| Abstract Submission No. | ABS-4.1-0158 |
| Title of Abstract | Novel P-Wave Based Parameters for Earthquake Magnitude Prediction Using Machine Learning in the Uttarakhand Himalaya |
| Authors | Monika*, Cyril Shaju, sandeep, Pankaj Kumar, Vikas Singh, Suraj kumar Pal |
| Organisation | Banaras Hindu University, Uttar Pradesh |
| Address | Depaartment of Geophysics, Banaras Hindu University,Varanasi varanasi, Uttar Pradesh, India Pincode: 221005 E-mail: yadavmonika3678@gmail.com |
| Country | India |
| Presentation | Oral |
| Abstract | Accurate and rapid estimation of earthquake magnitude is a critical component of earthquake early warning systems. In this study, seismic waveform data from the Uttarakhand Himalaya were utilized, with event records obtained from the P-Alert network. A set of novel parameters were introduced in addition to source¿receiver distance. These parameters are extracted from the initial 3 seconds P-wave window. These parameters were specifically designed to capture both spectral and temporal characteristics of early-arriving p phase, providing new insights into their relation with earthquake magnitude. The features were then employed to train and evaluate regression models. On the initial phase of this work three approaches were tested: multiple linear regression, Random Forest regression, and Gradient Boosting regression. Model performance was assessed using root mean square error (RMSE) and coefficient of determination (R²). Results demonstrated that nonlinear machine learning models, particularly Random Forest and Gradient Boosting, achieved higher predictive accuracy compared to linear regression, and feature importance analysis revealed the significant contributions of the proposed parameters to magnitude estimation. This study establishes the effectiveness of incorporating novel P-wave based parameters from the P-Alert network in the Uttarakhand Himalaya, offering a robust data-driven framework to improve rapid earthquake magnitude prediction and strengthen early warning capabilities in the region. |
| Keywords | EEWS, Machine Learning, P-alert, Himalaya |