An Exploration of CHRIS PROBA Hyperspectral Imagery for Obtaining Burnt Area Cartography
Petropoulos, George1; Kontoes, Charalambos2

Accurate burnt area delineation in a rapid and cost-effective way can provide significant assistance in the development of protection and restoration policies as well as in the future monitoring of fire activity in an affected region. Nowadays, Earth Observation (EO) data, often combined with Geographical Information Systems (GIS) are widely used in different types of natural hazards including wildfires. The recent technological advances in remote sensors technology have resulted in the development of hyperspectral remote sensing systems, which, in contrast to multispectral data, are able to collect spectral information in a large number of discrete, narrow spectral bands. As a result, those systems are able to provide very rich spectrally information content that enhances dramatically our ability to accurately identify different targets on the Earth's surface.

The Compact High Resolution Imaging Spectrometer (CHRIS) is a hyperspectral instrument onboard PROBA platform operated from the European Space Agency (ESA). The sensor is able to collect spectral data in up to five different viewing angles and at 19 spectral bands (fully programmable) in the VNIR range (400 - 1050 nm) and at a spatial resolution of 17m. It can also be configured to provide 63 spectral bands at a spatial resolution of about 34 m.

The present study aims at exploring the use of CHRIS instrument in burnt area delineation when combined with different image processing approaches. As a case study it selected a major destructive fire occurred during the summer of 2007 in the Prefecture of Evoia in Greece located north-east of Athens, for which CHRIS PROBA imagery was acquired shortly after the fire suppression. Evaluation of the burnt area mapping estimates from CHRIS was based on the computation of the error matrix statistics. Also, the McNemar test was used to evaluate the statistical significance of the differences between the approaches tested. Burnt area estimates from CHRIS were also further validated against validated estimates from the Risk-EOS Burnt Scar Mapping service, developed in the framework of the GMES-SE programme (Global Monitoring for Environmental Security/Service Element) of ESA.

In overall, results from our study exemplified the appropriateness of the spatial and spectral resolution of CHRIS sensor and of its multi-look capability in obtaining rapid and cost-effective post-fire analysis, producing enhancing the thematic accuracy of mapping and advancing the thematic content of the fire recovery products compared to the ones generated in the framework of RiskEOS operational service. It appears that if hyperspectral technology is incorporated into the operational development of services for the estimation of burnt areas, the resulting combination should prove highly useful and additional information as far as the analysis of changes to land cover types due to natural hazards are concerned, therefore adding to the common Burn Scar Mapping product a new information layer with damage assessments directly derived from the use of hyperspectral EO data.

KEYWORDS: CHRIS, PROBA, burnt area mapping, hyperspectral, Risk-EOS, Greece.