Razvoj metode bi-normaliziranog omjera opožarenosti na događaju katastrofalnog šumskog požara 14. srpnja 2022. godine kod Vodica, Hrvatska

Abstract

The use of multispectral sensors, such as MSI, on the Sentinel-2 satellite for rapid assessment of the area and the burning degree after large forest fires utilises standard method of determining the Normalized Burn Ratio (NBR) before and after the fire, from which the difference of the Normalized Burn Ratio (dNBR) is calculated. Applied to areas with sparse vegetation, the standard method of NBR gives poor results. To cope with low vegetation areas, a so-called “relativizing dNBR” was introduced where an empirical expression divides the difference of the NBRs in two channels by a quotient slightly greater than the pre-fire NBR. In this paper, a forest fire near Vodice, Croatia, which occurred on July 14, 2022, and which grew into a catastrophic fire that left about 30 km2 of burnt area, was studied. As this is an area of Mediterranean vegetation, relatively sparse, instead of the relativization, an attempt was made to introduce a new formula for calculating the normalized ratio of burning, which included two differences in spectral radiances (instead of one). In addition to the standard difference in the near-infrared and short-wave infrared part of the spectrum, a difference between the near-infrared and the green visible part of the spectrum was used simultaneously. These new, bi-Normalized Burn Ratio, retain information about the spectral radiance in the green visible band and react to surfaces where at least a little vegetation cover remained after the fire, which ultimately gives lower burn categories, with the same burned area detected compared to the standard method.