Fog detection in cold winter situations

Fog detection in cold winter situations

01 November 2006 00:00 UTC

Fog detection in cold winter situations
Fog detection in cold winter situations

Fog detection in cold winter situations using MSG.

Last Updated

21, October 2020

by Anna Eronn and Jochen Kerkmann (EUMETSAT)

On 1 November 2006 a deep low pressure system (979 hPa at 00 UTC, centred over the Gulf of Finland) moved eastward towards western Russia. After the low pressure a ridge of high pressure built over Norway and a cold northerly wind spread down over Sweden and Finland.

In this situation, northern and central Sweden was covered with fresh snow and from northern Sweden 2 m temperatures as low as minus 26 °C were reported. The Meteosat-8 images below show the weather situation over Scandinavia at 04:00 UTC (night-time) when patches of fog/low stratus were forming in the cold air over northern Scandinavia and Finland.

The most common way to detect fog/low stratus at night is the use of the brightness temperature difference (BTD) IR10.8 minus IR3.9, which is displayed on the green color beam of the so-called Fog RGB image (left image). For clear scenes, the BTD depends on differences in atmospheric absorption (namely CO2 and H2O) and surface emissivity between the two channels. For Scandinavia, the BTD IR10.8–IR3.9 is typically around +5 K when it is cloud free and the IR3.9 channel is not CO2 corrected.

As shown in the following figure , the overall range of this BTD for clear ocean targets is from about +2 K (moist atmosphere, small viewing angle) to about +7 K (dry atmosphere, large viewing angle). For low-level water clouds (fog or low stratus) the BTD IR10.8–IR3.9 gives higher values than for cloud free surfaces as the emissivity of water droplets is lower at IR3.9 than at IR10.8.

n general, the smaller the water droplets the lower the emissivity in the IR3.9 channel and thus the larger the BTD. For Scandinavia, the values for the BTD are in the order of +9 K (IR3.9 channel not CO2 corrected) and thus about 4 K higher than for cloud free scenes.

However, the above described method for detecting low-level clouds at night-time does not work in very cold winter situations because the IR3.9 channel is very noisy for cold scenes. This can be seen in the image on the left which is actually quite noisy. For example, the low-level clouds over northern Finland can hardly be distinguished from the cold, cloud free land surfaces.

Thus, under cold winter conditions it is recommended to use the IR8.7 channel instead of the IR3.9 channel, i.e. to replace the BTD IR10.8–IR3.9 with the BTD IR10.8 - IR8.7 as shown in the image on the right. The range for the BTD IR10.8– IR8.7 used to generate the image on the right side is from +2 K to +6 K. Thus, the image on the right side is very similar to the so-called Dust RGB, but with a different (smaller) range for the BTD IR10.8–IR8.7 (the Dust RGB uses a range from 0 to +15 K, with a Gamma correction of 2.5).

The physical principle for the detectability of night-time low clouds using the IR8.7 channel is the same as for the IR3.9 channel, namely the lower emissivity of water droplets at IR8.7, as compared to IR10.8 or IR12.0. The advantages are that the IR8.7 channel is signifcantly less noisy for cold scenes and that it has no solar, reflected component (24-h capability).

The disadvantage is that the BTD IR10.8–IR8.7 for fog/low stratus scenes is only about +2 K higher than for cloud free scenes (in Scandinavia about +5 to +6 K for low-level clouds; +3 K for cloud free scenes). This may seem a rather weak signal, but it has to be seen together with the short-term noise figures of the IR channels.

For cold scenes, e.g. temperatures around 250 K, the IR3.9 channel is significantly more noisy than the IR8.7 channel and thus a signal (with low noise) of 2 K for the BTD IR10.8–IR8.7 is better than a signal (with high noise) of 4 K for the BTD IR10.8–IR3.9. Precise figures for the short-term noise (in flight measurements using the on-board blackbody calibration) are only available for a scene temperature of 300 K (see radiometric performances of SEVIRI on MSG-1 ), but with the help of the Planck function the short-term noise for a scene temperature of 250 K can be estimated to be around 1.25 K for the IR3.9 channel, whilst it is only 0.2 K for the IR8.7 channel.

 

Meteosat-8 Images

 
Met-8, 01 November 2006, 04:00 UTC
Met-8, 01 November 2006, 04:00 UTC
RGB Composite
IR12.0–IR10.8, IR10.8–IR3.9, IR10.8
Large Area
Animation (00:00–06:00 UTC, AVI)
Met-8, 01 November 2006, 04:00 UTC
Met-8, 01 November 2006, 04:00 UTC
RGB Composite
IR12.–IR10.8, IR10.8–IR8.7, IR10.8
Large Area
Interpretation
Animation (00:00–06:00 UTC, AVI)

Related Content

Fog RGB with IR3.9 (in polar stereographic projection) (03:00 UTC, source: SMHI)
Fog RGB with IR8.7 (in polar stereographic projection) (03:00 UTC, source: SMHI)
First autumn fogs in Europe (8 October 2005)