Soil liquefaction poses a significant risk to distributed and critical infrastructure systems. Although liquefaction vulnerability is controlled by geologic and groundwater conditions, its regional assessment is almost exclusively based on geologic material properties, which introduces empirical uncertainty. In this talk, I show a multivariate methodology that incorporates groundwater conditions by introducing hydrological variables in a regional assessment of liquefaction hazards. I demonstrate the methodology for the 2010 El Mayor-Cucapah earthquake in Imperial County, for which I used field reconnaissance data to train a logistic regression model, and obtained probabilistic maps of liquefaction occurrence. Results indicate that the proposed approach may improve the modeling of regional liquefaction assessment. However, in order to draw quantitative conclusions, increased data resolution is necessary which may be possible by downscaling regional datasets. Downscaling may be possible by leveraging correlations between remotely sensed terrain data and geological and hydrological datasets. To that effect, I show a machine learning technique where I leverage spatial correlations to bridge microscopic and macroscopic length scales in the context of granular systems. I close by suggesting that such a technique may be repurposed to generate synthetic data and downscale regional datasets, eventually resulting in more accurate mapping of regional liquefaction hazard.
About the Speaker: Utkarsh Mital (Postdoctoral Scholar, California Institute of Technology (Caltech))
Utkarsh Mital is a Postdoctoral Scholar in Mechanical and Civil Engineering at California Institute of Technology (Caltech). He is interested in modeling problems in geomechanics and earth sciences using principles of probability and advances in machine learning. He obtained his Ph.D. in Applied Mechanics, also at Caltech, where his focus was on understanding the micro- and macro-mechanics of soil liquefaction. He also holds a Master’s degree in Applied Mechanics from Caltech and a Bachelor’s degree in Civil Engineering from V.J.T.I. (India).
Hosted By: Dipankar Dwivedi