用复杂气候模式研究ENSO预测的春季预报障碍及其目标观测敏感区

The rapid growth of initial errors is a major cause of the “spring predictability barrier” (SPB) of ENSO prediction. Within the Zebiak-Cane model, previous studies have revealed the initial errors that are more likely to induce the SPB and identified the sensitive areas of targeted observation associated with ENSO prediction. However, the Zebiak-Cane model is an anomaly coupled model of intermediate-complexity and cannot be used to explore the SPB and the sensitive areas of targeted observation for La Niña events. In view of this, using the Community Earth System Model (CESM), this project will firstly identify the initial errors inducing the SPB for ENSO events (including both El Niño and La Niña events) most probably from the perspective of initial error growth. By tracing the season-dependent evolutions of these initial errors, the dynamical mechanism responsible for error growth will be investigated. The difference between the results obtained by the CESM and that in the Zebiak-Cane model will also be discussed. Based on the spatial characteristics of the initial errors inducing the SPB, the project will further identify the sensitive areas of targeted observation associated with ENSO predictions and the role of sensitive areas in improving prediction skill will be examined. In addition, the Ensemble Optimal Interpolation will be adopted to conduct observation system experiment for realistic ENSO events, in order to investigate the role of sensitive areas in improving realistic ENSO predictions. These studies may be helpful for the implementation of the international program of “Tropical Pacific Observing System” (TPOS-2020) in optimizing the tropical Pacific observation network.

初始误差增长是ENSO预测发生春季预报障碍的主要原因之一。前人用Zebiak-Cane模式揭示了易导致El Niño春季预报障碍的初始误差及其初值敏感区。考虑到Zebiak-Cane模式较为简单，未能用于考察La Niña的预报障碍及目标观测问题，本项目拟采用复杂海气耦合模式（CESM），首先，从理想回报试验出发，针对模式中的ENSO事件（包括El Niño和La Niña），从初始误差增长角度寻找导致春季障碍的初始误差，揭示误差增长机制及其与Zebiak-Cane模式结果的异同；分析误差结构识别ENSO初值敏感区，探讨改善敏感区内初始场对提高预报技巧的有效性；其次，从实际预测思路出发，针对真实的ENSO事件进行观测系统试验，对初值采用集合最优插值同化方法，考察上述敏感区内初始场的改善对提高真实ENSO事件预报技巧的有效性，进而为热带太平洋观测系统（TPOS-2020）的优化设计提供指导。

ENSO预测的春季预报障碍及其目标观测敏感区研究对提高ENSO预测技巧具有重要的意义。前人就El Niño的春季预报障碍及其敏感区问题在Zebiak-Cane简单模式中开展了研究。本项目使用复杂的地球系统模式CESM，分别对El Niño和La Niña的春季预报障碍和目标观测敏感区开展了研究，取得的主要研究结果包括：通过大量的理想回报试验，从初始误差增长角度揭示了导致El Niño和La Niña春季障碍的初始海温误差；探讨了初始误差的三维空间结构及其发展特征和主要物理机制，尤其与Zebiak-Cane模式相比，CESM结果强调了赤道西太平洋次表层初始误差对春季预报障碍的重要影响；基于初始误差大值区的局地性特征和预报误差的演变情况，确定了ENSO预测的初值敏感区为赤道东太平洋上层（10°S–10°N，150°W–85°W，0–85m）和赤道西太平洋次表层（10°S–10°N，150°E–135°W，95–165m）两个区域；敏感性试验结果表明，在这两个敏感区内优先减小初始误差、提高初始场精度，能够有效地提高ENSO预报技巧，这一结果为ENSO实际观测实施时加强海洋观测网在西太平洋的布局提供了一个理论依据。