一种基于全球动力模式和SMART原理结合的统计降尺度区域季节气候预测方法

针对江苏夏季旱涝和高温热浪等异常气候的预测难题，以江苏夏季站点降水和气温为预测目标，建立了一种基于全球动力模式BCC_CSM1.1（m）和最优可预测气候模态和异常相对倾向（SMART）原理结合的统计降尺度季节气候预测方法。利用历史观测资料和SVD方法提取决定中国夏季降水异常相对倾向的同期热带地区向外长波辐射（Outgoing Longwave Radiation，OLR）和北半球中高纬500 hPa位势高度场异常相对倾向的最优可预测气候模态，并利用逐步回归法构建其与同期江苏站点降水和气温异常相对倾向同期关系的统计降尺度模型；将动力模式对最优可预测气候模态的预测带入统计降尺度模型，实现对区域降水和气温异常相对倾向的预测；最后通过引入观测的近期背景异常实现对降尺度的降水和气温总距平的预测。通过对1991—2019年江苏夏季降水和气温的回报检验表明，本文建立的统计降尺度模型效果较BCC_CSM1.1（m）动力模式的直接预测效果有显著提高，为区域精细化季节气候预测提供了一种有效的手段。

A statistical downscaling method for regional seasonal climate prediction with combined global dynamical model and SMART principle

To solve the problems of abnormal climate prediction such as summertime drought, flood and heat wave in Jiangsu, in this paper, a statistical downscaling method for regional seasonal climate prediction is developed with combined global dynamical model BCC_CSM1.1(m) and selected predictable climate modes (SM) and anomalous relative tendency (ART), i.e., SMART-based principle for summertime rainfall and air temperature. In this method, the principal climate modes in the ARTs of summertime tropical OLR and mid and high latitude 500 hPa geopotential height in the Northern Hemisphere that determine the simultaneous national-wide ART of rainfall in China are firstly extracted with historical observed data and SVD analysis, and a statistical downscaling model for the simultaneous relationship between those modes and ARTs of rainfall and air temperature in Jiangsu Province is constructed with stepwise regression method. Then, with the statistical downscaling model, the regional ARTs of rainfall and temperature are predicted with predicted SMs by the global dynamical model. Finally, the total anomalies of rainfall and air temperature are obtained by adding the recent background anomalies to the predicted ARTs. The hindcasts for 1991-2019 show that the method proposed in this study significantly improves the regional predictions as compared with those by dynamical model itself. Thus, this study provides an effective tool for fined regional seasonal climate prediction.