Utilization of Hyperspectral Remote Sensing Imagery for Improving Burnt Area Mapping Accuracy

Abstract

Disastrous wildfires, such as the ones occurring in New South Wales / Australia in 2019/20 or in California / USA in 2020, have gained wide recognition in the global media. Besides obvious effects, such as the loss of lives and property, wildfires also affect the global climate through carbon dioxide (CO2) emissions. Monitoring wildfires using satellite-based Earth Observation (EO) imagery has become an essential tool for wildfire management and mitigation. The assessment of burnt areas through remote sensing imagery is usually based on a combination of information of either the red / NIR (near infrared) or the NIR / SWIR (short wave infrared) domain. This is done by utilizing a normalized difference index such as the NDVI (Normalized difference vegetation index), or the NBR (Normalized Burn Ratio), often in their differential variants. The wavelengths of the available spectral bands are, however, determined by the utilized sensor, and may not be optimal for the specific task of burnt area derivation. Hyperspectral data, however, allows the utilization of specific wavelengths best suited for a given task. The presented study analyses to what extent an automatic burnt area processing chain can profit from the additional spectral information available in data of the DESIS hyperspectral instrument.