Forecasting the solar energy potential for the Adana Region using machine learning methods from satellite and meteorology data

Due to many reasons such as population growth, rapidly developing industrialization, development of technology, and increase in people's needs, the world's need for large amounts of energy has emerged. It has become necessary to meet this need either from power plants based on fossil fuels, which are traditional methods, or from sources called clean energy, such as solar energy, wind energy, which is a new generation energy source. In this study, it is aimed to calculate the amount of solar energy, based on Adana meteorology station, in the area covering the self region in order to overcome this energy deficit. For this purpose, remote sensing data and machine learning algorithms have been used in order to examine the solar radiation in this wide region in the fastest and most cost-effective way. NOAA-AVHRR satellite data was obtained from NASA' web site, calibration and geograpic processes were performed and land surface temperature values were calculated. 8 different data sets were prepared to be run in machine learning algorithms. These data sets were run with machine learning algorithms by using WEKA software, it was investigated with which data sets the algorithms gave appropriate results. When these results are examined, it has been determined that the linear regression algorithm has 79.9%, the multilayer perceptron algorithm has 88.4% and the support vector regression algorithm has 89.2% success rates.