Estimation of meteorological parameters by anfis method: The case of Kirkuk station, Iraq

The exploration and understanding of complex weather patterns to reliably forecast various climatic conditions has been an important part of the scientific process examinations globally. In this study, Adaptive Neuro Fuzzy Inference System (ANFIS) is used to create models to predict some of the most important climatic parameters. Relative humidity, pressure, temperature, and solar radiation parameters have been predicted by depending on the input variables which include daily and hourly data of temperature ( ℃), wind speed (m/sec), dew point (℃), relative humidity (%), solar radiation (watt/m^2) and pressure (hpa). The models set with data observed in Kirkuk station in Iraq, were tested with data of Mardin and Sanliurfa stations in Turkey. 2014-2017 and 2018 datasets were used as training and checking purposes, respectively. Moreover, the results of normalized and un-normalized dataset and ANFIS and Regression Analysis were compared. The models are evaluated by using root mean square error (RMSE), mean absolute error (MAE), and determination coefficient (R2). According to the results, ANFIS has much better performance than regression analysis for the estimation of daily normalized data, that had lower error values and higher R^2. Among all the parameters, the temperature parameter achieved the best performance using relative humidity and dew point as input variables. The results of humidity and temperature parameters have shown that these models showed a performance at Mardin and Sanliurfa stations at a similar level with Kirkuk station. If it aimed to predict in case of the lack of better models to use, these models can be used to get an idea about the parameters. Keywords: ANFIS, prediction, pressure, relative humidity, , solar radiation, temperature.