Change Detection Applying Landscape Metrics on High Remote Sensing Images

Abstract

The Haiti earthquake produced a big effort in the remote sensing data collection. Many international surveying organizations were involved to collect, process, and use the spatial information to rescue the population. Here, a change detection methodology based on landscape metrics, derived by the landscape ecology and implemented in FRAGSTAST software is presented. Some of these metrics were applied to Pre-event and Post-event classified images (QuickBird and Ikonos). The attention was focused on ten test areas, characterized by different spatial patterns and urban structures. The land-cover classification (Urban and Other) was provided by an unsupervised K-Means algorithm in ILWIS open source software. The final results and their temporal comparison have allowed localization of the earthquake damages and the occurred changes, and also analyzing the fragmentation of the urban morphology. The GIS management of the information has improved the thematic map production, with useful results very for damage assessment.