地下水数值模拟不确定性的贝叶斯分析

Bayesian analysis of uncertainties in groundwater numerical simulation

本文在常用的确定性地下水流数值模型的基础上,提出了基于贝叶斯原理的不确定性分析方法,并通过算例,针对几种常见的不确定性因素,定量地描述了其对模型参数和模拟结果的影响。相较于传统的蒙特卡罗法等不确定性分析方法,该方法不仅考虑了先验的参数不确定性,而且考虑了模拟结果与观测资料的相似度,因此其结果更加符合实际情况。此外,该方法能够被用于风险分析,可为地下水资源评价及开发利用提供科学的依据。

The uncertain factors in groundwater numerical simulation can affect the reliability of simulation results.Based on commonly used deterministic groundwater flow numerical model,a Bayesian uncertainty analysis method is proposed.A set of examples are given to quantitatively demonstrate the influences of several uncertain factors on model parameters and simulation results.Compared with the traditional uncertainty analysis methods such as Monte Carlo simulation,the proposed Bayesian uncertainty analysis method considers not only prior parameter uncertainties,but also the likelihood between simulation results and observations,thus the results of analysis are much more accord with reality.This method can be used in risk analysis to provide scientific decision-making basis for evaluation and exploitation of groundwater resources.