Development of neural model for precipitation considering climate change under economic growth

Abstract

Reliable forecasting of future precipitation influenced by climate change scenario is an important field of research. The present study is focused on simulating the impact of climate change on precipitation behavior of Kshipra River, Madhya Pradesh, India which is a southern tributary of Yamuna River basin. CGMs are the most reliable sources available for the future climate, downscaling involves conversion of large scale GCM outputs of climate variables to local scale hydrologic variables. An attempt has been made to downscale the GCMs using Artificial Neural Network (ANN).The developed model used HadCM3 monthly weather data under A1B scenario (Rapid Economic Growth, A balanced emphasis on all energy sources) to determine the monthly precipitation at a specific site. Model performance has been evaluated in terms of coefficient of correlation (R), Mean square Error (MSE), Root Mean Square Error (RMSE) and Mean Absolute Error (MAE). It was observed that the ANN model predicted precipitation dependably correlated with observed precipitation. (R=0.911 and 0.853 during training and validation runs)