Thin dust cloud/veil from field/agricultural land. Credit: Александра Замулина

Dust over the Canaries

22-27 February 2020, Various times

Thin dust cloud/veil from field/agricultural land. Credit: Александра Замулина
Thin dust cloud/veil from field/agricultural land. Credit: Александра Замулина

Record dense dust over the Canary Islands originates near the African coast.

Last Updated

21 February 2023

Published on

21 February 2020

By Jose Prieto and Vesa Nietosvaara (EUMETSAT), Andreas Wirth (ZAMG) and HansPeter Roesli (Switzerland)

An intrusion of fine Saharan dust covered the Canary Islands on 22-23 February 2020, causing general disruption in air traffic. This dust event was the worst this century, in terms of PM2.5 concentrations, for some of the islands.

 Meteosat-11 Dust RGB with 925hPa streamlines overlaid, 23 Oct 23:00 UTC
Figure 1: Meteosat-11 Dust RGB with 925hPa streamlines overlaid, 23 October 23:00 UTC

Both Meteosat-11 and GOES-16 offered a seamless temporal view of the dust evolution, after it was raised from the African ground to about 1km height during its course westwards of the source. Figure 2 is an animation of the Meteosat-11 Dust RGB superimposed with 925hPa streamlines, that shows the dust (in magenta) as it moved to higher levels.

Figure 2: Meteosat-11 infrared Dust RGB, 22 February 01:00 UTC–23 February 23:00 UTC

Sentinel-3A, through its solar ocean colour instrument OLCI (True Colour RGB at 0.86µm, 0.56µm and 0.44µm), also pinpoints the geographical origin of the dust, near the African coast (see Figure 3). With Sentinel-3's high horizontal resolution of 300m, the dust could be seen dust rising from the ground under cloud, then reaching cloud-free areas.

The dust appeared in yellow hues, which is easier to distinguish over the ocean.

 Sentinel-3A OLCI True Colour RGB, 10:40 UTC
Figure 3: Sentinel-3A OLCI True Colour RGB, 10:40 UTC

The dust helped visualise the wind dynamics around the Canary Islands, usually hidden from satellites. The short High Resolution Visible animation (Figure 4) showed gravity waves in the leeward air, caused by the islands' orography. In particular, a wake-like wave interference between Gomera and Tenerife, and the rings around Teide mountain in Tenerife, were obvious.

Figure 4: Meteosat-11 HRV, 23 February 14:27–17:27 UTC

The infrared-window, or Dust RGB, showed slight differences for GOES and Meteosat, due to the channel widths and centres, plus the ampler option for window channels in GOES.

Figure 5 compares Meteosat and GOES, in terms of perspective and channel width differences. The GOES signal for dust (magenta) shows a higher contrast than Meteosat due to the longer optical depth. The second example on Figure 5b shows a surprising hue shift from magenta to peach colour, due to different optical depths for both satellites, plus possibly to using three channels around 11µm in GOES rather than two of Meteosat.

 Comparison of Meteosat-11 Dust RGB (right) with GOES-16 Dust RGB (left), 23 Feb 14:45 UTC.
Figure 5a: Comparison of Meteosat-11 Dust RGB (right) with GOES-16 Dust RGB (left), 23 February 14:45 UTC
 
 Another comparison of Meteosat-11 Dust RGB  with GOES-16 Dust RGB, 27 Feb 10:40 UTC.
Figure 5b: Comparison of Meteosat-11 Dust RGB with GOES-16 Dust RGB, 27 February 10:40 UTC
 

A night inversion may be gleaned from the Night Cloud Microphysics RGB (NCMP) of Meteosat-11 on 22 February at 05:00 UTC (Figure 6, centre panel). Figure 6 shows a grey fog or low-cloud patch further south along the coast (yellow arrow in middle pane. While the Dust RGB of Meteosat-11 (left) marked this patch, GOES-16 (right) showed no signs of it. On the other hand, other fog or low cloud between Gibraltar and Morocco (top right corner of images) showed off quite well on both images.

