Using 5.15 micrometer imagery to see convective initiation
In July 2022, significant flooding affected metropolitan St. Louis as multiple convective cells trained through the area. (This blog post contains clean window imagery for 26 July portion of the two-day event; the St Louis WFO also discussed the event). This WPC discussion describes the atmospheric setup on 25 July.
A retrospective model run for this flooding event was created using a WRF simulation. The model had 2-km resolution and data were output every 15 minutes. The WRF data were input into a CRTM to convolve the spectral response function at 5.15 micrometers — the novel water vapor channel that is planned for the GeoXO Imager (the follow-on to GOES-R’s ABI) that is planned for launch in the mid-2030s. Weighting functions that compare the 5.15 µm channel with GOES-R channels at 6.19 µm, 6.95 µm and 7.3 µm (for approximately the same atmosphere), below, show much more information from the boundary layer (850 mb and below) sensed by a satellite that detects energy at 5.15 µm compared to longer wavelengths. The image below was created from this image for 5.15 and this image for GOES-R Bands 8-10.
What does the model output tell you? Consider the toggle below. The 5.15 µm imagery shows a region of cooler temperatures (yellow in the enhancement used vs. orange to the north and south) over Nebraska and Kansas into Missouri as highlighted by the blue arrows; a similar distribution is apparent in the 7.3 µm imagery — blue in the enhancement from Kansas/Nebraska into Missouri vs. yellow to the north and south. The cooler brightness temperatures mean that the height of the moisture layer detected is higher in the atmosphere; a likely reason for this moisture distribution is that more moisture is present in this corridor just to the north of a stationary front.
How do the fields change as convection starts to initiate? That is shown in the animation below that covers the times 0000 UTC to 0515 UTC on 25 July 2022. In particular, the 5.15 µm imagery, with a signal that includes more information from lower in the boundary layer, can detect the low-level development of convection over Kansas and Missouri at earlier times than occurs in the 7.3 µm imagery. Compare the cooler brightness temperatures especially between 0230 UTC and 0400 UTC over Kansas and Missouri. The 0345 UTC image, shown below the animation, shows significant 5.15 µm cooling in/around Kansas City (northeast Kansas/western Missouri) that is not as apparent in the 7.3 µm imagery. The lowest-level ‘water vapor’ imagery at 5.15 µm is giving a heads up for convective development about 90 minutes before it appears in the 7.3 µm imagery.
The mp4 animation below shows four water vapor channels from the model simulation every 15 minutes from 0000 UTC – 2345 UTC on 25 July. In addition to the earlier detection of developing convection over central Kansas/Missouri mentioned above, note how the convection initially over southern Missouri dissipates as it moves into regions with warmer brightness temperatures (where low-level moisture is perhaps not quite so abundant). Surface features are visible over Wyoming in the 5.15 µm, especially as the surface warms during the day, indicating information from the surface could be sensed in addition to information related to the distribution of water vapor. Keep in mind these types of very dry regions are not likely candidates for convective initiation. Regions with moderate to high amounts of low-level moisture are less suspectible to this type of surface contamination, making the 5.15um band very useful for early detection of severe storms.
Thanks to Zhenglong Li, Nate Miller and Mat Gunshor, CIMSS for these images! For more information on GeoXO, click here.