Application of soft computing techniques to predict reservoir water level

Abstract

In this work, the reservoir water level has been predicted using one of the soft computing techniques named Artificial Neural Network. The reservoir water level is influenced by many parameters. Among which the most influencing parameters have been considered here: amount of rainfall, temperature and evaporation. For this analysis, the reservoir made on Shetrunji River Dam in Dhari, Amreli district, Gujarat, India has been chosen as it was overflown seven times in last ten years. This shows the importance of water level prediction at this particular reservoir. The Neural Network is trained using the past data collected and further used to predict water level for the unknown data. The approach of the multiple regression is also shown for its comparison with the Soft computing approach. Computations and experimental works were done by programming in software MATLAB. Such modeling is useful for planning and decision making of opening gates for reservoir operation particularly during monsoon and water scarcity