提出了一种基于遗传神经网络算法的主被动遥感协同反演地表土壤水分的方法.首先，建立一个BP神经网络，并采用遗传算法对BP网络的节点权值进行了优化.然后分别将TM数据(TM3,TM4,TM6)、不同极化和极化比的(VV，VH，VH/VV)ASAR数据作为神经网络的输入，土壤水分含量作为网络的输出，用部分实测数据对网络进行训练并反演得到研究区土壤水分布图.最后，利用地面实测数据分别对遗传神经网络优化算法的有效性和主被动遥感协同反演的效果进行了验证，结果表明，新优化算法是有效可行的，且TM和ASAR协同反演的结果比两者单独反演的结果明显要好，体现了主被动遥感协同反演土壤水分的优势与潜力.

active andA new semiempirical model is presented for soil moisture content retrieval, using ENVISAT ASAR and LANDSATTM data collaboratively. Firstly, a back propagation(BP) neural network algorithm(GA) is introduced, and a genetic algorithm is applied to optimize the weights of the node of BP neural network. Then the TM bands (TM3, TM4, TM6) and ASAR data(VV, VH, VH/VV) are taken as the input of the GABP neural network, and the output corresponds to the ground soil moisture. The partial field measurements of soil moisture are used as training samples to train the network and to achieve the map of soil moisture distribution. The field measurements are used to test the validity of the BP neural network algorithm and effectiveness of the active and passive remote sensing cooperative inversion. The comparison between the inversion using single data set(TM or ASAR), and the cooperative inversion of active and passive remote sensing data demonstrates that the new algorithm is more effective, and shows considerable potential in soil moisture retrieval by integrating active and passive remote sensing data. passive remote sensing; GABP neural network; soil moisture; inversion