Study on Initialization Method for Regional Ensemble Forecast Based on Dynamical Downsacling

Using Global Ensemble Forecast data of GEFS, a regional ensemble forecast system based on WRF model is constructed, two initialization strategy are tested, one is direct dynamical downscaling of initial states of GEFS(namely DOWN ensemble), the other is extract the downscaled initial perturbations of GEFS and combine with the analysis of regional High-resolution numerical weather prediction system, to form the initial state of regional ensemble(since the High-resolution NWP system is the Rapid Update Cycle system of Beijing, namely BJ-RUC system, we call this initialization strategy “Down-RUC” ensemble). Using the two methods of Down and Down-RUC, a series of ensemble forecast tests are conducted and compared.

 Results indicate that the small scale components of Down-RUC perturbations grow more rapid than those of DOWN perturbations. The DOWN perturbations in short term forecast tend to underestimate the forecast error while the Down-RUC perturbation tend to better represent the forecast error, identify where the forecast error is large from where the forecast error is small; ensemble forecast verification shows that Down-RUC ensemble have larger spread and smaller root mean square error than DOWN ensemble at short forecast lead time, while the probabilistic scores of Down-RUC are also outperform that of DOWN for short term forecast. Typical precipitation case study shows that Down-RUC ensemble can provide better probabilistic precipitation forecast than DOWN ensemble.