M9 histogram

Smoke from California fires above Europe seen by NOAA-20

12 September 2020 11:37-13:07 UTC

M9 histogram
M9 histogram

Smoke from California fires above Europe on 12 September 2020, as seen in the NOAA-20 VIIRS 1.38 micron band, at 11:37 UTC.

Last Updated

26 February 2021

Published on

24 February 2021

By Martin Setvák (CHMI), Jochen Kerkmann and Ivan Smiljanic (EUMETSAT)

The 1.38 µm spectral band is primarily known for its power to detect very thin cirrus, and also some of the higher aerosol/ash plumes. Both, thin cirrus and aerosol plumes are typically located within the darkest areas of the scenes in this band, which makes their efficient visualisation somewhat difficult — either in single 1.38 µm band images, or, for example, in the RGB Cloud Type, in which the band is used as the red component. Attempts to clearly show the darkest parts of the images in this band (typically between 0.5 to 2.5 % reflectivity) lead to oversaturation of the brighter image areas, such as tops of convective storms or any other opaque high clouds.

The pseudo True Color RGB (Figure 1), in this case composed of VIIRS bands M5, M4, and M1 (in which the aerosol band appears somewhat better than in the traditionally used M3 band) shows an optically thin, but horizontally wide aerosol plume crossing France, Germany, Poland and Baltic states. By origin, this aerosol is a smoke plume, generated by fires in California.

NOAA-20 True Color 12 Sept 2020
Figure 1: NOAA-20 True Color, 12 September 11:37 UTC.

Now let’s see how this smoke plume appears in the M9 band. The first of the images shows the M9 reflectivity range from 0 to 5% (using standard linear enhancement) (Figure 2), and the plume is almost invisible here.

 

M9 linear 0-5
Figure 2: M9 band reflectivity range from 0 to 5%, 12 September 11:37 UTC.

The plume begins to be better visible at reflectivity range 0–2.5% (Figure 3), and gains on visibility at even lower reflectivity ranges, 0–1% (Figure 4), 0–0.75% (Figure 5), and namely at 0–0.5% (Figure 6). However, all other high clouds in the area are oversaturated at these enhancements.

Figure 3: M9 reflectivity range from 0–2.5%
Figure 3: M9 band reflectivity range from 0–2.5%, 12 September 11:37 UTC.
Figure 4: M9 reflectivity range from 0–1%
Figure 4: M9 band reflectivity range from 0–1%, 12 September 11:37 UTC.
Figure 5: M9 reflectivity range from 0–0.75%
Figure 5: M9 band reflectivity range from 0–0.75%, 12 September 11:37 UTC.
Figure 6: M9 reflectivity range from 0–0.5%
Figure 6: M9 band reflectivity range from 0–0.5%, 12 September 11:37 UTC.

One of the ways out how to show well the darkest parts of this band, without sacrificing structure of everything else, is use of the histogram equalization enhancement (Figure 7) or piecewise linear enhancement (Figure 8). Principle of these enhancements is shown in (Figure 9). Figure 10 shows an example of the histogram equalization method when applied to Cloud Type RGB, to its red (M9) component.

 

 

Figure 7: M9 rhistogram equalization enhancement
Figure 7: M9 band histogram equalization enhancement, 12 September 11:37 UTC.
Figure 8: M9 piecewise linear enhancement
Figure 8: M9 band piecewise linear enhancement, 12 September 11:37 UTC.
Figure 9: M9 principle of histogram equalization or piecewise linear enhancement
Figure 9: M9 band principle of histogram equalization or piecewise linear enhancement
Figure 10: M9 example of the histogram equalization method when applied to Cloud Type RGB
Figure 10: M9 band example of the histogram equalization method when applied to Cloud Type RGB, 12 September 11:37 UTC.

A confirmation that the plume is indeed an aerosol plume, detected by CALIOP instrument of the CALIPSO satellite, is shown in Figure 11.

M9 CALIPSO
Figure 11: M9 band and CALIPSO comparison.

For a reference, notice how smoke is basically transparent for Meteosat satellite, unless seen by the very slant angle (using Meteosat-8 at 41.5° E) at the slant Sun angle (evening), with strong forward scattering process. Figure 12 shows Natural Colour RGB product which also had to be very much enhanced (individual RGB components’ range only 0-30% reflectivity) in order to detect smoke (in faint cyan/blue shades) at late evening hours.

Figure 12: Smoke seen at slant angle of, both Sun and satellite. Meteosat-8 (IODC) Natural Colour RGB (enhanced to 0-30% reflectivity) animation on 12 September 2020, 06:00–18:00 UTC, 15 min time step.