Measuring Spatio-temporal Landscape Pattern Change After Severe Flood with SAR and Machine Learning

Keywords: natural hazards, spatial pattern analysis, landscape metrics
Session Type: Virtual Paper Abstract
Day: Sunday
Session Start / End Time: 2/27/2022 09:40 AM (Eastern Time (US & Canada)) - 2/27/2022 11:00 AM (Eastern Time (US & Canada))
Room: Virtual 46

Abstract

Severe floods are becoming increasingly frequent and can have great influences on the landscape pattern of the stricken areas, some of which are short-term while others can last very long. They can further affect the configuration of different land cover patches, changing the ecological functionality of the area, and how species distribute and disperse. Therefore, this research aims to enhance the understanding of flood-induced landscape pattern change by creating a series of landscape pattern portraits of Henan Province, China, after it suffered from a severe flood event in July, 2021. I will map flooded areas with SAR data from Sentinel 1 every 6 days after the flood. Then I will identify patches of high ecological value (i.e., green spaces and protected areas) and corridors to analyze their change in quantity and configuration using GeoAI-based algorithms. Changes in landscape configuration can provide implications on changes in ecological functionality at different parts of the region.

This research can hopefully provide a highly automated and easily reusable workflow to map landscape pattern change that does not get affected by data limitation of optical satellite imagery which tend to have large cloud cover on flood days. There has been much more research on the effect of landscape pattern on flood but not the other way around. This research can fill the research gap. Both intensive short-term impacts and long-term impacts can be used to provide information on planning and management of the area for policymakers and practitioners.