提出一种基于集合卡尔曼滤波的一维数据同化系统, 对不同深度土壤层的水分含量进行同化, 该系统的模型算子为分布式水文模型, 观测算子是积分方程模型和条件温度指数模型.于2008年6月1日至7月2日在黑河流域进行了同化实验, 结果表明, 集合卡尔曼滤波能较好地处理强非线性问题, 与单独DHSVM模型模拟土壤水分含量相比, 同化的表层和根层土壤水分含量精度有明显提高, 其中盈科站表层的均方根误差和平均误差分别减小了0.0217和0.0329, 根层的均方根误差和平均误差分别减小了0.0193和0.025；临泽站的精度也有明显提高, 表明多源遥感数据的同化在地表土壤水分含量的估计中具有较大的潜力.

This paper proposed a one-dimensional soil moisture content data assimilation system based on the ensemble Kalman filter (EnKF), the distributed hydrology-soil-vegetation model (DHSVM), microwave radiative transform model (advanced integration equation model, AIEM) and optically semi-empirical model (temperature-vegetation dryness index, TVDI) for soil moisture content retrieval in bare soil. Numerical experiments were conducted at the middle reaches of the Heihe River Basin from June 1 to July 2, 2008. The results indicate that EnKF is an efficient approach to handle the strongly nonlinear problem. By assimilating multi-source remote sensing observations, the assimilation method works successfully with DHSVM and significantly improves the soil surface moisture estimation in the surface layer and root layer, the root mean square error (RMS) and mean bias errors (MBE) decrease 0.0217 and 0.0329 in surface layer and 0.0193 and 0.025 in root layer respectively, both in Yingke station. In the Linze station, the retrieve precision was also improved. It is practical and effective for soil moisture content estimation by assimilation of multi-source remote sensing data.