Site selection of raingauges using entropy methodologies

Abstract

Data on precipitation – one of the important processes of the hydrological cycle – are necessary in water projects. Recognition of precipitation characteristics is necessary in dealing with flood control, plant water requirements, drainage systems design, and so on. Rainfall network design, defined as the proper density and distribution of raingauges in a region's rainfall network, is thus a necessary step toward the success of water projects and regional programming by effective use of data. Monitoring network design involves determining an adequate number and location of stations and providing design criteria for the stations, which may be different even for the same network. In this paper, the location of new rainfall stations in the Gav-Khuni basin rainfall network is determined using the concept of entropy with different criteria. For raingauge site selection, both sequential and genetic algorithms (SAs and GAs) are used. Maximisation of minimum entropy and maximisation of mean entropy objectives are defined for each algorithm. From these results, some stations can be nominated for installation by considering different criteria (e.g. economic, efficiency, precision) and other goals in the management of a watershed. The results indicate better performance and relative primacy of the GA in both objectives of minimum entropy and mean entropy maximisation compared with those obtained by the SA. Moreover, the results show that the nominated optimisation algorithm (GA) significantly improves the minimum and the average of the transinformation entropy, which indirectly influence an increase in the precision of rainfall sampling.