BP-CCA方法用于中国冬季温度和降水的可预报性研究和降尺度季节预测

Statistical downscaling based on BP-CCA: Predictability and application to the winter temperature and precipitation in China.

中文关键词:  可预报性， BP-CCA，降尺度，海气耦合模式，季节预测

英文关键词:Predictability, BP-CCA downscaling, CGCM seasonal prediction

中文摘要:

      使用变形的典型相关分析（BP-CCA）方法，基于交叉检验的结果建立了东亚冬季500 hPa大尺度环流和中国冬季温度、降水的最优BP-CCA降尺度预测模型，并在此基础上进行了中国冬季温度和降水的可预报性研究，表明用东亚冬季500 hPa高度场降维后的大尺度环流来解释中国冬季温度，平均距平相关系数（ACC）为0.7左右，最高可达0.9，解释中国冬季降水的平均ACC为0.3左右，最高可达0.7，温度的可预报性远高于降水，且二者的可预报水平存在明显的区域差异。可预报性研究表明东亚冬季500 hPa大尺度环流异常与中国冬季温度、降水异常有密切的联系，BP-CCA方法可以很好地揭示大尺度环流与温度、降水的内在联系，并且物理意义清晰。在东亚大尺度环流系统中，东亚大槽和西太平洋副热带高压是影响中国冬季温度、降水异常的重要系统。进一步利用国家气候中心海气耦合模式（CGCM/NCC）回报和预测的500 hPa环流场和BP-CCA方法对温度和降水进行降尺度预测应用，对温度和降水的预测效果明显高于模式直接输出的结果，而且对温度预测的改善高于对降水的改善。对模式预测的环流进行EOF(经验正交函数)分析，表明BP-CCA方法对降尺度要素预报的可预报性来源于CGCM/NCC对500 hPa高度场主要大尺度特征的模拟能力较好。

英文摘要:

      The BP-CCA is used to build the optimal statistical downscaling model between the large- scale circulation over East Asia and the temperature/precipitation over China in boreal winter based on one year out cross validation. The predictability of temperature/precipitation in winter is analyzed. It shows that the average anomaly correlation coefficient (ACC) between the 500 hPa circulation and the temperature in winter is 0.7 with the highest ACC of 0.9. And the average ACC between the 500 hPa circulation and the precipitation in winter is 0.3 with the highest ACC being 0.7. Though the predictability of temperature/precipitation over China shows regional features, the 500 hPa potential heights still have close relationship with the temperature/precipitation in winter. The East Asia trough and western Pacific subtropical high are the most important systems which affect the temperature /precipitation in winter. Combining the BP-CCA method with the hindcast results of CGCM/NCC, we can obtain higher skill of forecasting temperature/precipitation via downscaling temperature/precipitation by means of the BP-CCA downs caling model than that in case only the CGCM/NCC model is used. The ability of the downscaling model comes from the high forecast skill of large scale circulations by the CGCM/NCC. 