基于卡尔曼滤波的中国区域气温和降水的多模式集成预报

摘要

Based on the data from the TIGGE datasets of European Centre for Medium-Range Weather Forecasts (ECMWF), Japan Meteorological Agency (JMA), National Centers for Environmental Prediction (NCEP), China Meteorological Administration (CMA) and United Kingdom Met Office (UKMO), the Kalman filter method was applied to conduct multimodel ensemble forecasts of the surface air temperature and precipitation.The results showthat the multimodel ensemble forecasts by using Kalman filter are superior to those of the bias-removed ensemble mean (BREM) and other individual models.However, the forecast results of Kalman filter method vary for different meteorological elements and different forecast lead times. For the surface air temperature forecast in China, Kalman filter method shows the best forecast capability while for the precipitation forecast, it has a higher TS score than the BREM.However, with longer forecast lead time, the TS scores for heavy rains are approximately equivalent to those of the best individual model UKMO.So the Kalman filter method does not improve the forecast capability of heavy rains significantly.To sum up, the root mean square error (RMSE) of surface air temperature and precipitation forecasts based on Kalman filter is the smallest among those of the multimodel ensemble forecasts and each individual model forecasts.%利用TIGGE资料集下欧洲中期天气预报中心 (ECMWF) 、日本气象厅 (JMA) 、美国国家环境预报中心 (NCEP) 、中国气象局 (CMA) 和英国气象局 (UKMO) 5个模式预报的结果, 对基于卡尔曼滤波的气温和降水的多模式集成预报进行研究.结果表明, 卡尔曼滤波方法的预报效果优于消除偏差集合平均 (BREM) 和单模式的预报, 但是对于地面气温和降水, 其预报效果也存在一定的差异.在中国区域2 m气温的预报中, 卡尔曼滤波的预报结果最优.而对于24 h累积降水预报, 尽管卡尔曼滤波在所有量级下的TS评分均优于BREM, 但随着预报时效增加, 其在大雨及以上量级的TS评分跟最佳单模式UKMO预报相当, 改进效果不明显.卡尔曼滤波在地面气温和24 h累积降水每个预报时效下的均方根误差均最优, 预报效果更佳且稳定.