Meteosat Second Generation nowcasting imagery product guide

The Meteosat Second Generation Satellites (MSGs) are operated cooperatively between EUMETSAT and the European Space Agency (ESA). On board the MSG the Spinning Enhanced Visual and InfraRed Imager (SEVIRI) instrumentation provides full disk imagery of the Earth with a new set of images (across a number of different wavelengths) being taken every 15 minutes. This imagery is provided to users in near-real-time and allows for the production of multispectral/processed imagery that is useful for nowcasting. As part of the GCRF SWIFT catalogue we are generating a number of these images in the hope that they will be useful for operational nowcasting.

Of the images being generated for this catalogue, most are standard methods that have been used in many research and forecasting settings to provide insight into meteorological conditions. However, also included here are some new techniques that might prove useful in producing nowcasts. All methods are described below.

IR10.8 Enhanced Brightness Temperature Imagery (Standard)

Aim:
To highlight regions of cloud that are colder than a particular threshold. This will enable forecasters to quickly identify storms that have reached maturity and identify the coldest parts of the cloud top, potentially indicating the location of extreme weather.

Creation:
The background image is simply a greyscale image of the IR10.8 BT. Overlaid on this are colours for IR10.8 BT that are colder than -40 ℃. The colourscale is from -40 ℃ (blue) to -80 ℃ (red).

Use:
The advantage of this approach to using greyscale IR10.8 BT imagery is the ability to very quickly identify the coldest parts of the clouds that meet the cold conditions criteria. The structure of the cloud top can be examined and useful information gleaned from the structures and way in which cloud top temperatures are changing.
IR10.8 Enhanced Brightness Temperature Imagery (Standard)

Dust RGB (standard EUMETSAT algorithm).

Dust RGB (standard EUMETSAT algorithm).

Aim:
To allow users to identify regions where the atmospheric dust load is high. However, alongside this use these images are also very useful as a general purpose indicator of the meteorology, especially the location and development of deep convective storms.

Creation:
Dust RGB imagery is entirely produced using infrared channels. As such it is available 24 hours a day). The red channel is the brightness temperature (BT) difference between the IR12.0 and IR10.8 channels and gives some indication of optical thickness. The green channel is BT difference between the IR10.8 and IR8.7 channels and gives information about cloud phase (ice or water). The Blue channel is the IR10.8 BT and indicates surface temperature/height above the surface. The combination of these three colour channels leads to the ability to identify many dust and cloud features from dustRGB imagery.

Use:
As can be seen in the image below the dust RGB imagery highlights where dust has been raised from the surface with pink/magenta. This is generally very easy to identify against the light blue background of a dry airmass over a hot desert surface. However, identifying dust can be more difficult under less favourable conditions. When column water vapour is high and the surface temperature is lower (than the desert) the surface appears darker blue and dust can be masked. Cloudy conditions also make identifying dust more difficult as clouds block the dust below. The colour of clouds in dust RGB imagery is determined by optical thickness, phase and cloud top height. Thin cirrus clouds are indicated by black, deep convective clouds such as mesoscale convective systems are red while deep mid level clouds are orange.

Convection RGB (standard EUMETSAT algorithm).

Aim:
To highlight the presence of severe convective storms. Including helping to identify regions of updrafts and overshooting tops.

Creation:
Convection RGB imagery combines the BT difference between the WV6.2 and WV7.3 channels (on red), the BT difference between the IR3.9 and IR10.8 channels (on green) and the reflectance difference between the NIR1.6 and the VIS0.6 channels (on blue). As this product contains Visual and Near Infrared channels it can only be used properly during daylight hours.

Use:
Severe convective storms appear bright yellow because of the near zero BT difference WV6.2-WV7.3 (red channel) of overshooting tops. The strong updrafts in these clouds produce small ice particles at cloud tops due to homogeneous freezing of cloud drops, resulting with large BT differences in IR3.9-IR10.8 (green channel). Finally, large negative values of NIR1.6-VIS0.6 because of the large absorption at NIR1.6 by ice particles keeps the blue very low.
Convection RGB (standard EUMETSAT algorithm).

Overshooting tops RGB (standard South African Weather Service; SAWS algorithm).

Overshooting tops RGB (standard South African Weather Service; SAWS algorithm).

Aim:
To help identify overshooting tops within larger convective storm systems. These regions are linked to strong convective updrafts and therefore are closely associated with extreme weather at the surface.

Creation:
The overshooting top RGB uses Water Vapour and Infrared channels (channel differences) to attempt to highlight overshooting tops. The red channel is the BT difference between the WV6.2 and IR10.8 channels. The green channel is the brightness temperature difference between the IR9.7 and IR10.8 channels. The blue channel is the BT of the WV6.2 channel. As this is a relatively underused method (and information on the creation of images is not easy to find) the channels ranges and gammas for each channel are given below. As this method only relies on IR and WV channels it can be used 24 hours a day.
   
Beam   
   
Channel   
   
Range    
   
Gamma   
   
Red   
   
WV6.2-IR10.8   
   
-50 to 5K   
   
1.0   
   
Green   
   
IR9.7-IR10.8   
   
-30 to 25K   
   
0.5   
   
Blue   
   
WV6.2   
   
193 to 243K   
   
1.0   

Use:
The use of the overshooting tops RGB is very simple. Deep convective clouds show up as pink while mid level clouds are darker (red). The surface and very low clouds are almost black. Overshooting tops embedded within cold clouds appear whiter than the surrounding clouds and important structures (such as rings or horse shoe shapes) can be made out too. These can help users to identify the location of the most intense updrafts within a storm and therefore where convection (and extreme weather) are most active. One drawback can be the subjective nature of this method and the fact that overshooting tops in shallower clouds will not appear white (even though these could indicate rapid development).

IR10.8 Overshooting top identification - with enhanced brightness temperature imagery (non standard technique).

Aim:
The aim of this product is to quickly highlight the location and intensity of overshooting tops within MSG-SEVIRI satellite imagery. The region surrounding an overshooting top is shown in the same colours as the enhanced BT imagery on top of a greyscale IR10.8 BT image.

Creation:
In order to create this type of imagery there is a need to use an overshooting top detection algorithm. Several of these already exist but the method used here is very simple but seemingly robust. The data used in this algorithm is IR10.8 BT, this is thresholded using a -60 ℃ value to produce a mask. This mask is then further modified using the image processing techniques of binary erosion and dilation. Initially the mask is eroded (as shown in part c of the image below) using a disc (marked B in the image below) with a radius of 5 pixels. This gets rid of any very small regions that exceed the threshold. Following this the mask dilated using a disc with a radius of 7 pixels (but with a check to make sure that all included pixels still exceed the initial -60 ℃ threshold). After this each contiguous region of qualifying pixels is designated as being a “cell” and is now treated independently from all other cells. Within a cell the average BT and the horizontal gradient of the BT is calculated. For a pixel to be considered an overshooting top it must be colder than the “cell” average and have a horizontal BT gradient greater than 3 ℃/pixel. Once these pixels have been identified they are used to generate a second mask which is then dilated using a disc with a radius of 7 pixels. This second mask is what it then used to describe the regions plotted in colour in the final images.

IR10.8 Overshooting top identification - with enhanced brightness temperature imagery (non standard technique).

Use:
It is intended that these images could be used to highlight regions that have overshooting tops within larger well developed storms. Therefore indicating the parts of the storms that pose the greatest risk at the surface. However, due to the use of a variable “cell” specific temperature threshold it is also possible that these images could aid in the identification of storms that are rapidly developing.
Overshooting tops RGB (standard South African Weather Service; SAWS algorithm).

IR10.8 combined metrics RGB (cloud top roughness, BT and cooling rate of BT; non standard)

IR10.8 combined metrics RGB (cloud top roughness, BT and cooling rate of BT; non standard)

IR10.8 combined metrics RGB (cloud top roughness, BT and cooling rate of BT; non standard)

Aim:
To include a variety of useful metrics for indicating the behaviour of mesoscale convective systems into a single RGB image to help to interpret the current and possible future behaviour of storm systems.

Creation:
The three different metrics being represented within this RGB product are (1) the surface “texture” of the cloud, (2) the cloud top temperature and (3) rate of cooling of the cloud top (from one SEVIRI image to the next). All different metrics are based on the IR10.8 BT. As such there is no dependence on time of day so images can be produced 24 hours a day. The cloud top texture (red channel) is represented by the horizontal BT gradient of the image. Clouds with very bumpy tops will have high gradient values and so the red value will be high in these regions. The cloud top temperature (green channel) is simply represented by the IR10.8 BT value. Cold clouds will have a strong green component and warm clouds have a green value of 0. The rate of cooling (blue channel) is indicated by the difference in IR10.8 BT between the latest image and the image preceding it. Only regions that have cooled have blue values and to limit the impact on cloud motion the range is limited to -10 ℃ to 0 ℃. The details required to replicate these images are given in the table below. Once the RGB values are generated they are masked using a IR10.8 BT threshold of -40 ℃. These are then plotted over a greyscale IR10.8 BT image for context.
   
Beam   
   
Channel   
   
Range    
   
Gamma   
   
Red   
   
Horizontal   gradient of IR10.8 BT   
   
0 to 7 ℃/pixel   
   
0.5   
   
Green   
   
IR10.8 BT   
   
-90 to -40℃   
   
1   
   
Blue   
   
IR10.8 BT   cooling (since last image)   
   
-10 to 0℃   
   
0.5   

Use:
The interpretation of these images might require some time for familiarisation with the images. However they are quite intuitive in their use once this barrier has been overcome. Due to the thresholding only cold clouds are shown in colour. The green channel indicates the cloud temperature so if an MCS is mostly bright green then this mean that it has a well developed anvil. However if green is dominant this doesn’t suggest a cloud that is very convectively active. The regions within a storm that are light blue and white colour indicate locations where the cloud top is still very cold but also where there is significant cloud top texture and cooling. This is indicative of active convective updrafts and overshooting tops. These regions are likely to be connected to extreme weather at the surface. Often the same structures that are visible in the convection RGB and overshooting top RGB are also present in these images making the identification of well developed, active, convective storms very easy. Another feature of importance is the colour of the leading edge of the storm. If there is a pink to blue band along the edge of a storm then this storm is moving rapidly and generating fresh convective cells along the leading edge. If the storm is surrounded by these colours then you can infer that the storm is growing in all directions rapidly. However if the leading edge of a storm is darker (magenta/purple) then you can infer that the storm is less vigorous and possibly slower moving. Regions of a storm (or even entire convective storms) that appear darker green fading to browns and red can be expected to be dissipating. Often regions of a storm that take on these colours break apart and only the more active parts of a storm continue. Regions of cold cloud that fade to dark green/brown in all regions are likely to dissipate over the next 2 to 4 hours. Obviously reinvigoration can happen but this will be clear early in the process as the rapidly developing cells will show up as white/light blue regions surrounded by a pink fringe. White overshooting tops and ring and horseshoe structures might also be evident.