Development of Ground Motion Prediction Equations for the Makran Interface Subduction Zone

One of the main tools in seismic hazard analysis is Ground Motion Prediction Equations (GMPEs). Selection of appropriate GMPEs is an important step in hazard evaluation, which can cause accurate seismic design of structures. The GMPEs have been developed based on the local or regional or global data. Iranian plateau is a shallow crustal environment except the Makran region in south-east Iran in which the subduction events can happen. Due to the tectonic characteristics of the Makran subduction zone, different categories of GMPEs are required for seismic hazard assessment including GMPEs for shallow crustal events and subduction zone earthquakes (both in-slab and interface events). Taking into account that most of GMPEs in Iran have been provided for shallow crustal earthquakes, development of a GMPE model is needed for this subduction zone. Therefore, a new ground motion prediction model (GMPE) is developed based on Makran interface subduction events in this study. Due to the lack of recorded data in the Makran zone, this study is based on records of strong ground motions in other subduction zones, including events in Japan and Mexico from 1985 to 2018, as well as a database compiled by Atkinson and Boore (2003) including events recorded in Japan, Mexico, Alaska, Peru and Chile from 1968 to 1998. The database contains 1424 records of interface subduction events with M_w 5 to 9, distances less than 300 km and focal depth   less than 40 km. Since all records should be the same in terms of magnitude and measured by M_w, records that were reported with M_b and Ms were separated. Then, by examining the relationships and relevant articles and the conditions of this database, these values were eventually converted to M_w. Soil type of some records was reported according to geological characteristics of the region. The soil type of some others, due to the lack of geological characteristics information, using the information of the stations in that area, V_s30 values were extracted in different depths of the soil and using the studies, coefficients and relationships provided, the V_s30 value for each station was obtained. The soil type of some others was determined according to NEHRP classification. Finally, using presented studies by previous researchers to match this available information, this unification has been done. Existing records are categorized by soil type, M_w and focal length, and diagrams representing them are also plotted. The soil type of the records is A, B, C, D and E according to NEHRP classification. Processing of the database have been performed and the spectral accelerations in different periods have been obtained. In this work, some functional forms have been tested to understand the best model according to the most possible accuracy to fit our dataset. Then, using regression analyses, the ground motion prediction model is developed. Therefore, based on the comparison of the models, the dispersion of residues and the standard deviation of the models, best model has been selected for development. The intra-event and inter-event residuals for the proposed model have been obtained. Since the intra-event residual values are different for each earthquake record, it has been investigated in several different periods and also the fitting line is plotted in the diagrams. Inter-event residues have constant values in each earthquake. The residual value in each earthquake is considered average between the residues of different periods, the corresponding diagram is plotted and the fitting line is plotted in these diagrams. Spectral values ​​in the proposed model and the value of standard deviation are compared with other models in the world that shows that the proposed model has good accuracy for predicting spectral values and it is consistent with other models available.