MJO预报研究进展

热带大气季节内振荡 (Madden-Julian oscillation，MJO) 是次季节-季节时间尺度气候变率的支配模态。它不仅对低纬度地区天气气候产生重要影响，还能够通过经向传播和激发大气遥相关波列对中高纬度地区产生影响，是延伸期尺度最重要的可预报性来源。因此，MJO预报是次季节-季节气候预测中极为重要的部分，近年来受到国际学术界广泛关注。该文回顾了MJO预报发展历史，概述了当前国际上主要科研业务机构的MJO预报发展现状。目前基于统计方法和气候模式的MJO预报研究取得了较大进展，特别是多个耦合气候模式和一种基于时空投影方法的统计模型均能够显著提升MJO预报技巧 (有效预报可达20 d以上)。该文还介绍了中国气象局国家气候中心在MJO预报技术发展和业务系统研制方面的新进展，当前基于第2代大气环流模式的MJO业务预报填补了国内空白，技巧为16~17 d，而耦合气候模式试验的技巧已达到约20 d。总体来看，利用耦合模式预报MJO是未来发展的主要方向，其中，面向MJO的模式初始化和集合预报新方法研究将是关注重点。

Progresses of MJO Prediction Researches and Developments

Madden Julian oscillation (MJO) is the dominant mode of the sub seasonal to seasonal (S2S) time scale variability. It has great impacts on weather and climate events at low latitudes, and also influences the circulation at mid high latitudes by its meridional propagation and stimulating teleconnection wave trains, which presents a primary source of predictability on extended range time scale. Therefore, the MJO prediction, which is the crucial part of S2S climate prediction, has been paid much attention in recent years. Firstly, the history of MJO prediction is reviewed, and then the current status of MJO prediction in main international research and operation institutions is summarized. Furthermore, the latest progress of the MJO prediction technique development and operation system establishment in National Climate Center of China Meteorological Administration (NCC/CMA) is focused on. The development goal and research plan for the MJO prediction on next step in NCC/CMA are prospected finally.Basic methodologies for the MJO prediction include the statistical and dynamical models. In recent years, big progresses have been made for two methodologies. For the former, the so called spatial temporal projection method (STPM) can extend the valid length of the MJO prediction to 25-30 days in terms of pentad mean. For the latter, atmospheric and coupled general circulation models (GCMs) have significantly pushed skills of the MJO prediction forward. Until now, the major research and operational institutions in the world have the valid prediction length longer than 20 days generally. More and more evidences indicate that the dynamical prediction based on GCMs is the most promising direction of MJO prediction. Indeed, it plays the crucial role in the model prediction that the GCMs need more adequate initial values including strong MJO signals, well designed initialization schemes making initial values dynamically consistent with model, and effectively perturbed ensemble members in terms of uncertainties from both initial values and model physics.n NCC/CMA, its own MJO prediction system using both the statistical and model is developed. On one hand, the STPM is now applied to the quasi operational MJO prediction and the countrywide extended range pentad mean prediction of China station precipitation. On the other hand, methods and techniques of using BCC_AGCM2.2 and BCC_CSM1.1(m) models are being developed quickly in NCC/CMA, where particularly, a real time MJO monitoring prediction system based on BCC_AGCM2.0 is established, fed by observations which all from CMA. By this system, daily real time monitoring and prediction operational products are issued for forecaster use. Recently, a new ensemble method has been put forwards by averaging prediction results of BCC_CSM 1.1(m) with three different initialization schemes, which can significantly improve the MJO prediction and make the valid length reach about 20 days. Further studies are necessary for the MJO prediction, especially for that using the coupled GCMs.