双通道微波辐射计大气折射率剖面反演的神经网络算法  

Neural Network Method to Retrieve Atmospheric Refractivity Profiles By using Dual-channel Microwave Radiometer Measurements

作　　者：王波[1,2] 赵振维[2] 

机构地区：[1]中国海洋大学信息科学与工程学院,青岛266071 [2]中国电波传播研究所,青岛266071

出　　处：《飞行器测控学报》2007年第5期1-4,共4页Journal of Spacecraft TT&C Technology

摘　　要：基于地基双通道微波辐射计（23．8GHz和31．65GHz）所测天顶方向的亮温测量数据和地面气象参数，给出了一种利用微波辐射测量反演大气折射率剖面的神经网络算法。利用青岛地区历史探空数据仿真的大气辐射亮温对神经网络进行了训练，回归得到了三段折射率剖面模型，并对利用实测亮温反演的大气折射率剖面与探空实测折射率剖面和模式剖面进行了比较分析，分析结果表明利用微波辐射计反演的折射率剖面与实测剖面间有很好的一致性，较三段折射率剖面模式具有更好的反演精度。A neural network method to retrieve atmospheric refractivity profiles using brightness temperatures measured by ground-based 23.8GHz and 31.65GHz dual-channel microwave radiometer and surface temperature, pressure and water vapor density is proposed in this paper. The BP neural network is trained with brightness temperatures and refractivity profile data simulated by the historical radiosonde data of Qingdao, and a three-segment model of atmospheric refractivity profiles is also obtained with the same historical radiosonde data. The retrieved atmospheric refractivity profiles are compared with the measured profiles of radiosondes and profiles of three-segment model. The compared results show that the retrieved profiles have good agreement with the measured profiles, and have better retrieval accuracy than the three-segment model.

关 键 词：反演 微波辐射计 折射率 神经网络 

分 类 号：TN011[电子电信—物理电子学]