Submitted By Sara Khoshnevisan
Assc. Research Fellow, RESA
Liquefaction is a phenomenon where vibrations from an earthquake cause fully or partially saturated soil to lose strength. For us non-geologists, think of the soil acting like a liquid during events like an earthquake. If you ever get bored, may I suggest looking into liquefaction videos on YouTube. It is only moderately terrifying, but completely fascinating. As you can imagine, liquid soil is not really a great foundation for large structures, making it a major cause of damage to buildings, bridges, and lifelines.
How do we fight this? In order to understand exactly what happens during the liquefaction process, we must evaluate and validate the site’s condition. Most times this happens through liquefaction quantification models. It is critical that these models continue to be updated and improved upon, so that events like the 1964 Niigata Earthquake in Japan do not happen again. Additionally, these models can help estimate earthquake-related losses, which can help communities become resilient to natural disasters.
At RESA, we crave being a part of the movement towards creating resilient communities. Therefore, we are conducting research focused on developing a high resolution database of site condition (quantified by Vs30) at several selected regions in the United States with a high risk of seismic activity. Particularly, we are interested in identifying and filling in gaps maintained in the current site condition data as well as providing validated data for earthquake models.
To evaluate the models a reference point is needed. Geologists and other cool people like to call this a benchmark. The benchmark database RESA has complied was achieved through screening available Vs30 database with high resolution Vs30 measurements. The methodology for evaluation of different database is complicated and the next couple of sentences will be densely packed full of information, but don’t say I didn’t warn you. To evaluate the measurements, a maximum likelihood principle methodology was adopted. In addition, the Bayesian Information Criterion (BIC) and/or Laplace’s approximation methodology has been adopted for model ranking. In cases where it is impossible to determine which model combination is overwhelmingly better than the others, we have opted to use a reliability analysis. Using this criteria, models are evaluated and ranked for each assigned region. The recommended database of Vs30 for each region is then presented in the form of an Excel spreadsheet. The data within that file can be easily imported into software like ArcGIS for viewing.