SPATIAL CONFIGURATION OF POTENTIAL OF EARTHQUAKES MAGNITUDE USING KRIGING METHOD WITH SEMIVARIOGRAM OF 3-DIMENSIONAL ANISOTROPIC

• Main Author: (20114035), SALIM
• Format: Theses
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/24174
• Institution: Institut Teknologi Bandung

Indonesia is an archipelago located in the conjuction of four tectonic plates, which Eurasia plate, Indo-Australia plate, Philippine plate, and Pacific plate. Relative movement of those slabs also create regional cesareans which then can develop to be source of earthquake. This research was intended to provides an overview of spatial configuration of potential earthquake in locations around observation of unknown using kriging method with semivariogram of 3-dimensional anisotropic. Earthquake data obtained from the website IRIS (Incorporated Research Institutions For Seismology). Data historic earthquakes that have occurred in Indonesia which in Banda Sea, on January 31, 2008. There are 21 directions to be analyzed of semivariogram of 3-dimensional anisotropic, it is based on direction of main angles 0o, 45o, 90o, and 135o. Assumed that process is stationary, fitting proses used by Exponential semivariogram model, thus obtained minor range (𝑎𝑥 ) is 0,1 for directions Latitude, second-major range (𝑎𝑦 ) is 0,3 for directions Longitude and longest major range (𝑎𝑧) is 0,4 for directions Depth (respectively in units of degrees). In anisotropic semivariogram, ellipsoid is transformed into a ball, so that anisotropic is simplified to be isotropic then it is used to kriging analysis. It is obtained that Simple Kriging (SK) is more suitable for data estimation of earthquake magnitude then compared with Ordinary Kriging (OK). It can be seen from Mean Square Error (MSE) Simple Kriging (SK) which is smaller than Ordinary Kriging (OK). Kriging analysis with 3D-anisotropic semivariogram of potential earthquakes magnitude provide an overview of spatial configuration that each estimation is influenced by neighbor spatial observation. If neighbor observations have large mean then estimation of unobserved location will be large as well, and reverse. Moreover, predicted results are also influenced by changes in the direction and angle of spatial lacations configuration.