基于PEST的地下水数值模拟参数优化的应用

本文介绍了参数优化工具PEST的工作流程与基本原理,并建立了一个理想地下水流动模型作为实际观测资料对4个参数进行优化。通过分析优化过程与结果:在参数设置为不同初值的情况下,PEST均能获得与真实值非常接近的优化结果,若初值与真实值越接近,优化迭代次数及调用模型次数越少。通过分析优化过程还发现:参数分区所占比例与参数敏感度有一定的相关性,分区面积越大,该参数敏感度越高;观测资料越丰富,优化结果与真实值越接近,反之优化结果误差较大,这对地下水网布设与优化起到一定的指导作用。

Application of groundwater numerical simulation for parameter optimization based on PEST

The workflow and the basic principle of the parameter estimation tool PEST are introduced briefly, and an ideal model for groundwater flow is established based on MODFLOW. It is used to obtain the actual observations which can be used in parameter estimation process by PEST. The parameter estimation results provided by PEST are almost equal to the actual values which are set in the ideal model. By comparison, the numbers of the iteration and the call model are very few if the initial value of the parameter is close to the actual value. By analyzing the optimization process of PEST, it is found that the sensitivity of the parameter is high when the proportion of the corresponding parameter area is large, so there is the correlation between them. Moreover, there is a positive correlation between the amount of observational data and the estimation results. Therefore, it is also important to guide the collection of the hydrogeological data for increasing the reliability of groundwater model.