Seismic Hazard Assessment using Machine Learning For Balochistan Region

Abstract

The Balochistan Region is in a seismically active belt that has been exposed to many destructive earthquakes in the past in the boundary of Pakistan. In this study, Peak Ground Acceleration (PGA) in the Balochistan Region has been predicted using Gene Expression Programming (GEP) after comparing its efficiency with Artificial Neural Network (ANN) and Adaptive Neuro Fuzzy Inference System (ANFIS) within range of 28.070º to 31.000º N latitude to 66.710º and 69.330º E. This is a simple approach to assess seismic hazard assessment using machine learning algorithm. A homogeneous catalogue was obtained by collecting data from Pakistan Meteorological Department (PMD) which comprises 2021 earthquake events in the past divided into 600 data sets. An Expression Tree (ET) of six inputs (longitude, latitude, depth, magnitude, seismic energy, and logarithmic seismic moment) and one output (PGA) was run on GEP Expo tool 5.0. The most suitable model with a correlation function R2 of 91% was obtained after several iterations of GEP. All results were statistically evaluated and validated from data of United States Geological Survey (USGS) and International Seismic Centre (ISC). The present study explicitly provides the applicability of GEP to predict Peak Ground Acceleration.