[ghsc-seminars] GHSC Seminar *HYBRID* - Weiwei Zhan, UT Austin, January 31, 2023

[Llenos, Andrea L]

Updating Global Geospatial Liquefaction Models: with Focus on Uncertainty Quantification and Feature Engineering

Speaker: Weiwei Zhan
                 Texas Advanced Computing Center
          University of Texas Austin
Location: HYBRID (online and in-person at 1711 Illinois St, Golden, CO, Rm 204)
                 *Non-USGS visitors are welcome! Please bring a photo ID and plan to arrive ~15 min early to allow
                   time to go through building security.*
Date/Time: Tuesday, January 31, 2023 10-11am MST

Abstract: Global geospatial liquefaction models take advantage of broadly available geospatial proxies to predict liquefaction hazards. It remains less explored on how to quantify uncertainties for such geospatial natural hazard models. The first part of this presentation will share an uncertainty quantification framework that consists of uncertainty source characterization, sensitivity analysis, and forward uncertainty propagation. We have identified three sources of uncertainty in such predictive models as parametric estimation uncertainty, modeling error, and geospatial input uncertainty. A Bayesian inference algorithm is used to quantify model parameter estimation uncertainties which are found to be negligible when a large amount of data is used in the parameter estimation process. Modeling errors are characterized based on the observed residuals between model predictions and observations and by fitting a normal distribution to the liquefaction probability residuals. The geospatial input uncertainties are characterized using the literature and expert judgment and propagated into model output. Then, we investigate the sensitivity of model output to different uncertain inputs and find that the variance of model output is largely controlled by the geospatial input uncertainties and modeling errors. Last, we propose an approximate forward uncertainty propagation method, which provides comparable results to a Monte Carlo simulation-based method with better computational efficiency. The proposed UQ framework provides a measure of uncertainty on model predictions and can be applied to any logistic regression models and other geospatial modeling problems.

The second part of this presentation will show our updates on geospatial liquefaction hazard modeling with a focus on feature engineering. Multiple feature engineering techniques are employed to find the best combinations of geospatial features for predicting liquefaction occurrence, including the feature filter that removes low-correlated and redundant features, exhaustive feature selection that creates and evaluates all possible feature combinations, and principal component analysis (PCA) that reduces the dataset dimensionality while preserves the maximum amount of information. The results suggest that the exhaustive feature selection method achieves less information loss and better interpretability than the PCA method for this case. We conclude that such feature engineering methods could be applied to other geospatial modeling tasks that can benefit from finding the optimal feature combinations.

Bio: Dr. Weiwei Zhan is a postdoctoral fellow in the Texas Advanced Computing Center at the University of Texas at Austin. Previously, he worked as a postdoc in the Geohazards Research Lab at Tufts University. He earned his Ph.D. degree in Civil Engineering from Clemson University and bachelor’s degree in Engineering Geology from Chengdu University of Technology. His research interests lie primarily in integrating multidisciplinary data, machine learning, and computational geomechanics to understand mechanisms of and develop predictive models for multiple natural hazards, such as liquefaction, landslide, and earthquake.

For questions, comments, or speaker suggestions, please contact the GHSC Seminar Committee: Andrea Llenos <allenos at usgs.gov>, Benjamin Murphy <bmurphy at usgs.gov>, Lauren Schaefer <lschaefer at usgs.gov>


