Skip to main content Skip to main navigation

Publication

Effect of Terrain Information on Multimodel Deep Learning for Flood Disaster Detection

Takashi Miyamoto; Marco Stricker; J. Ogishima; Kevin Iselborn; Marlon Nuske; Andreas Dengel
In: 2023 IEEE International Geoscience and Remote Sensing Symposium (IGARSS). IEEE International Geoscience and Remote Sensing Symposium (IGARSS-2023), June 16-21, Pasadena, CA, USA, IEEE, 2023.

Abstract

The utilization of multimodal analysis techniques, combining satellite imagery with terrain information, has gained promi- nence in fl ood detection. This study focuses on the inclu- sion of elevation data into the Sen1fl oods11 dataset, an open dataset for fl ood damage detection, to investigate the influ- ence of terrain information on fl ood detection tasks. Among the considered terrain information, the inclusion of elevation data resulted in a bias of overestimating the presence of water in relatively low-lying areas, without contributing to accuracy improvement. However, the utilization of slope information, derived from differentiating the elevation data, mitigated such bias and yielded a slight improvement in accuracy. This fi nd- ing aligns with the utilization of derivatives in physical equa- tions describing fl ood fl ow, suggesting the explicit incorpo- ration of physics-based principles, such as the fl ow of water based on slope, to enhance model accuracy in future research endeavors.

Projekte

Weitere Links