Brightness Temperature, Reflectivity, Cloud Phase RGB, Fire Temperature RGB, Natural Colour RGB, Night Microphysics RGB, IR3.9
A massive wildfire outbreak in the second half of August impacted parts of northern Greece.
Last Updated
07 September 2023
Published on
01 September 2023
By Djordje Gencic and Ivan Smiljanic
Following three very warm and dry summer months, another round of wildfires happened in vicinity of Alexandroupoli in northeastern Greece.
During 21 and 22 August, the fire intensity grew rapidly under the influence of strong north-easterly winds. Fire fronts expanded rapidly outwards, creating a large burnt area in the centre. This is very much apparent from higher resolution imagery from the SLSTR and MODIS instruments.
The fire area was still growing after more than 10 days of non-stop fire burning, re-intensifying on 30 August under the influence of south-easterly winds. Fire dynamics are captured by a 10-day loop of combination of Natural Colour RGB and a Night Microphysics RGB (Figure 1).
Figure 1: Meteosat-11 Natural Colour RGB (daytime) and Night Microphysics RGB (nighttime) 10-day loop, 21 August 06:00 UTC-31 August 06:00 UTC
SEVIRI thermal imagery reveals very intense burning, apparent through saturation of IR3.9 channels, which starts to show artificial linear dark signals when sensing the hottest fires. This can be seen through comparison with more advanced instruments/imagery, eg SLSTR Fire Temperature RGB (Figure 2), which even outlines the fire fronts (not clear with SEVIRI), but also provides qualitative fire intensity information.
Figure 2: Comparison of the Meteosat-11 SEVIRI IR3.9 um (left) and Sentinel-3B SLSTR Fire Temperature RGB (right), 22 August 08:30 UTC
However, none of the images under Figure 2 provide a good smoke information. This is due to the fact that smoke becomes transparent when seen through longer wavelengths (near-IR and IR channels). For that reason, we show the product that uses both near-IR and visible (short wavelength), ie Cloud Phase RGB (Figure 3). The visible channel is placed on the blue RGB component, hence. smoke is seen in blue hues. The other two RGB components are made of near-IR channels, with most of the fires detected through more fire-sensitive NIR2.25 on green RGB beam, and less sensitive NIR1.6 channel on red. That way the gradient from lower to higher intensity fires is given in green to yellow shade gradient (green and red in RGB give yellow). Burnt ares are seen in green shades.
Figure 3: Fires, fire fronts, fire intensity and smoke extent seen through Cloud Phase RGB, Terra MODIS, 22 August
This unnatural view on fires and burnt areas, and also the surrounding ground, can be mitigated through swapping the red and green components of the standard Cloud Phase RGB (Figure 4). Thus, the fire intensity colour gradient becomes closer to what the human eye would expect (similar to Fire Temperature RGB). Also, burnt areas turn into more natural brown shades, surrounding ground into green shades, and smoke keeps similar blue hues.
Figure 4: Fires, fire fronts, fire intensity and smoke extent seen through 'Natural' Cloud Phase RGB, Terra MODIS, 22 August
Different clouds types, for which the Cloud Phase RGB was tailored, still keep the cloud phase and particle size information with this 'Natural' Cloud Phase RGB. Although, water and ice cloud swap the spectrum (depending on the particle size) of purple and blue shades, respectively. This is seen while looking at the Cu and Cb clouds, that developed over Sicily parallel to the Greece fire development (Figure 5).