TITLE:
Comparison of Different Ann Approaches in Daily Pan Evaporation Prediction
AUTHORS:
Yildirim Dalkiliç, Umut Okkan, Nesrin Baykan
KEYWORDS:
Daily Evaporation; Pan Evaporation; Evaporation Prediction; Penman; LMANN; RBNN; GRNN
JOURNAL NAME:
Journal of Water Resource and Protection,
Vol.6 No.4,
March
27,
2014
ABSTRACT:
Nowadays,
one of the most important effects on water resources under climate change is increasing
of free water surface evaporation which depends on the increasing of
temperature. In basins, where there are no observed data, free water surface
evaporation is taken into account depending on historical temperature and
similar data and their long-term statistics. Predicting of real value of
evaporation contains some uncertainties. The modeling of evaporation with a
small number of predictors has crucial importance on the regions and basins
where measurements are not sufficient and/or not exist. In this presented
study, daily evaporation prediction models were prepared by using empirical
Penman equation, Levenberg-Marquardt algorithm based on "Feed Forward Back
Propagation Artificial Neural Networks (LMANN)", radial basis neural
networks (RBNN), generalized regression neural networks (GRNN). When the models
were compared, it was noticed that the results of neural network models are
statistically more meaningful than the Penman equation.