← Back to Sessions

Region-Specific Geospatial Liquefaction Model for Alaska by Bayesian Model Updating of the Global Liquefaction Model

Description: 

Global geospatial liquefaction models are currently being used to predict the probability of liquefaction occurring after an earthquake. However, these models are limited in their ability to generalize to new regions because they are only trained on data from previous events using a global dataset. Additionally, the geology, saturation, and seismic activity of different regions can vary greatly, making it desirable to develop specific liquefaction models for each region. A recent example is the Mw 7.1 Anchorage, Alaska earthquake that occurred on November 30, 2018. This earthquake caused widespread liquefaction in natural fluvial-estuarine environments and damaged and tilted residential buildings in southern Anchorage. In this study, a regional geospatial liquefaction prediction model is developed for Alaska using liquefaction observations from the 2018 Anchorage earthquake. For this purpose, a Bayesian logistic regression model was trained to combine the knowledge from the existing global model and the region-specific relations between geospatial explanatory variables for the Anchorage region. This can be done by selecting the prior distribution of the model coefficients from the global model and then sequentially updating the model for each liquefaction observation from the region to find the posterior probability distribution of the model parameters. To create a balance between the information gained from the global model and regional data, prior and likelihood functions are weighted to create an optimal liquefaction prediction model. The resulting model shows a more accurate liquefaction detection in the Alaska region.

Session: Regional-Scale Hazard, Risk and Loss Assessments - I

Type: Oral

Date: 5/2/2024

Presentation Time: 05:15 PM (local time)

Presenting Author: Hooman

Student Presenter: Yes

Invited Presentation: 

Authors