面向同化应用的红外高光谱探测资料局地综合通道选择方案及在FY-3D/HIRAS中的初步应用

通道选择是红外高光谱探测资料同化的关键技术。为了最大限度提取红外高光谱探测资料观测信息，减少模式在青藏高原等常规观测稀少地区的初始场的误差，不同区域需要选取不同通道进行同化。基于信号自由度的通道选择方法提出一种面向资料同化的红外高光谱资料的局地综合通道选择方案，该方案综合考虑了局地的大气温度垂直分布特征、背景误差协方差、仪器通道的雅克比函数、权重函数和其他影响红外高光谱模拟和同化的因素。针对CMA_GFS（原GRAPES_GFS）全球背景误差协方差，在高原和海洋两个典型区域对FY-3D/HIRAS红外高光谱资料的温度通道进行局地综合通道选择，并通过一维变分同化评估了局地综合通道选择方案对分析场的影响。结果表明，高原和海洋两个典型区域的大气温度垂直分布特征、背景误差协方差、模式垂直分层以及各通道的雅克比函数和权重函数均有明显的差异，选出的敏感通道也明显不同，相比较在其他区域选择出的通道，在对应地区选择的通道能够显著提高红外高光谱资料的同化效果。 

LOCAL COMPREHENSIVE CHANNEL SELECTION SCHEME FOR INFRARED HIGH-SPECTRAL SOUNDER DATA AND ITS PRELIMINARY APPLICATION IN THE FY-3D/HIRAS

Channel selection is key to infrared high-spectral sounder data assimilation. To fully extract observational data from the infrared high-spectral sounder data and reduce errors of the initial field of the model for areas with few conventional observations, e.g., the Qinghai-Tibet Plateau, we should choose different channels when assimilating different regions. In this paper, a local comprehensive channel selection scheme for infrared high-spectral data was proposed based on the channel selection method of signal freedom. The scheme comprehensively considered the local vertical distribution characteristics of atmospheric temperature, background error covariance, the Jacobian matrix and the weight function of the instrument channel, and other factors that affect infrared high-spectral sounder data assimilation. Based on the global background error covariance of CMA_GFS, local comprehensive channel selection was carried out on the FY-3D/HIRAS data of two typical regions, one in the plateau and the other in the ocean. The influence of local comprehensive channel selection on analysis field was evaluated using one-dimensional variational assimilation. The results show that the vertical distribution characteristics of atmospheric temperature, background errors covariance, the Jacobi matrix, and the weight function in the two regions are significantly different. The sensitivity channels selected are also different. Compared with the selected channel sets in other areas, the selected channel set in the corresponding region can significantly improve the effect of infrared high-spectral data assimilation.