利用NCEP的第二代气候预测系统（CFSv2）提供的2000 -2009年降水场历史回报试验资料以及川渝182个测站的降水实况资料。采用时间相关系数、 均方根误差、 距平相关系数、 距平符号一致率以及PS评分等方法， 对模式在川渝地区夏季降水以及夏季降水异常的次季节尺度预测技巧进行检验， 并进一步分析了模式在概率密度和降水频次方面的预报偏差特征。结果表明： 该模式对川渝夏季降水的可用预报时效为3候左右， 能够较好地模拟出夏季降水的高值中心， 但量级偏大。预报技巧高值区主要位于四川盆地西北部及渝东北地区， 对攀西地区南部及川西高原部分地区也有一定的预报技巧。该模式也能够较好地把握川渝地区夏季降水异常偏少的趋势， 有效预报技巧为2候以内。模式各时效预报与观测的降水概率密度主要集中在10 mm以下量级； 模式预报各量级降水频次与实况相比均偏高得较为明显， 且随着预报时效延长， 偏差越大， 其中偏高最为明显的是小雨频次。

In this paper， the hindcast precipitation fields of the NCEP's second-generation climate prediction system （CFSv2） and the observed precipitation data of 182 meteorological stations in Sichuan and Chongqing from 2000 to 2009 were utilized.Sub-seasonal forecasting ability for summer precipitation and its anomaly in Sichuan-Chongqing region of this model was evaluated by use of Temporal Correlation Coefficient （TCC）， Anomaly Correlation Coefficient （ACC）， Root Mean Square Error （RMSE）， sign coincidence rate （SCR） and PS scores methods.Meanwhile， bias characteristics on probability density and frequency of precipitation was analyzed.The results show that the available forecast lead time for summer precipitation in Sichuan and Chongqing is about 3 pentads， which can simulate the high-value center of summer precipitation well， but the magnitude is too large.The high-value areas of forecasting skills are mainly located in the northwestern Sichuan Basin and northeastern Chongqing， and some forecasting skills are also available for the southern Panxi region and parts of the western Sichuan Plateau.The model can also better grasp the trend of abnormally less precipitation in summer in the Sichuan-Chongqing region， and the effective forecasting skill is within 2 pentads.The precipitation probability densities predicted and observed in each time period of the model are mainly concentrated in the order of magnitude below 10 mm； compared with the actual situation， the precipitation frequency of each magnitude predicted by the model is significantly higher， and with the extension of the forecast time， the deviation is greater， and the most obvious one is the light rain frequency.