Using Fuzzy Logic Approach on Evaporation Modeling

Abstract

Due to chaotic behavior of weather, meteorological variables are controlled by many parameters which should be analyzed by nonlinear approaches. Thus, long-term prediction is almost impossible. In this study, it is aimed to simulate evaporation by using three different independent
meteorological variables. In this context, daily data taken from Guzelyurt meteorological station in the Northern Cyprus is used. The measured
data includes observations of temperature, relative humidity and atmospheric pressure. The observations are between the years of 2005 and
2014. Adaptive Neuro Fuzzy Inference System method is more useful for the abovementioned meteorological modeling structed in general
stochastical ways in order to obtain a rule based evaporation model with membership functions. Fuzzy Logic model is structured with three
inputs: temperature, relative humidity and atmospheric pressure and one output: evaporation. Three membership functions are defined for each
input information, therefore we have 27 rules. These rules are defined to model the relations between the inputs and output. Here, Sugeno type
Fuzzy Inference System is chosen for modeling the evaporation. Then, fuzzy based output are compared to real weather data and observations.
The determination coefficient (R2) is obtained as 0.74, which is a statistically significant value. It will be planned to extend the study by considering other factors such as wind and solar radiation.