Overshooting tops

By definition, overshooting top stands for protuberances from the top of a Cumulonimbus cloud which rise above the general upper level of the cloud or anvil before sinking back, their appearance being that of cupolas (International Meteorological Vocabulary, WMO - No. 182), or (also anvil dome, penetrating top) a domelike protrusion above a cumulonimbus anvil, representing the intrusion of an updraft through its equilibrium level (level of neutral buoyancy). It is usually a transient feature because the rising parcel's momentum acquired during its buoyant ascent carries it past the point where it is in equilibrium; the air within it rapidly becomes negatively buoyant and descends. Tall and persistent overshooting tops are frequently observed with strong or severe thunderstorms in which there is a nearly continuous stream of buoyant updrafts (AMS Meteorology Glossary, http://glossary.ametsoc.org/).

Appearance and location of overshooting tops with respect to the overall storm structure (in side-view) is shown in the scheme and photo below:

Figure 9a. Scheme of a convective storm, showing - among other - a strong rotating updraft, which penetrates the cloud top equilibrium level (at which the storm anvil spreads horizontally) and next forms a bubble-like overshooting top, extending well above the storm anvil. Despite the fact that this specific scheme shows a supercell storm with strongly rotating updraft (forming a "mesocyclone"), the link of the overshooting top to storm updrafts is similar even for weaker, non-supercell convective storms. From http://commons.wikimedia.org/wiki/File:Tornadic_supercell.jpg.

For a full size view, click on the image.
Figure 9b. Side view of a convective storm above the Black Sea, taken on 22 June 2007 (14:35 UTC, OK423 flight), with a distinct overshooting top penetrating the storm anvil. Downwind of the overshooting top (in the photo to the right) are remains of older overshooting tops, including faint "jumping cirrus".  Photo © Martin Setv├ík, 2007.

The main importance of overshooting tops from the perspective of satellite meteorology and nowcasting is that these typically manifest the storm core location and storm activity. Storms with stronger and more persistent updrafts will generate more pronounced and longer-lived overshooting tops than the weaker storms. This is the basis for utilizing the overshooting tops in nowcasting - if we are able to detect distinct overshooting tops (either subjectively, using our eyes for diagnosis of satellite imagery, or automatically, using various overshooting tops detection algorithms), the storm is more likely to produce severe weather than a storm without overshooting tops, or with smaller or short-lived ones only. Also, as the overshooting tops in satellite imagery indicate the location of the storm core, the area near the overshooting tops is more prone to turbulence as compared to storm tops (or their parts) without these.

As the overshooting tops ascend above the cloud-top equilibrium level of the storm, they keep cooling down (by about 0.6 - 0.8 K per 100 meters), until they loose their energy - which can be up to about 2 - 2.5 km  above the surrounding anvil top. This is a very important fact, a key factor  for their detection in satellite imagery. In visible and near-IR imagery, they cast distinct shadows on the surrounding anvil top - provided that the Sun is not near the local zenith, almost overhead, and typically they resemble bubble-like, smaller scale features. In thermal IR bands they can be much colder than the surrounding anvil top - depending on the satellite pixel resolution, they can be colder by about 15-20 K than the surrounding anvil; the coldest part of the overshooting top can be somewhat smaller as compared to the overshooting top size as inferred from the visible or near-IR bands. However, as they loose their energy, they begin to collapse back to the anvil top, warming up rather quickly. For this reason it is easier to detect them during their ascent and at their culmination, rather than during their collapse phase. Of course, to reveal them by eye in thermal IR bands, it is essential to use the color-enhanced IR BT imagery. Examples of overshooting tops as seen in satellite imagery are shown in the figures below.

For a full size view, click on the image.

Figure 10. A "textbook" appearance of an overshooting top as seen in the NOAA/AVHRR imagery. In visible and near-IR bands (RGB composite image left) it can be recognized due to the shadow it casts at the lower anvil cloud top, while in the color-enhanced IR image (right) it can be revealed because of its very low temperature and also the steep BT gradients around it.
Figure 11. Overshooting tops as seen in MSG imagery. The same storm ( 2013-05-18, 12:30 UTC, Meteosat-10, Nigeria) as shown in Figures 6 and 8 in the previous sections. The overshooting tops are indicated by arrows; while in the color-enhanced image (right) the overshooting tops can be easily localized, it is not that easy to see these in the HRV image (left). As this image was captured close to the local noon, and the Sun is north of the Equator at this time of the year, the almost overhead position of the Sun is not favorable for casting more distinct shadows, and makes the overshooting top detection rather difficult in the HRV band. 

As the typical lifetime of the overshooting tops ranges from about 5 to 20 minutes, detecting the overshooting tops by satellites with longer scan cycle (such as the 15 minute interval of the MSG satellites when performing the full disk scan) is somewhat random process, some of the overshooting tops may escape their detection, or are captured at their later stage (when can be warming up again). The same applies even more to the satellites on polar orbits. Therefore, one should be aware that if a satellite image (or automated detection algorithm) does not indicate any overshooting top, the storm still may be severe, we just may be unlucky missing the overshooting top due to timing of the image. The shorter the scanning interval, the higher the chance of providing more representative information about the storms.

Big variability of the storm tops and namely their overshooting tops can be seen in the loops based on the MSG image data, captured during the 2.5-minute rapid scan experiments in 2013. The movie files and corresponding presentation devoted to these can be found here. Despite the fact that these movies show storms above central Europe, they nicely document how transient the overshooting tops can be.

Besides the temporal issues, also the pixel size (geometrical resolution) plays a significant role. The better the image resolution (the smaller the pixel size), the more likely the chance of detecting the overshooting tops more easily. This applies not only to images in visible bands, but even more so to those in thermal IR bands. Lower pixel resolution in the IR bands means that a sub-pixel-sized BT minimum may escape its detection due to averaging of the BT within the pixel, averaging of the temperature minimum with its warmer surroundings. For this reason it is also necessary to display the satellite image data at their full resolution - zooming out can mean a loss of important information.