MODELING OF GEOMETRIC AND ZONAL ANISOTROPIC SEMIVARIOGRAM CASE STUDY OF EARTHQUAKE DATA IN MALUKU

• Main Author: Muthi'ah Rahmah, Nabila
• Format: Final Project
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/39134
• Institution: Institut Teknologi Bandung

Indonesia is a country with high earthquake potential, because it located in the Ring of Fire, the region with high earthquake and volcanic activity potential. Not only that, Indonesia, especially the east region, is also the meeting point of three plates: Eurasia, Indoaustralia, and Pacific. Earthquakes that occurred cause many disadvantages. Therefore, mitigation efforts needed, and one of the efforts that can be done by researcher is modeling magnitude and depth of the earthquake using spatial analysis. This research using earthquake data in Maluku at 1998, consists of coordinate data in longitude and latitude, earthquake magnitude (SR), and depth (km). Semivariogram is a measurement of spatial relation between locations. This research using 3 types of semivariogram models: exponential, spherical, and Gaussian. When model the data, directions between the pair of earthquake locations are also be counted, so the anisotropy can be detected. Anisotropy is a condition when the direction affects semivariogram value. There are 2 types of anisotropy: geometric anisotropy, when the range differs at the different directions; and zonal anisotropy, when the sill differs at the different directions. In this research, by fitting the model to experimental semivariogram, the anisotropy that occurs is a mixed one, when the range and sill differs for every direction group that used, that is West-East (W-E), North-South (N-S), Southeast-Northwest (SE-NW), and Southwest-Northeast (SW-NE). Anisotropy axes that obtained from the scatter plot of the pair of observations and from the model fitting is on SW-NE and SE-NW directions. Anisotropic semivariogram model obtained as the sum of anisotropic model of the anisotropic directions and two isotropic models for the rest. From the sum square error of the model, obtained that the best model for magnitude data is Gaussian, meanwhile the best model for depth data is spherical.