论文部分内容阅读
为了估计具有垂向空间异质性的田间尺度土壤水分运动参数,提出利用原位多点土壤含水率观测数据反演多层土壤持水和导水特征参数的Bayes方法。该方法基于Hydrus-1D模拟模型建立Bayes推断模型,采用Markov Chain Monte Carlo(MCMC)方法求解Bayes联合概率分布,得到参数的后验边缘分布。将该方法应用于华北平原一典型农田土壤剖面,参数反演结果反映了空间变异性,与室内试验值具有较好的一致性,相应的土壤水分运动模拟具有较高精度,从而验证了该方法的有效性。
In order to estimate the soil moisture movement parameters with vertical spatial heterogeneity, a Bayesian method for retrieving multi-layer water and soil hydraulic conductivity parameters from in situ multi-point soil moisture data was proposed. The method builds a Bayesian inference model based on the Hydrus-1D simulation model, and uses the Markov Chain Monte Carlo (MCMC) method to solve the Bayesian joint probability distribution and obtains the posterior edge distribution of the parameters. The method was applied to a typical farmland soil profile in the North China Plain. The results of parameter inversion reflected the spatial variability, which was in good agreement with the indoor test values. The corresponding soil moisture movement simulation was of high precision, which verified the method Effectiveness.