________________________________________________________________________________
Microsoft Teams meeting
Join on your computer or mobile app
Click here to join the meeting<https://gcc02.safelinks.protection.outlook.com/ap/t-59584e83/?url=https%3A%2F%2Fteams.microsoft.com%2Fl%2Fmeetup-join%2F19%253ameeting_ZWI3OWFlOTQtZTVkZS00OGIzLTg3MzUtMThkMTgwMDI0Zjc5%2540thread.v2%2F0%3Fcontext%3D%257b%2522Tid%2522%253a%25220693b5ba-4b18-4d7b-9341-f32f400a5494%2522%252c%2522Oid%2522%253a%252250f012ff-efbf-4efc-b34b-d5361c4364e5%2522%257d&data=05%7C01%7Cghsc-seminars%40geohazards.usgs.gov%7C8a13d89e66984dd8939008daffe336ed%7C0693b5ba4b184d7b9341f32f400a5494%7C0%7C0%7C638103648579984872%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C3000%7C%7C%7C&sdata=3W8gW1Q%2BAzLPO%2B5%2BQ5NWUQcwTznNj1eNMcnOtA%2BqmUM%3D&reserved=0>
Meeting ID: 276 852 793 971
Passcode: F6GHjw
Download Teams<https://gcc02.safelinks.protection.outlook.com/?url=https%3A%2F%2Fwww.microsoft.com%2Fen-us%2Fmicrosoft-teams%2Fdownload-app&data=05%7C01%7Cghsc-seminars%40geohazards.usgs.gov%7C8a13d89e66984dd8939008daffe336ed%7C0693b5ba4b184d7b9341f32f400a5494%7C0%7C0%7C638103648579984872%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C3000%7C%7C%7C&sdata=cNCqUFIIGfKLvK%2FJHiBuTQTwlOK9MF43psFBvFt%2FA6w%3D&reserved=0> | Join on the web<https://gcc02.safelinks.protection.outlook.com/?url=https%3A%2F%2Fwww.microsoft.com%2Fmicrosoft-teams%2Fjoin-a-meeting&data=05%7C01%7Cghsc-seminars%40geohazards.usgs.gov%7C8a13d89e66984dd8939008daffe336ed%7C0693b5ba4b184d7b9341f32f400a5494%7C0%7C0%7C638103648579984872%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C3000%7C%7C%7C&sdata=9kKCibpILec1Z%2BfyoWrpWPOCJPEUz0QpYFEQewzG8w0%3D&reserved=0>
Learn More<https://gcc02.safelinks.protection.outlook.com/?url=https%3A%2F%2Faka.ms%2FJoinTeamsMeeting&data=05%7C01%7Cghsc-seminars%40geohazards.usgs.gov%7C8a13d89e66984dd8939008daffe336ed%7C0693b5ba4b184d7b9341f32f400a5494%7C0%7C0%7C638103648579984872%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C3000%7C%7C%7C&sdata=EVpx0lB1hZjqxwvZYAD35lQ9gkCQNTyVPi%2BzTPh%2BvRg%3D&reserved=0> | Meeting options<https://gcc02.safelinks.protection.outlook.com/?url=https%3A%2F%2Fteams.microsoft.com%2FmeetingOptions%2F%3ForganizerId%3D50f012ff-efbf-4efc-b34b-d5361c4364e5%26tenantId%3D0693b5ba-4b18-4d7b-9341-f32f400a5494%26threadId%3D19_meeting_ZWI3OWFlOTQtZTVkZS00OGIzLTg3MzUtMThkMTgwMDI0Zjc5%40thread.v2%26messageId%3D0%26language%3Den-US&data=05%7C01%7Cghsc-seminars%40geohazards.usgs.gov%7C8a13d89e66984dd8939008daffe336ed%7C0693b5ba4b184d7b9341f32f400a5494%7C0%7C0%7C638103648579984872%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C3000%7C%7C%7C&sdata=SD4Y3%2BgotmkLmYBxwHPw82lUuScyjD6dGL4BCd6%2Fj3c%3D&reserved=0>
________________________________________________________________________________
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://geohazards.usgs.gov/pipermail/ghsc-seminars/attachments/20230126/9868abbb/attachment-0001.html>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: text/calendar
Size: 10377 bytes
Desc: not available
URL: <http://geohazards.usgs.gov/pipermail/ghsc-seminars/attachments/20230126/9868abbb/attachment-0001.ics>