Geographia Technica, Vol 14, Issue no.2/2019, pp. 20-38

EVALUATION OF BURNED AREAS WITH SENTINEL-2 USING SNAP: THE CASE OF KINETA AND MATI, GREECE, JULY 2018

Kamill Daniel KOVÁCS (Cluj-Napoca, Romania)

DOI: 10.21163/GT_2019.142.03

ABSTRACT: The aim of this paper is to compare the two largest forest fires that occurred in Greece in July 2018 using metrics for burned area and burn severity mapping, derived only from free satellite data. Sentinel-2 satellite images of the European Space Agency (ESA) within the Copernicus program provide a spatial resolution of 10 m, which facilitates more accurate monitoring of environmental phenomena such as forest fires. The processing of the satellite images and the calculation of the metrics was performed using SNAP software, which is an open-source software developed by ESA. The mapping of the obtained results was performed in the QGIS software, which is also an open-source software. The delimitation of the burned area and the classification of the severity of both wildfires was performed using the Relativized Burn Ratio (RBR) satellite index. These results were contrasted with the Copernicus Emergency Management Service (EMS) maps related to these two events. Our results obtained in relation to the size of the burned area show smaller affected areas than the Copernicus Emergency Management Service maps. This is explained by the different methods used in the delimitation of the burned areas. In the case of Mati’s wildfire the EMS has created the thematic layer by means of visual interpretation using post-event satellite image and in the case of Kineta’s wildfire was applied a semi-automatic approach. Moreover, in this study is proposed and evaluated a new burn severity metric, the burned vegetation index (BVI) which shows where the most significant changes in healthy vegetation occurred. This new index was compared with RBR, dNDVI and dNBR using statistical correlation. The results indicate that BVI shows better the burned vegetation and its statistical correlation with RBR is significant (R2 = 0.92).


Keywords: burned area mapping, burn severity, RBR, BVI.

Full article here