Monitoring severe convective storms in Texas in May 2020 with the help of the Convection RGB (ice particle size information).
08 December 2022
10 March 2021
By Jochen Kerkmann (EUMETSAT) and Daniel Lindsey (CIRA)
The Day Convection RGB, also called Severe Storms RGB or Ice Particle Size RGB, was originally devised to monitor convective storms. However, it is also very useful for detecting high-level wave clouds, dusty cirrus, pyro CBs, high supercooled clouds, and for monitoring hurricane development.
As it focuses on high-level clouds, it cannot be used for low-level features — all low-level clouds and land/sea appear blue, i.e. not much contrast.
The Convection RGB uses the following inputs (for the GOES ABI):
- Red: WV6.2-WV7.3 (as a proxy for cloud height)
- Green: IR3.9-IR10.3 (as a proxy for cloud particle size)
- Blue: NIR1.6-VIS0.6 (as a proxy for cloud phase)
Although it is the most complicated EUMETSAT RGB in terms of number of channels, its interpretation is much more straightforward than of the other EUMETSAT RGBs. In total, there are four main cloud colours: red = thick ice cloud with large ice crystals, yellow = thick ice cloud with small ice crystals, pink = thin ice cloud and bright grey/green = mid-to-high level supercooled water cloud (see EUMeTrain RGB Colour Guide).
The most important colour is yellow, which in case of a Cumulonimbus (Cb) cloud indicates:
- Cb cloud with strong updraft, or
- Cb cloud with a high,cold cloud base (common over high mountain areas, like the Rockies or the Himalaya mountains), or
- very cold Cb cloud (cloud tops colder than -70 °C), or
- a combination of the above.
Many European (MSG) convective cases have been published that demonstrate the good use of this RGB product in convective situations. However, so far, we have seen only a few GOES-16/17 case studies of severe storms with displays of this RGB product.
One of the reasons may be that this product is not included in display tools like the CIRA slider. Another reason may be that forecasters in the US don't rely heavily on ice particle size signals in severe storm analysis. There are two possible reasons for this:
- The High Plains, with relatively dry boundary layers (the area just east of the Rocky Mountains), often have storms with rather small ice crystals owing more to cold cloud base temperatures than to updraft strength. So this complicates the signal, meaning forecasters don't really know if a small ice signal means a storm with a strong updraft or not.
- The US ground radar network is very good, and after storms form, forecasters use it almost exclusively for storm warnings and analysis.
This case study is an attempt to highlight how the Convection RGB could be used in the US. Ice particle size information (i.e. updraft strength), as given by the Convection RGB, can give extra warning lead time, compared to radar data, in severe hail cases.
For this study, we checked all severe storm cases of May 2020 (using EUMeTrain ePort). The most interesting case we found is the case from 7-8 May 2020 (Figure 1).
On this day, an isolated supercell thunderstorm produced a long swath of large hail across far northern Texas. The storm produced hail as large as 8cm (3.25in) in diameter. The case is well described in two CIMSS blogs: Hail-producing supercell thunderstorm in Texas and Where will convective initiation occur?. The case is also shown on the CIRA GOES-16/17 Loop of the Day page, with the title 'Supercell splits, marches along Texas-Oklahoma border'.
Figure 2 shows the rapid scan animation of GOES-16 ABI VIS0.6 channel.
The loop reveals a number of important features:
- Moist boundary layer with low-level cumulus and gravity waves in the eastern part of the image.
- High-level jet marked by cirrus clouds.
- A dry line (north-south oriented cumulus line).
- Convective initiation in the area of the intersection of the dryline and the jet (Figure 3).
- Mature phase of the supercell with pulsing overshooting tops.
- The split of the storm (Figure 4).
- Large above-anvil cirrus plume (AACP) (Figure 5).
As an add-on to the above mentioned blogs, we examined the period immediately following convective initiation, i.e. the first 60 minutes of its lifecycle as mature storm, and took a closer look at some well-known satellite signals of severe convective storms like small ice, overshooting tops, cold ring shape and AACP.
In the GOES-16 'clean' infrared window (10.35µm) images (Figure 6, 7 and Figure 8), the pulsing overshooting tops (OT) and the development of a cold ring can be seen. The OTs exhibited infrared brightness temperatures in the -70 to -80ºC range (see Figure 7). For colour scale on Figure 6 and Figure 8 see Convection Working Group color enhancements page. For Figure 7, the infrared colour scale can be seen as a colour bar at the bottom of the images.
Figures 9 and 10 show the Convection RGB loops at 10-minute and one-minute intervals, respectively.
As discussed above, only high-level features are seen, namely the cirrus cloud band and the supercell storm. The latter appears with a yellow/orange colour. In particular, yellow dots can be seen in the very early phase of the storm, between 21:20 and 21:40 UTC (see 'rocking' animation), indicating small ice particles/strong updrafts. Note that, in this phase, the cloud top temperature was well above -70°C, i.e. the the yellow colour was due to small ice particles and not to very cold cloud tops.
In summary, from the VIS0.6, IR10.3 and Convection RGB loops the following timeline of satellite signals can be drawn:
- First appearance of small ice at 21.20 UTC.
- First overshooting top at 21.40 UTC.
- First cold ring at 21.50 UTC.
- First plume (AACP) at > 22.00 UTC (at about 22.30 UTC).
From the CIMSS blog we have the following information:
- Lighting activity began at 21.34 UTC
- First hail report (2.5 cm) at 22.29 UTC
- Giant hail (8 cm) reported at around 22:45 UTC
And from the local weather forecast office in Lubbock (TX) and the Storm Prediction Center (SPC) in Norman (OK) we have the following watches and warnings: -
- SPC at 21:45 UTC: high risk of severe hail up to 3.5 inches in diameter.
- NWS Lubbock 1st warning at 22:07 UTC: warning for large hail and wind.
Given the first hail report was at 22:29 UTC, there was a 22-minute lead time of the first NWS warning. The small ice particle size (from Convection RGB or 3.9 effective radius products) signal provided additional evidence that the storm exhibited a strong updraft and may produce large hail, potentially with additional lead time: 47 minutes if the forecaster issued a hail warning upon the first appearance of small ice (with a very high risk of false alarm), 27 minutes if the forecaster waited for confirmation of small ice in consecutive slots and the first appearance of overshooting tops.
In conclusion, extensive experience with MSG over the years has shown that for severe hailstorms the satellite signal (small ice, OT) often precedes the radar signal. Basically, the updraft signal comes first (satellite signal), before you see the downdraft/rain (radar signal). In the very early phase of the storm, the yellow signal in the Convection RGB is very small (only a few pixels) but this is crucial for the classification of the storm. One-minute satellite imagery, like in this case, is ideal, so you don't miss the updraft signal — in 10-min or 15-min satellite imagery you may miss the signal.