United States Department of Transportation
i
Informing Predictions From Above With Data From Below: An AI-Driven Seismic Ground Failure Model for Rapid Response and Scenario Planning
-
2022-06-01
[PDF-8.20 MB]
Details:
-
Creators:
-
Corporate Creators:
-
Corporate Contributors:
-
Subject/TRT Terms:
-
Publication/ Report Number:
-
Resource Type:
-
Geographical Coverage:
-
Edition:Final Project Report
-
Corporate Publisher:
-
Abstract:Geospatial models for predicting soil liquefaction infer subsurface traits via satellite remote sensing and mapped information, rather than directly measure them with subsurface tests. This project developed geospatial models that are driven by algorithmic learning but pinned to a physical framework, thereby benefiting from both machine and deep learning, or ML/DL, and the knowledge of liquefaction mechanics developed over the last 50 years. With this approach, subsurface cone penetration test (CPT) measurements are predicted remotely within the framing of a popular CPT model for predicting ground failure. This has three potential advantages: (i) a mechanistic underpinning; (ii) a significantly larger training set, with the model principally trained on in-situ test data, rather than on ground failures; and (iii) insights from ML/DL, with greater potential for geospatial data to be exploited. Models were trained using ML/DL and a modest U.S. dataset of CPTs to predict liquefaction-potential-index values via 12 geospatial variables. The models were tested on recent earthquakes and were shown—to a statistically significant degree—to perform as well as, or better than, the current leading geospatial model. The models are coded in free, simple-to-use Windows software. The only input is a ground-motion raster, downloadable minutes after an earthquake or available for countless future scenarios. To demonstrate model use, ground failure probabilities were computed for 30 ground-motion simulations of a magnitude 9, Cascadia Subduction Zone earthquake, at every bridge site in Washington state. These analyses, as further detailed and mapped herein, indicated that: 13 bridges had at least a 70 percent probability of ground failure; 218 bridges had at least a 60 percent probability of ground failure; and 795 bridges had at least a 50 percent probability of ground failure. These analyses did not consider specific asset designs or site-specific ground improvements. The analyses did, however, provide a ranked list of bridge sites most likely to be damaged by ground failure. Select ground-truthing lent credence to the developed models and results.
-
Format:
-
Funding:
-
Collection(s):
-
Main Document Checksum:
-
Download URL:
-
File Type:
