Assessment of response surface method and hybrid models to predict evaporation (case study: Chahnimeh and Pishein reservoirs in Sistan and Baluchestan Province of Iran)

Evaporation is one of the major components of the hydrological cycle and its accurate estimation affects water resource management, irrigation planning, and environmental studies. In this paper, an explicit mathematical function was proposed based on the response surface function (RSF) to predict pan evaporation. The second-order RSFs were found to be more efficient and accurate than the linear RSF. The ability of the proposed RSF-based method was investigated with a wavelet-based adaptive neuro-fuzzy inference system (WANFIS) and wavelet-based support vector regression (WSVR) to estimate the daily evaporation of the pan. Daily climatic variables such as wind speed, sunshine hour, air temperature, and relative humidity from two climate stations (Chahnimeh and Pishien reservoirs) in Iran were used as inputs in all models. The modelling results were evaluated using RMSE, d, NSE, CI, and R2 in both regions. RSF, WANFIS, and WSVR all could be successfully employed in modelling the evaporation estimation. WANFIS model shows the best results and its estimated statistics factors in Chahnimeh are RMSE = 3.05, NSE = 0.90, d = 0.98, CI = 0.88, and R2 = 0.98 and for Pishien are RMSE = 1.51, NSE = 0.84, d = 0.96, CI = 0.81, and R2 = 0.92. Also, the RSF model provides satisfactory results similar to the WSVR model.