Geohazard Risk Assessment

In the event of an earthquake, rescue workers use rapid response maps to assess intensity and hazards such as soil liquefaction, which occurs when an earthquake exerts stress on water-saturated soil. Current liquefaction hazard maps are valuable in a disaster situation, but many at-risk areas lack the necessary geological data and resources to effectively use current models.

Professor Laurie Gaskins Baise in the Geohazards Research Lab specializes in empirical and theoretical models to describe and predict natural hazards. Most recently, she has been developing a better model to predict the probability of soil liquefaction in the event of an earthquake. Baise and graduate student Jing Zhu recently published details on their model in the journal Earthquake Spectra.

“Liquefaction is predicted either at a specific location using in situ tests (standard penetration tests or cone penetration tests) or regionally using surficial geology maps. Both are difficult if not impossible to get on a global scale,” said Baise, “Our model allows you to predict whether liquefaction is expected anywhere in the world as soon as the earthquake happens.”

Since many parts of the world do not have detailed, comprehensive soil data, Tufts researchers are testing the viability of more easily obtainable information through remotely sensed data such as digital elevation models to help identify risk areas. Zhu examined well-documented cases of soil liquefaction in the 2011 Christchurch, New Zealand and 1995 Kobe, Japan earthquakes in order to build a predictive model based on geospatial variables, specifically distance from and elevation above bodies of water, and earthquake-specific parameters such as peak ground acceleration.

Zhu tested the reliability of the model using data from the 2010 Haiti earthquake. Since Haiti had insufficient preexisting data on soil properties, traditional liquefaction maps were not available to predict what areas were susceptible to liquefaction after the earthquake. However, Zhu’s model correctly predicted known instances of soil liquefaction using existing geographic data.

Zhu said, “I am currently working on expanding the database to include over 15 earthquakes and updating the model. This time, we explore many new parameters and perform a more thorough statistical analysis. The updated model will be more reliable in different geological environments.”