Application of Gustafson-Kessel algorithm in finding seismically co-pattern zones, A case study on Meybod city

The current paper deals with a new use of fuzzy logic in the domain of seismic zonation methodologies with a case study in Meybod, Iran. Determining the risk function of an earthquake is an important issue and for the complete solution, the seismic specialist shall necessarily find the soil response and present an appropriate zonation output to the civil engineer. For tackling this problem, a fuzzy clustering method has been applied in acquiring microtremor data more specifically in the city of Meybod. Dataset is divided into four subsets based on its intrinsic complexity by GK fuzzy clustering. Features in this classification practice are including the coordinates, the output dominant frequency of H/V method and the related amplitudes. Euclidian distance matrix norm is constructed to detect hyper ellipsoidal clusters with different orientations (shape and size) in the dataset. The cluster means are then refreshed in an iterative manner so as to identify the uniform seismic answer into the isolated gatherings. We used this method to determine the four divided seismicity regions with different range of frequencies. In addition, various type of soil structures in Meybod city with high and weak risky area have been cleared and can be applied in hazard and earthquake engineering projects. This approach was significantly well matched with clay and silt dominant on soil observed in the boreholes.  Finally zonation maps based on this new method is provided.