Common visualisation of the upper layer and background images - 'sandwich' product


In the figures in the previous chapter we observed cloud top features independently in the two single channels - in a solar single channel (or RGB image using solar channels) and a colour-enhanced IR window brightness temperature image. In this chapter we present the 'sandwich' product, which merges (blends) the two bands, allowing observation all of these features simultaneously, in one single product.

Principle of the method

The product consists of two layers (as shown in Fig. 5.1):

  • The upper layer is a colour-enhanced IR window image, showing the details of the brightness temperature field. Alternatively, another coloured image, e.g. RGB image can be used.
  • The bottom layer (background image) is a black and white single solar channel image. Alternatively, true colour RGB image, or other almost black and white RGB image based solar channels can be used.

Figure 5.1: Principle of the method. (Courtesy of Martin Setvák, CHMI)

There are several options for how to 'blend' these two layers.

  • The simplest way is to use the partial transparency of the upper layer, setting the layer opacity somewhere between 40 to 80%. Left panel of Fig. 5.2 shows an example of this approach.
  • Better results (brighter, more saturated colours) can be obtained by using a more effective type of 'blending' of the two layers together:
    • For interactive processing:
      • in Adobe Photoshop one can use either the 'Multiply' or 'Linear Burn' functions (in 'Blending Options'), in combination with the layer opacity set as above.
      • the GIMP (GNU Image Manipulation Program) freeware program can be also used in a similar way.
    • Similar products can be generated automatically, e.g. using the ImageMagick software. Right panel of Fig. 5.2 shows an example.

For operational, fully automatic processing the ImageMagick (freeware) software can be used, for example by running the following ImageMagick script:

# alpha blending of the IR-BT layer
alpha=70
# sets the alpha blending of the IR-BT layer
convert ${ir_bt} -alpha On -channel Alpha -evaluate set ${alpha}% ${ir_bt_png}
# merging (blending) the two images together
composite ${hrv} ${ir_bt_png} -compose Multiply -quality 90 ${output}

The two layers 'ir_bt' and 'hrv' should be enhanced before running the above script. The user can modify the alpha parameter. It defines the weighting of the upper layer in the final product.

In Fig. 5.2 both panels present SEVIRI HRV/IR10.8 blended images of the same scene created in different ways. This figure demonstrates the difference between the methods. In the left image the transparent upper layer is overlaid on the background image, while in the right image the above ImageMagick script was used to blend the same input files. Alternatively, it is possible to use any other standard or specialised graphics editor with an option of blending two images together.

Figure 5.2: SEVIRI HRV/IR10.8 blended image processed without (left) and with (right) the ImageMagick script, taken on 8 July 2015 at 13:40 UTC

Figs. 5.3 - 5.6 show the upper layer, the background image and the resulting sandwich image together. Note that the storm of Fig. 5.3 was presented in Fig. 4.3a and the storms of Fig. 5.4 was presented in Fig. 4.4a.

Figure 5.3: Overshooting tops, cold-U and above-anvil ice plumes in the form of 'ship waves' in MODIS 250m band 01 image (upper left), 1 km colour-enhanced band 31 image (upper right) and band01/band31 blended image (bottom), taken on 16 September 2016 at 12:22 UTC over Tyrrhenian Sea and Italy. (Courtesy of Martin Setvák, CHMI)

Figure 5.4: Meteosat SEVIRI HRV (upper left), IR10.8 (upper right) and HRV/IR10.8 blended (below) images of Germany, taken on 12 July 2011 at 17:40 UTC. (Courtesy of Martin Setvák, CHMI)

Figure 5.5: Meteosat SEVIRI HRV (upper left), IR10.8 (upper right) and HRV/IR10.8 blended (below) images of Tyrrhenian Sea and Italy, taken on 16 September 2016. (Courtesy of Martin Setvák, CHMI)

Figure 5.6: Meteosat SEVIRI IR10.8 (left), HRV/IR10.8 blended (middle) and HRV (right) images, taken on 20 May 2008 at 16:45 UTC of a tornadic storm over Hungary. (Processed at OMSZ with the ImageMagick script using 200-240 K BT region for the colour enhancement.)

Figs. 5.4 and 5.6 show storms with very complex cloud top structures: overshooting tops, cold ring or cold-U shape, above anvil ice plumes, and anvil-top waves. One can see how more informative the sandwich product is than the HRV or the IR10.8 separately!

Benefits of the sandwich products compared to the 2-panel visualisation

The primary advantage of sandwich products is that they merge the two input images into one single image. So, one can observe the characteristics of both images simultaneously in one single product. The spatial fitting of the different characteristics is perfect. In the case of the visible/IR-window sandwich combination, the visible band adds the cloud-top 'morphology' (shadows and textures) to the final image, while the colour-enhanced IR-window band adds the BT information.

Additionally, it is much easier to follow the evolution of convective storms (or any other weather phenomena) in one product, rather than in two windows, showing the two input bands separately.

All this makes the sandwich product very attractive and effective for operational applications. It becomes even more attractive and useful when used in loops of satellite imagery (e.g. the best way to present rapid scan animation of zoomed details of storm tops).

Solar and IR window bands combinations which can be used to create sandwich products

Table 5.1 shows some solar and IR window band combinations of present, and future, satellite imagers which can be used to create sandwich products from their image bands.

Satellite Imager Background layer Spatial resolution Upper layer Spatial resolution
Geostationary Satellites
Meteosat (MSG) SEVIRI HRV 1 km IR10.8 3 km
GOES-R series ABI Band 2 (0.64 μm) 0.5 km Band 13 (10.3 μm) 2 km
Himawari-8/9 AHI 0.645 μm 0.5 km 10.45 μm 2 km
MTG-I FCI VIS0.6 0.5 km IR10.5 1 km
FY-4 series AGRI 0.65 μm 0.5 km 10.7 μm 4 km
Polar satellites
Metop, NOAA AVHRR Band 1 (0.63 μm) 1 km Band 4 (10.8 μm) 1 km
AVHRR Band 2 (0.865 μm) 1 km Band 4 (10.8 μm) 1 km
Terra, Aqua MODIS Band 1 (0.65 μm) 0.25 km Band 31 (11.0 μm) 1 km
S-NPP, NOAA-20 VIIRS Band I1 (0.64 μm) 0.375 km Band I5 (11.45 μm) 0.375 km
EPS-SG METimage 0.670 μm 0.250 km 10.790 μm 0.5 km

Figure 5.1: Solar and IR window bands combinations to create sandwich products.

Summary

Cloud top features can be easily, effectively analysed in the sandwich product. It merges the information of the two input images into one single image. One can observe the characteristics of both images simultaneously in one single product.

It is much easier to follow the evolution of convective storms (or any other weather phenomena) in one product. Operational use of sandwich products - rapid scan animations of convective storms, shown in zoomed images.

The high resolution of polar satellite images has the advantage of more detailed information, but for operational use the geostationary satellites, with their much higher resolution in time, are preferable.