基于贝叶斯融合的土壤含水量估计

提出了一种基于贝叶斯融合的土壤含水量估计方法。该方法首先利用散射计与辐射计协同试验数据分别测量后向散射系数和亮温，并利用主被动模型提取农作物下垫面土壤含水量；然后利用贝叶斯融合算法将主被动反演结果进行融合，在农作物完整生长期，融合后土壤含水量与真实值相比，平均平方误差（MSE）小于3.56、平均绝对差值（MAD）小于1.36、平均相对误差（MRE）小于13.92%，同时分级贝叶斯与经典贝叶斯同真实土壤含水量的决定系数为0.77和0.60，证明基于贝叶斯理论的融合算法能够在整个生长期土壤含水量估计优于单一传感器。

Estimation of Soil Moisture Based on Bayesian Assimilation

This paper presents an estimation of soil moisture based on Bayesian assimilation. Firstly, the backscattering coefficient and brightness temperature are measured by scatterometer and radiometer in the joint experiment. Then the Bayesian assimilation is utilized for the results from active and passive retrieval models. The assimilative results of the mean square error (MSE), mean absolute deviation (MAD) and mean relative error (MRE) are 3.56, 1.36, and 13.92% smaller than real value, respectively. Moreover, the coefficient of determination result of hierarchical Bayesian (HB) and classic Bayesian (CB) are 0.77 and 0.60, respectively, which prove that the Bayesian assimilation results are better than inverse model which based on single sensors during the growth season.