摘要
地下水数值模拟受到众多不确定性因素的作用,直接影响了模拟结果的精度与可靠性.因此,分析这些不确定性因素对模型输出的影响十分必要.目前针对地下水模型输入参数不确定性的分析与研究已取得了一系列的成果,相对比较成熟.但对地下水概念模型不确定性的关注与研究不足.基于此,本次研究基于一个理想的地下水流模型,并针对该模型建立了三个不同结构的概念模型进行试验对比分析.基于蒙特卡洛模拟技术,分别对所建三个模型进行参数识别,并获取了地下水流模型输出的预测概率分布.研究结果表明当地下水流概念模型越接近真实模型结构时,模型输出的预测分布更加准确与稳定.单一考虑模型输入参数的不确定性无法弥补由其概念模型偏差所导致的预测不确定性.概念模型是地下水流数值模拟模型的基础,应当加强对地下含水介质空间结构方面的资料收集,以减少地下水模拟的不确定性.
Groundwater numerical simulation is influenced by many factors. The simulations are always deviate from observations. Therefore,it is necessary to analysis the influences of these factors on groundwater simulations. Recently,plenty of researches have been made for estimating model parameter uncertainty,and so,a series of meth- odologies have been proposed or developed for uncertainty analysis. Nevertheless, a little of attention is paid for assessing conceptual model uncertainty. Therefore, for the purpose of analyzing conceptual model uncertainty, a synthetical groundwater flow model is constructed in this paper. Three alternative conceptual models are constructed by considering incompleted model structures. Based on Monte Carlo simulation, these three models are calibrated for model parameters. The predictive probability distributions of groundwater budget terms for the three models are obtained from the results of Monte Carlo simulations. The results show that conceptual model is essential for groundwater numerical simulation. The more reliable the structure of conceptual model is, the more reliable the predictive distribution of model output is. Moreover,the conceptual model uncertainty cannot be compensated by parameter uncertainty. As a result,{or the real groundwater simulation, more resources and attentions should be paid [or the data collections on the structure of groundwater media.
出处
《南京大学学报（自然科学版）》
CAS
CSCD
北大核心
2012年第6期746-752,共7页
Journal of Nanjing University（Natural Science）
基金
国家自然科学基金(41172207
41030746
40725010
41071018
40730635)
环保公益性行业专项(201009011)
环保公益性行业专项(201009009-04-03)
关键词
地下水模拟
概念模型
不确定性
蒙特卡洛模拟
groundwater simulation, conceptual model, uncertainty, Monte Carlo simulation
