← Back to Sessions

Moment Magnitude Estimation Using Machine Learning Algorithms for Western United States

Description: 

Accurately estimating the moment magnitude (M_w) earthquakes are fundamental for risk assessment, seismic hazard analysis, other seismological studies, and structural design. This study presents a comprehensive analysis of moment magnitude estimation using a dataset from the NGA-West2 database of about 21,000 earthquake records. We use machine learning regression techniques to train and test a model to predict M_w considering the following input features: Peak Ground Acceleration (PGA), 5%-damped pseudo-spectral acceleration (PSA) at 21 different periods, hypocentral distance (R_hypo), and the timed-average shear-wave velocity of the upper 30 m of soil (V_s30), fault mechanism, and the depth to the top of the rupture plane (Z_tor). The machine learning models considered in this study include Artificial Neural Networks, Support Vector Regression, Random Forest, and Gradient Boosting, each known for their predictive ability in nonlinear and complex datasets. Tree-based algorithms like Random forest and Gradient Boosting offer a more accurate model than Support vector and Neural networks. The proposed model uses a stacking method to combine different machine-learning techniques to provide a more accurate and robust framework. The proposed approach overcomes the limitations of conventional techniques by creating a rapid and straightforward process for estimating M_w. We use mean square error (MSE), mean absolute error (MAE), and the coefficient of determination (R²) to indicate that machine learning regression models can significantly improve moment magnitude estimations.

Session: Network Seismology: Recent Developments, Challenges and Lessons Learned [Poster Session]

Type: Poster

Date: 5/1/2024

Presentation Time: 08:00 AM (local time)

Presenting Author: Najme

Student Presenter: Yes

Invited Presentation: 

Authors