Estimation of monthly evaporation in hilly regions of Uttarakhand using machine learning techniques

Evaporation estimation is one of the complicated and important measures of the hydrological cycle due to its complex behavior for planning, management, and development of water resources. In this study, the Multivariate Adaptive Regression Splines (MARS) and Random Forest (RF) models were utilized to estimate mean monthly evaporation at Ranichauri. The mean monthly meteorological variables as, maximum and minimum temperature, morning and afternoon relative humidity, rainfall, morning and afternoon vapor pressure, wind speed, morning and afternoon wind direction, solar radiation, and evaporation were used for the development of the models. The results obtained by MARS and RF models were compared based on statistical indices root mean square error (RMSE), correlation coefficient (r), percent bias (PBIAS), and Nash-Sutcliffe efficiency (NSE). The results indicated that the performance of the RF model is better than the MARS model during the training period and the MARS model is better than the RF model during the testing period with nine input variables in estimating mean monthly evaporation at Ranichauri. Based on the present results when compared with the previous study's results done by many researchers, reveal that the methods can easily be implemented in hilly regions with greater accuracy.