年代际预测和耦合资料同化研究综述

年代际预测是近年来气候变化研究的一个迅速发展的新兴热点领域，其首要步骤是进行初始化，目的是为年代际预测提供包含观测变率信息的初值.发展效果好且省时的初始化方法是年代际预测的重大挑战之一，目前国际上主流的初始化方法是耦合资料同化，即在耦合模式框架下进行同化.在年代际预测时，由于模式偏差和初始化方法性能的限制会产生初始冲击问题.目前国际上的各模式机构普遍对北大西洋、热带东西太平洋和印度洋海表温度的年代际预测水平高，而对全球平均近地面气温和北太平洋海表温度的年代际预测水平相对较差.本文主要从初始化方法和年代际预测这两方面的研究现状进行全面回顾，指出存在的问题并讨论未来的发展趋势和挑战.

Review on the study of decadal prediction and coupled data assimilation

Decadal prediction has been a rapidly developing new field in the research on climate change these years. The first and foremost procedure in decadal predictions is initialization,which aims to provide initial conditions which include the observed internal climate variabilities for decadal predictions. One of the biggest challenges in decadal predictions is the design of a high-quality and time-saving initialization approach. A commonly-used kind of initialization approach is coupled data assimilation (i.e.,assimilate observations under the framework of the coupled model). During decadal predictions,an initial shock problem often turns up due to model bias and the limitation of initialization method. Currently,the modeling groups around the world achieve high decadal prediction skills of sea surface temperature (SST) in the North Atlantic,tropical eastern and western Pacific and Indian Ocean,but relatively poor decadal prediction skills in forecasting global mean surface air temperature and North Pacific SST. In this paper,we provide a detailed review of the current status of the initialization approaches and decadal predictions. The existing problems and future developments and challenges are also discussed.