Skip to main content Skip to navigation

Lecture

Department of Mathematics & Statistics Post-Ostrom Lecture Colloquium: Dr. Randall LeVeque, Rapid Tsunami Forecasting Using Fake Quakes and Machine Learning

Spark
  • Colloquium 4:00-5:00 p.m., Spark 335 |
  • Open Discussion and Refreshments 2:00-3:30 p.m.,       Neill 216 (Hacker Lounge) |
  • Zoom Meeting ID: 946 8199 2093 |
  • Zoom Meeting Passcode: 300086
 
View location in Google Maps

About the event

Rapid Tsunami Forecasting Using Fake Quakes and Machine Learning

Abstract: Producing rapid real-time forecasts for tsunamis in the first few minutes of an earthquake is a challenging problem. Accurate forecasts often rely on direct measurements of the tsunami, which are only available at sparse locations, and only after the tsunami has passed the sensors. Real-time numerical modeling of the tsunami is also time consuming. We attempt to bypass these restrictions by training a convolutional neural network (CNN) that can forecast tsunami wave heights based only on Global Navigation Satellite System (GNSS) data, which is available within minutes at many existing stations in earthquake-prone regions. Training the model requires a large set of hypothetical earthquakes that are drawn from a geophysically reasonable probability distribution. These are generated using a Karhunen-Loeve expansion, as implemented in the FakeQuakes software. Tsunami simulations are performed for each using the GeoClaw software. Studying adversarial examples
designed to fool the model leads to some new insights on the robustness and stability of CNNs.

Speaker Bio: Dr. LeVeque has held appointments at the Courant Institute, UCLA, and ETH-Zürich, and has been on the faculty at the University of Washington since 1985. Dr. LeVeque enjoys teaching, and concentrates on classes in the fields of numerical analysis, partial differential equations, and nonlinear phenomena. He has written textbooks and lecture notes that are used at many universities. His research interests span many areas, including numerical analysis, computational fluid dynamics, nonlinear partial differential equations, mathematical theory of conservation laws, and software development, including the CLAWPACK software for solving conservation laws and other hyperbolic systems modeling wave propagation. He is also involved in research in many applications areas, including astrophysics, geophysics, and biophysics. Dr. LeVeque is a Fellow of SIAM, the Society for Industrial and Applied Mathematics, and a Fellow of the American Mathematical Society.

 

Contact