Estimating daily pan evaporation from some existing time-series models

Nowadays, wise scheduling of water resources is an important task for sustainable development. In this scope, accurate knowledge about daily evaporation from surface water is important for optimum agricultural products. Shortage or lack of evaporation sites, quality of the data and gaps in pan observations force the researchers to use prediction models. In this work, three different time series were applied for estimating daily pan evaporation in Hamedan province (Ekbatan dam, Kooshk-Abad and Agh-Kahriz). Evaluations of time series data indicated that pan evaporation of all stations, contains no seasonal trend. Based on the results, the ARIMA showed the highest performance in estimating the daily pan evaporation for the selected sites. With respect to the obtained errors, the best time-series structure were ARIMA(1,1,1) for Agh-Kahriz and ARIMA(2,0,1) for Ekbatan dam and Kooshk-Abad, respectively. The calculated R2 and RMSE for selected sites were 0.93 and 2.44 mm/day for Agh-Kahriz, 0.93 and 1.23 mm/day for Ekbatan dam and 0.92 and 1.03 mm/day for Kooshk-Abad. For the selected model, the Mean Percentage Errors (MPE) varied between 5.85 to 6.92 percent, which denotes models overestimation versus the measurements. The standard deviations of the best performance model are: 0.15, 0.14 and 0.13 mm for Agh-Kahriz, Ekbatan dam and Kooshk-Abad, respectively.