本文讨论了35通道地基微波辐射计反演温度、水汽廓线的BP神经网络反演方法,对神经网络的层次结构、样本选择、训练方法等进行了分析,并将反演结果与探空数据进行了比对,分析了误差产生的原因。结果表明,BP神经网络反演的廓线数据具有很好的精度,满足实际使用要求。%A method of BP neural network used for inversion of temperature and water vapor profiles with 35 -channel ground-based microwave radiometer is discussed; and its hierarchical structure, sample selection and training method is analyzed. The result of retrieval is compared with that observed with radiosonde and the cause of the error is analyzed. The result shows that the data of temperature and water vapor profiles retrieved by using BP neural network is accurate and using this method can meet the requirement of practical applications.
展开▼
