Schedule
MINDS Seminar Series | Donsub Rim (Washington University in St. Louis) - Early Tsunami Warning using Convolutional Neural Networks
period : 2022-11-08 ~ 2022-11-08
time : 10:00:00 ~ 11:00:00
개최 장소 : Online streaming (Zoom)
Topic : Early Tsunami Warning using Convolutional Neural Networks
개요
| Date | 2022-11-08 ~ 2022-11-08 | Time | 10:00:00 ~ 11:00:00 |
| Speaker | Donsub Rim | Affiliation | Washington University in St. Louis |
| Place | Online streaming (Zoom) | Streaming link | ID : 688 896 1076 / PW : 54321 |
| Topic | Early Tsunami Warning using Convolutional Neural Networks | ||
| Contents | In a tsunamigenic earthquake event, predicting tsunami waveheights early in the event can aid emergency management. However, such prediction is a challenging problem: The tsunami wave propagates according to nonlinear partial differential equations with unknown initial conditions. In this talk, we will discuss a data-driven approach: we simulate random earthquakes and resulting tsunamis to form a training dataset, then train convolutional neural networks to make predictions. We show two recent results that focus on events originating from the Cascadia Subduction Zone and reach the Strait of Juan de Fuca. We first show that 30-60 minutes of waveheight measurements at the entrance of the Strait is sufficient to predict 6 hours of tsunami waveform in the interior of the strait. Next, we show that 4 minutes of seismic measurements from Global Navigation Satellite System (GNSS) is sufficient to make predictions of comparable accuracy. | ||
MinDS
·
2022-10-04 10:22 ·
Views 439