Planning of runoff retarding measures at the watershed scale using genetic algorithms

摘要

Management of surface runoff and protection of water and environment are key elements of any watershed-wide development. The necessity of analyzing the watershed area and the receiving water as one single system is obvious. In order to mitigate the flood peak at the downstream, micro-management and source control measures that include scattered small-scale infiltration facilities and subbasin outlet type detention/retention ponds are commonly used in watershed. The optimal planning and design of this networks belongs to the large combinatorial optimization problems that are very difficult to handle using conventional operation research techniques. Genetic Algorithms (GA) forms a radically different approach to optimization. The aim of this paper is to apply this technique to determine the most cost-effective placement strategy of runoff retarding facilities under constraints of least-cost and desired level of flood peak reduction requirement. Genetic algorithms that were interfaced with the hydrologic model through encoding-decoding processes performed optimization of flood retarding schemes. Wudu watershed located in upstream of Keelung River basin in northern Taiwan was described in the paper to illustrate the methodology.