SAMPLING SITE LOCATION PROBLEM IN LAKE MONITORING

Monitoring is a task of actual water collection in water quality assessment from lakes. In this paper, we consider the problem of locating water sampling sites which is a significant issue in the design of lake monitoring. We formulate the location problem so that it can handle multiple purposes and constraints arising in monitoring tasks, and design an algorithm based on iterated local search (ILS). In our experiments, we apply our formulation to the real situation of Lake Biwa, the largest freshwater body in Japan. The ILS based algorithm finds such a location of sampling sites from which we can achieve better estimation of water quality distribution over the entire lake than the existing location, where we assume the existence of the true distribution and generate it by Lake Biwa Basin Hydrological and Material Cycle Model. Also, we observe that some points are newly selected in the output solutions more frequently than others; such points can be interpreted as potential sampling sites