Applicability of Genetic Algorithm for Identifying the Groundwater Contamination Source

概要

論文の詳細を見る
Estimating the characteristics of groundwater contamination source is a very important topic in geo-environmental studies. In the present study, an applicability of Genetic Algorithm (GA) to the problem of identifying the contamination source is examined. Six parameters of the contamination source are identified by this method, which are its location, its size (the length of the source that is assumed to extend linearly), its initial concentration, reduction factor, and the time when contamination begins. The effect of the data acquisition intervals and the number of observation points on GA performances is investigated. It is found that increasing the number of observation points is more effective than increasing the observation frequency at one point in order to improve the accuracy of estimation. The variations of concentration at a point far from the source are calculated using estimated parameters to examine the predictability. As a result, it is shown that the accuracy of estimated parameters greatly affect the prediction. Moreover, a limit of application of GA on heterogeneous aquifers generated based on geo-statistics is also examined. It is found that if the hydraulic conductivity is distributed within two orders of the heterogeneity, GA provides the parameters of practical precision.
日本雨水資源化システム学会の論文