Thermal Anomaly Level Algorithm for Active Fire Mapping by Means of Sentinel-2 Data

In this study, we propose a new approach to detect and map active fire area by means of Sentinel-2 data. Sentinel-2 Multi-Spectral Instrument (MSI) is comprised of visible spectral bands, near infrared (NIR), and short wave infrared (SWIR) spectral bands. These spectral bands, band 8A (865 nm), band 11 (at 1.61 μm), and band 12 (at 2.19 μm), are employed to roughly approximate the temperature within the forest fire range (400 - 1500°C) using Planck's blackbody radiation law. The Normalised Difference Indices (NDI) of band 11 with 8A and band 12 with 11 are formulated, namely NDI1 and NDI2, respectively. The criteria of NDI1, NDI2, and band 12 ranges are estimated empirically to identify forest fire area into four thermal anomaly levels, i.e., high-temperature crown fire, typical crown fire, surface fire or smoldering area and remains of the fire. For the high-temperature crown fire area, band 12 value is very high, all indices are positive, and the NDI2 is lower than NDI1, following Planck's curve for temperature higher than 1200 K. For the typical crown fire area, the band 12 value is high, and the NDI2 can be higher than NDI1. For the smolder area, the band 12 value is in the middle range, and the NDI2 must be larger than 0.2. Finally, for remains area, the band 12 value is in the lower range, and NDI1 must be higher than -0.27. This work shows that the active fire area can be identified and mapped into four thermal anomaly levels. This approach can be employed for rapid fire management as soon as the image data is obtained.