← Back to Sessions

Epistemic Uncertainty in Ground-Motion Prediction in the Indian Context: Evaluation of Ground-Motion Models (GMMs) for the Himalayan Region

Description: 

Empirical ground-motion models (GMMs), the essential element of probabilistic seismic hazard analysis (PSHA), exhibit two types of uncertainties for a given rupture scenario: aleatory and epistemic. The former is captured by the residual standard deviation (σ) of GMM, whereas the actual epistemic uncertainty associated with ground-motion prediction is unknown. Epistemic uncertainty increases significantly with less recorded data in magnitude and distance ranges that dominate hazards at a particular location. In the Indian context (mainly along the Himalayan arc), constraining epistemic uncertainty poses a major challenge due to less recorded data. The present study investigates two components of uncertainty in ground-motion prediction in the Indian context. We use a recently compiled strong-motion dataset to evaluate the published GMMs for the active crustal regions, mainly across the Himalayas. First, we provide a brief review of the GMMs developed, followed by statistical analysis. The inherent uncertainty, due to regional variations and incorrect parameter estimates, is also explored. The residual analysis involves decomposing the total residuals into between and within-event components using a mixed-effects regression algorithm. For evaluation, 13 GMMs with 519 records for distance ≤ 300, magnitude ≥ 3, years 2005-2017, were utilized. For most GMMs, the current functional form can’t capture the observed magnitude and distance scaling exhibiting a significant bias. The residual analysis indicates that GMMs performance decreases with increases in periods, particularly beyond T ≥ 0. 8s and total aleatory variability, σ is lower at shorter periods but increases at higher periods.  Our analysis shows that a large aleatory and epistemic uncertainty is associated with the Indian GMMs. Further studies are required to better understand different uncertainty sources with more recorded data to provide constraints on the developed GMMs along with ground-motion simulations.

Session: Understanding and Modeling the Uncertainties in Earthquake Ground Motions

Type: Oral

Date: 4/20/2023

Presentation Time: 11:00 AM (local time)

Presenting Author: Sanjay Bora

Student Presenter: No

Invited Presentation: 

Authors