Published June 6, 2024 | Version 1.0
Peer review Open

AI-driven rapid landslides mapping following the 2024 Hualien City Earthquake in Taiwan

Creators

  • 1. ROR icon British Geological Survey
  • 2. ROR icon University of Padua
  • 3. ROR icon State Key Laboratory of Geohazard Prevention and Geoenvironment Protection
  • 4. ROR icon University of Warwick

Description

This repository contains the co-seismic and pre-seismic landslide inventories from the Mw 7.4 earthquake that struck the eastern coast of Taiwan on April 2, 2024. The earthquake triggered numerous landslides and caused severe damage to infrastructure, providing a unique opportunity to test various automated landslide detection methods. In this news article, we utilized state-of-the-art Earth Observation (EO) techniques to create the first preliminary inventory of earthquake-induced landslides. We processed 13 Sentinel-1 (radar) images and 33 Planet and Google Earth Pro (optical) images covering approximately 3,300 km² of Eastern Taiwan using Convolutional Neural Networks and a Vision Transformer model. The inventory identifies approximately 8,700 landslide events with a co-seismic landslide area exceeding 75 km². More details can be found in the original publication.

Files

new_coseismic_landslide_area.zip

Files (5.1 MB)

Name Size Download all
md5:66aee1a4a2cad8e57821669cbf656c0c
2.3 MB Preview Download
md5:c18069eea9dab0134d0a83b10e414969
2.8 MB Preview Download

Additional details

Dates

Submitted
2024-06-07