Estimating daily pan evaporation using data mining process

This study investigates the applicability of the data mining process in estimation of daily pan evaporation, a fundamental element in the hydrological cycle. Firstly, the models were developed using autoregressive modeling, frequently preferred in hydrological studies, for Lake Egirdir in the southern part of Turkey, and the suitability of the AR(3) model was shown. Hence, the previous 1-, 2- and 3- day, daily pan evaporation values of Lake Egirdir were used to develop the other DM models. The correlation coecient and root mean square error criteria were used for evaluating the accuracy of the developed models. When the results of the developed models were compared to observed pan evaporation according to these criteria, it was determined that the AR(3) model is a little more appropriate in estimation of daily pan evaporation. Consequently, it was shown that DM models are useful, as they are based on only daily pan evaporation data and do not include meteorological parameters.