 2004 to 2015. Up to now, the low operational forecast skills on warm-sector heavy rainfall in South China have been just qualitatively attributed to the lack of upstream observation over the ocean and the limited capability of representing convective-scale forcing in planetary boundary layer such as land/sea breeze and low-level jets (LLJ) in numerical model. Based on the numerical simulations of these 20 cases, we will try to pin on key reasons for the poor capability of numerical models in capturing the warm-sector rainfalls in South China in more details.

Our preliminary results show that WRF simulations of 4.5-km horizontal grid spacing can well capture the accumulated rainfall distributions and evolution in about 50% of the cases. The simulated MCSs tend to be more widespread, poorer organized and located more to the north compared with observation. The initiation of the heavy rainfall can be barely captured correctly in neither timing nor location. By classifying the 20 cases into two groups based on whether there was a LLJ or not, the forecast skills of precipitation and their behind reasons will be examined for the two different scenarios respectively. The impact of digesting more observations into the initial field on the performance of the model will be also explored.

- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner