Neural Network Models for Prediction of Evaporation Based on Weather Variables

Artificial Neural networks (ANNs) is a computation method that can be utilized for predictions. In this study prediction of evaporation using ANN's multilayer perceptron (MLP) is attempted considering different weather variables viz. Relative Humidity Morning & Evening, Bright Sunshine Hours, Rainfall, Maximum & Minimum temperature, Mean Temperature and Mean Relative Humidity. The analysis is done over different parts of India viz. Raipur, Pantnagar, Karnal, Hyderabad and Samastipur. Weather of four lag weeks from week of forecast is considered for the model development. The lag periods were also utilized to develop weather indices. Subsequent two years were not included while developing the model for predicting evaporation for different locations. The performance of the developed models was evaluated based on Root Mean Square Error (RMSE).