Detecting historic informal settlement fires with Sentinel 1 and 2 satellite data - Two case studies in Cape Town. (September 2019)

• Record Type: Journal Article
• Authors: Gibson, Lesley; Engelbrecht, Jeanine; Rush, David
• Is Part Of: Fire safety journal. Volume 108(2019)
• Publication Date: 2019-09
• Subjects: Informal settlements -- Remote sensing -- Fire -- Sentinel-1 -- Sentinel-2; Fire prevention -- Periodicals; Incendies -- Prévention -- Recherche -- Périodiques; Fire prevention -- Research; Periodicals; 628.92205
• DOI: 10.1016/j.firesaf.2019.102828
• Languages: English
• ISSNs: 0379-7112

Abstract

Abstract: Increasing global urbanisation is leading to a rise in the number of people living in informal settlements, challenging our ability to achieve sustainable development goals. As a consequence of high building density, inadequate building methods and flammable building materials, informal settlements are highly vulnerable to the devastating impacts of fire. Databases on historic fire occurrence, location and extent are scarce, especially in the Global South. This paper explores the potential for remote sensing technologies to fill this gap. Two case studies in Cape Town representing fire of different extent and build back characteristics, are used to demonstrate that Sentinel-1 and Sentinel-2 data can be used to detect known historic informal settlement fire. A pixel based approach applied to Sentinel-2 band 2 reflectance and Sentinel 1 backscatter and interferometry are highlighted. The concept of spatial autocorrelation is explored with both Sentinel-2 and 1 data showing that a 3 x 3 pixel standard deviation kernel and hotspot analysis can complement the pixel approach. Further research is required to test these methods within a time series change detection algorithm to identify unknown historic informal settlement fires. . Highlights: Informal settlement fires are characterised and the case for remote sensing to detect historic fires is proposed. A case study of two fires in Cape Town demonstrates the potential of Sentinel-2 and Sentinel-1. Spatial autocorrelation methods can assist through disruption of Tobler's Law and hotspot analysis. Settlement, fire and rebuild characteristics vary regionally and thus methods should be adapted to suit local conditions.