Study of advanced data assimilation and cloud resolving ensemble technique

为了克服在没有强天气和地形强迫的情况下预报局地对流雨的困难，开展了中尺度资料同化和集合预报的研究和开发工作，开发了风暴尺度4DVAR资料同化方案、JMA非静力模式的局地集合变换卡尔曼滤波方案、集合变分资料同化方案，并将其应用于局地暴雨和台风资料同化试验。对中尺度观测资料(如GPS资料和多普勒激光雷达资料)进行了中尺度同化试验。发展了一个中尺度集合预报系统的初始和侧边界摄动方法，并将该系统应用于一个云分辨模式。通过中尺度集合预报的验证，验证了局地暴雨概率预报的可行性。

To overcome the difficulties to predict local convective rains without strong synoptic or orographic forcing, research and development on mesoscale data assimilation and ensemble forecast were conducted.A storm scale 4DVAR data assimilation scheme, local ensemble transform Kalman filter for the JMA nonhydrostatic model, an ensemble variational data assimilation scheme were developed and applied to data assimilation experiments for local heavy rainfall events and typhoons. Mesoscale assimilation experiments for mesoscale observation data (e.g., GPS data and Doppler lidar data) were conducted. Initial and lateral boundary perturbation methods for a mesosocale ensemble prediction system were developed and the system was applied to a cloud resolving model. Feasibility of the probabilistic forecast of local heavy rainfall was shown through validation of mesoscale ensemble forecasts.