中国业务动力季节预报的进展

ADVANCE IN SEASONAL DYNAMICAL PREDICTION OPERATION IN CHINA

利用动力模式开展季节到年际的短期气候预测 ,是目前国际上气候预测的发展方向。自 1996年以来 ,经过 8a多的研制和发展 ,国家气候中心已建立起第 1代动力气候模式预测业务系统 ,其中包括 1个全球大气 海洋耦合模式 (CGCM )、1个高分辨率东亚区域气候模式 (RegCM_NCC)和 5个简化的ENSO预测模式 (SAOMS) ,可用于季节—年际时间尺度的全球气候预测 ;全球海气耦合模式与区域气候模式嵌套 ,可以提供高分辨率的东亚区域气候模式制做季节预测。CGCM对 1982～ 2 0 0 0年夏季的历史回报试验表明 ,该模式对热带太平洋海表面温度和东亚区域的季节预测具有较好的预测能力。RegCM NCC的 5a模拟基本上能再现东亚地区主要雨带的季节进展。利用嵌套的区域气候模式RegCM NCC对 1991～ 2 0 0 0年的夏季回报表明 ,在预报主要季节雨带方面有一定技巧。 2 0 0 1～ 2 0 0 3年 ,CGCM和RegCM NCC的实时季节预报与观测相比基本合理。特别是 ,模式成功地预报了 2 0 0 3年梅雨季节长江和黄河之间比常年偏多的降水。SAOMS模式系统的回报试验表明 ,该系统对热带太平洋海表面温度距平有一定的预报能力 ,模式超前 6～ 12个月的回报与观测的相关系数明显高于持续预报。 1997～ 2 0 0 3年 。

Application of dynamical model in seasonal-interannual climate prediction is currently a developing direction of climate prediction in the world. After more than 8 years' research and development, the first generation of dynamical climate model prediction operation system has been established in National Climate Center (NCC), Chinese Meteorological Administration. It consists of a global coupled atmospheric-oceanic circulation model (CGCM), a high-resolution East Asian regional climate model (noted as RegCM_NCC) and five simplified ENSO prediction models (noted as SAOMS), which can be used to perform climate predictions on seasonal-interannual time scales. RegCM_NCC nested with CGCM can provide high-resolution seasonal prediction of climate over East Asia. The hindcast experiment results of the CGCM for the summer of 1982-2000 show that the model has reasonably good performance in predicting the tropical Pacific sea surface temperature (SST) and the climate over East Asia on seasonal time scale. By nesting with CGCM, 10-year hindcast experiments of RegCM_NCC have been undertaken for the summer of 1991-2000. The preliminary results have shown that the RegCM_NCC has some capabilities in predicting the major seasonal rain belts over China. The real-time experimental predictions for summer in 2001-2003 with CGCM as well as the nested regional climate model have shown certain prediction capabilities. In particular, the prediction of 2003 summer was successfully made with the much above-normal precipitation in the regions between the Yellow River and the Yangtze River valleys. Hindcast experiments of SAOMS shows it has a certain skill to predict the SST anomalies over the tropical Pacific Ocean, with the correlation coefficients considerably exceeding the persistence prediction at 6-12 lead months. The real-time predictions of the tropical Pacific SST anomalies were conducted based on the five models of the SAOMS during 1997-2003. The correlation coefficient between the multi-model ensemble predictions and observations for the SST anomalies in Nino3 region is 0.7. The model system predicted both El Nio (1997/1998) and La Nina (1998/2000) events reasonably.