Evaluation of geomorphometric parameters produced using remote sensing and geographic information systems in terms of climate

Since the formation the Earth, has been shaped and continues to be shaped due to sudden and long-term changes in climatic conditions, meteorological events, tectonism, and the constant movement of the earth's crust. Sudden and long-term climate changes experienced on a global or regional scale are among the important factors that shape landforms under the influence of physical and chemical processes. These processes create different erosion and storage patterns on the earth. The aim of this thesis is to investigation the relationship between landforms and climate classes by using geomorphometric parameters. In this context, Supervised Machine Learning Classification models were developed using geomorphometric parameters and climate classes of the study areas where Köppen-Geiger climate classes are typically observed in Turkey. Within the scope of the study, machine learning models were developed with random forest, support vector machines and k-nearest neighbors classification algorithms using the training data set. Classification results were obtained using the developed models and test data. As a result, the overall accuracies for the random forest model, support vector machines model and k-nearest neighbors models were obtained as %98,80, %99,70 and %99,30, respectively, while the kappa statistic values were obtained as 0,98, 0,99 and 0,99. Within the scope of the study, it was determined and evaluated that the geomorphometric parameters that are important in distinguishing climate classes are height, maximum height and valley depth.