Akademik Veri Yönetim Sistemi
ARABIAN JOURNAL OF GEOSCIENCES, cilt.15, ss.15-21, 2022 (Hakemli Dergi)
Temperature data is one of the basic inputs of meteorological, hydrological and climatic studies. The completeness of this data is of great importance for reliability in research. This study aimed to compare the performances of various machine learning methods such as support vector machines (SVM), adaptive neuro-fuzzy inference system (ANFIS) and decision tree (DT) to infill missing air temperature data. Monthly average temperature data from 1968 to 2017 (50 years) was used to develop the models. In the established model, the data is divided as 80/20% (1968–2007 training/2008–2017 testing). Neighbouring stations, like Sarıkamış, Tortum and Ağrı, which have a high correlation with Horasan, were used as inputs to estimate the temperature data of the Horasan station. The most suitable machine learning method was chosen according to the mean square error (MSE), root mean square error (RMSE), mean absolute error (MAE) and determination coefficients (R2) of the training and test results. The ANFIS model with four sub-sets, triangular membership function, hybrid learning algorithm and 300 iterations was selected as the most suitable model. It was recommended using ANFIS to estimate monthly air temperatures in the northeastern part of Turkey and perhaps in other semi-arid climatic regions around the world.