IMPLEMENTATION OF DSS USING REGRESSION ANALYSIS FOR DAM MANAGEMENT

Abstract

Reservoirs created behind the dam on a river are meant for the storage of water which can be utilized for the purpose of irrigation, drinking water, power generation, industrial use, recreation etc. Besides fulfilling the requirement of the society, these reservoirs also helps in mitigation of floods during heavy rainfall in the catchment. The releases through reservoir for various purposes are controlled by gates. The decision to operate the gates is taken on the basis of observation of discharge at immediate upstream gauging site as per prevailing practice. The warning time to reach the water from these sites to reservoir called ‘inflow, is generally few hours depending upon the ‘time of travel’ of discharge from upstream site to the reservoir. In the event of very high rainfall for consecutive days it sometimes become difficult to accommodate and manage the large quantum of water reaching the dam site even with full opening of gates. This leads to catastrophic floods in downstream of reservoir. This calls for an efficient warning system having advance warning of inflow coming to reservoir so that regulated supply through gates could be released in advance to accommodate the incoming flood. An attempt is therefore made to develop an advance warning system with C# based on data mining tools. This paper describe the use of real time historical rainfall data to first predict discharges at two upstream siteswhich in turn used to predict real time inflow into the reservoir three days in advance. In order to develop a dedicated computer program for this warning system the data of a typical dam in Maharashtra, India, on the confluence of two rivers is utilized. It is found that the predictions made with the developed program were 75 to 86% correct and thus can be successfully utilized to control the gates and in turn will save the mankind from the danger of flash floods.