Illustration of SAFNWC/MSG v2011 cloud detection improvement over cold ground at night

Illustration of SAFNWC/MSG v2011 cloud detection improvement over cold ground at night

The evolution in Cloud Mask (PGE01) algorithm at night over cold ground is illustrated for a winter situation in February 2011. This case illustrates that such situations may involve wide areas and be settled for a long period. Of course the cloud mask behaviour is very sensitive to the Numerical Weather Prediction model used as their forecated surface temperature may differ. The v2010 Cloud Type obtained on Meteosat-9 (MSG2) at 02h45 UTC on 22 february 2011 is displayed below together with SYNOP observations. A wide area over Russian federation and Kazakhstan where clear-sky ground based observations are reported in many SYNOP is wrongly depicted as a high thick cloud deck or mid-level clouds. In this area, the reported air temperature is between -35°C and -15°C, air is dry and visibility is generally good (never less than 4 km). Another area at the same hour over Poland, Lithuania and Belarus is also wrongly depicted as cloudy. Air temperature reported there, between -15°C and -22°C, is lightly warmer. Light haze may be sometimes observed.

	Not processed Clear land Clear sea Snow land Sea ice Very low Low Mid. level A High opaque Very high opaque Thin semi transparent Meanly thick semi transparent Thick semi transparent Semi transparent above Fractional Undefined
	Not processed Clear land Clear sea Snow land Sea ice Very low Low Mid. level High opaque Very high opaque Thin semi transparent Meanly thick semi transparent Thick semi transparent Semi transparent above Fractional Undefined

When processed at daytime on 22 February CT depicts that in general ground is snow covered for these two areas as illustrated below on the image produced for 08h45 UTC.

The difference between V2011 and V2010 is emphasized by the animation of the two CT at 02h45.

Such a situation is also observed on 23 and 24 february 2011. We analyse again the cloud cover as reported by SYNOP together with V2010 CT on 24 february. Obviously one can observe again persisting false cloud detections with V2010.

	Not processed Clear land Clear sea Snow land Sea ice Very low Low Mid. level High opaque Very high opaque Thin semi transparent Meanly thick semi transparent Thick semi transparent Semi transparent above Fractional Undefined
	Not processed Clear land Clear sea Snow land Sea ice Very low Low Mid. level A High opaque Very high opaque Thin semi transparent Meanly thick semi transparent Thick semi transparent Semi transparent above Fractional Undefined
	Not processed Clear land Clear sea Snow land Sea ice Very low Low Mid. level A High opaque Very high opaque Thin semi transparent Meanly thick semi transparent Thick semi transparent Semi transparent above Fractional Undefined

An animation loop between 23/02/2011 14h00 UTC and 24/02/2011 08h45 UTC shows that the v2011 cloud detection is better matched with SYNOP reports than V2010 CT. The SEVIRI spectral information behaviour is also illustrated by images of brightness temperature difference T108-T39 (scale[-10K,6K]) and T108-T87( scale[-2K,4K]). The false detection of clouds that occured rapidly with the sun set in V2010 CT is not observed any more when V2011 algorithm is used. Moreover we can note that the detection of low clouds in not hampered by the relaxation of the thresholds (for instance the fog over Vistula between the Gulf of Gdansk and Wloclawek is caught). Some high clouds may sometimes be missed, but their area remains limited in the fringe of thin semi-transparent high clouds.

Not processed

Clear land

Clear sea

Snow land

Sea ice

Very low

Low

Mid. level

High opaque

Very high opaque

Thin semi transparent

Meanly thick semi transparent

Thick semi transparent

Semi transparent above

Fractional

Undefined

For instance, if we analyse in details the image at 02h00 we can see that the fog formation above Vistula between the Gulf of Gdansk and Wloclawek is caught by V2011. We can also note that there is still place for improvement to remove false cloud detections such as over sea ice on the Baltic sea, or over frozen lakes as over lake Balaton in Hungary