Also, both Dust RGB images illustrate the change in the blue shades over the circulation. It appeared in GOES-16 (green arrows in right panel) but not in Meteosat-11. This may be explained by the Dust RGB scheme used for GOES-16, which takes advantage from the availability of two IR-window bands around 11µm.

 Dust and NCMP composites from Meteosat-11 (left and centre) and GOES-16 (right panel), 22 Feb 05:00 UTC.
Figure 6: Dust and NCMP composites from Meteosat-11 (left and centre) and GOES-16 (right panel), 22 February 05:00 UTC.

There were two main dust waves, on 22 and 23 February. Both started in the morning after the dissipation of the night inversion under a low-level jet, which is a usual mechanism. The second dust release reached far out into the Atlantic and, encountering a deformation zone, part of it moved back eastward. This part arrived over the northern part of the Iberian peninsula on 27 February (where it moved off the FOV of GOES-16) (see the dust\s progress in Figure 7). The other parts of the dust circulated anti-clockwise around the low pressure system. More dust was injected on 24 February from the coast of Senegal and Mauritania, which remained mostly hidden by cloud.

Figure 7: GOES-16 Dust RGB, 22 February 06:00–27 February 12:00 UTC

By Tuesday 25 February the dust had spread southwards and mixed with condensation to form cloud. Figure 8 is a comparison of the standard Dust RGB from GOES with its channel at 2.24µm, which generally enables cloud or aerosol particle size estimation.

Image comparison

GOES ABI channel 2.2µm compare1
compare2
 

Figure 8: Comparison of GOES Dust RGB images, 25 February 12:00 UTC.

Alternative image recipes

Introducing solar channels in the standard recipe adds value and colour to the Dust RGB. The result pinpoints details, in this case of the structure of the dust cloud. It can only be used during the day however, which is a high price to pay. Two partly-solar alternatives are analysed: (a) solar information on the green beam (Figure 9, left) and (b) solar on green and blue (Figure 9, right). The ABI Channels used were 15-13, 5-6 and 13-0 in the left-hand image and 15-13, 5-6 and 3-1 in the right-hand image.

 

Image comparison

RGB with solar information on green and blue compare1
compare2
 

Figure 9: Comparison of two experimental mixtures of solar and infrared sources in the same RGB, 23 February 14:45 UTC.

The difference 1.6 µm–2.2µm at the green beam does not generate a good contrast between dust and clear areas (just 4% reflectivity in the difference, vertical axis on Figure 10). The small pixel differences (around 0.6K) between infrared channels, used as blue beam, indicate low flying dust.

Including the blue beam solar information from the channel at 0.55µm (left in Figure 9), creates a more detailed image for the dust distribution. The contrast of dust with ice cloud is now clear, but not if dust is compared with the oblique view of the clear ocean.

In a contrasted close-up of the GOES image (Figure 11, the orographic blocking effect on dust downstream of the islands was clear for Tenerife (T), and also Hierro and Gomera — pointing to low level flying aerosol.

 Scattering graph for pixel values from the infrared (x-axis) and solar (y-axis) for clear ocean and floating dust above
Figure 10: Scattering graph for pixel values from the infrared (x-axis) and solar (y-axis) for clear ocean and floating dust above
 
 Colour enhancement on the Dust RGB from GOES image, 23 Feb 14:45 UTC
Figure 11: Colour enhancement on the Dust RGB from GOES image, 23 February 14:45 UTC

The ocean surface reflectivity was lower in the wake of the islands, where it was sheltered from the wind. Northwest of Gran Canaria (GC) there was a cold cloud development of -45°C, but it did not interfere with the dust crossing below.

 
 

Additional content

Blowing dust across the Canary Islands and Atlantic Ocean (CIMSS Blog)
El episodio de calima en Canarias a vista de satélite (AEMET, in Spanish)
Sandstorm Wreaks Havoc in Canary Islands (New York Times)

COPERNICUS LOGOS