Comparision of local geoid determination methods with artificial neural networks, kriging and polynomials in the example of Amasya province

Instead of conventional surveying techniques, GNSS measurement methods, which are frequently preferred today, enable the ellipsoidal height to be determined in a short time. However, the ellipsoidal height does not make any physical sense. Therefore, orthometric heights, which have a physical meaning, are used in engineering applications. In order to transform the ellipsoidal height to orthometric height, the geoid model must be known. The geoid model can be determined locally as well as within the scope of this thesis. Within the scope of this thesis, it is aimed to determine the local geoid model in Amasya region. Sixty-two C3 grade GNSS/levelling station were used in the study. The data were divided into two groups as model and test points. Polynomials, kriging and artificial neural network methods were used to create the local geoid model in the study area. The statistics determined as a result of the application were evaluated according to the square mean error criterion. It is observed that the best results have been obtained with the multilayer perceptron artificial neural networks method in the study area. The closest results to this method have been obtained with the kriging method.