不同目标函数对土壤水力参数反演效果评价

Inversion Effect Evaluation of Determining Soil Hydraulic Parameters with Different Objective Functions

土壤水力参数是表征土壤水力性质的重要参量,是准确预测土壤水分及溶质运移等过程的基础。基于GML非线性参数优化算法,构造了三个不同的目标函数,同时反演van Genuchten-Mualem模型中三个土壤水力参数(进气吸力倒数α、孔隙体积大小分布指数n和饱和导水率Ks),并分析了同一目标函数下不同降雨条件对反演结果的影响。结果表明,选取不同的目标函数对反演结果有显著影响,以累积入渗量作为目标函数能有效反演参数Ks,以压力水头和累积入渗量作为多目标函数则能同时有效反演α、n、Ks三个参数;此外,降雨条件对反演结果也有一定影响,相对于均匀降雨,在非均匀降雨条件下以压力水头作为目标函数对反演结果的精确度和准确度有较大的提升;以累积入渗量作为目标函数对反演结果的精确度略有提升但降低了反演结果的准确度。

Soil hydraulic parameters are the important manifestation of soil hydraulic properties.Knowledge of the soil hydraulic parameters is especially important when numerical models are used to simulate variably saturated water flow and solute transport.By using the Gauss-Marquardt-Levenberg（GML）nonlinear parameter optimization method,the unknown soil hydraulic parameters（α,n,Ks）in van Genuchten-Mualem＇s model were estimated with different objective functions under different rainfall conditions.The results show that the parameter Kscan be estimated exactly by using the cumulative infiltration data only.The multi-objective function based the combination of pressure head and cumulative infiltration data can inverse these three parameters effectively.Under non-uniform rainfall condition,the objective function based the pressure head data can improve the precision and accuracy of the inversion results.While the objective function based cumulative infiltration data can improve the precision of the inversion results but reduce its accuracy.