Crop water requirement under micro irrigation systems using different evapotranspiration estimation techniques

Knowledge of exact amount of water required by different crop in a given set of climatological condition of a region is great help in planning of irrigation scheme, irrigation scheduling, effective design and management of micro irrigation systems. Crop water requirement is generally estimated by multiplying the reference evapotranspiration (ETo) by crop coefficient. The Penman–Monteith FAO 56 (P-M) model is recommended for estimating ETo across the world. However, the use of the P-M model is restricted by the unavailability of input climatic variables in many locations and the option is to use simple approaches with limited data requirements. In the current study, linear regression (LR) and Artificial neural network (ANN) techniques were used for estimating ETo in a semi-arid environment of Rahuri region Maharashtra, India. The four types of LR and ANN models were developed by varying the independent variables viz., Model1 (evaporation), Model2 (Tmax and Tmin), Model3 (Tmax, Tmin and SSH), Model4 (Tmax, Tmin, RHmax, RHmin and SSH). The comparison of the models were evaluated by statistical measures viz., correlation coefficient (R), Index of agreement d(IA), root mean square error (RMSE), mean absolute error (MAE), mean absolute percentage error (MAPE) and coefficient efficiency (CE) and ranking of models were assigned as per the statistical criteria. As per the results of overall comparison, it was observed that ANN4 with five inputs (Tmax, Tmin, Rhmax, RHmin and SSH) has secured 1 rank and exhibited overall best performance for performance criteria as R (0.940), d(IA) (0.968), RMSE (0.505), MAE (0.376), MAPE (8.419) and CE (0.884) followed by ANN3, ANN1, LR1, ANN2, LR4, LR3 and have secured 2nd, 3rd, 4th, 5th, 6th, 7th ranks respectively. It reveals that the proposed LR models may be adopted satisfactorily for estimation of ETo, the accuracy in ETo estimation may further be improved using corresponding ANN models for Rahuri region. Based on the overall results it was recommended that all ANN models can be used for estimation of ETo followed by all LR models as per data availability and simplicity of users and it can be useful for determination of crop water requirement, irrigation scheduling and design of micro irrigation systems.