PRELIMINARY DOWNBURST CLIMATOLOGY AND WARNING GUIDELINES FROM A SINGLE CELL THUNDERSTORM DATABASE
Ken Falk, SOO
Lee Harrison, MIC
John Elmore, Forecaster
NWSO Shreveport, LA
A downburst is defined as a rapid downdraft of wind from a single cell thunderstorm which produces a sudden outflow of horizontal winds at the surface (Fujita 1981). Beginning with a downburst- producing single cell thunderstorm over Waskom, Texas in July, 1996 (Falk and Harrison 1996), and continuing through the summer of 1997, we compiled a database of over 30 single cell thunderstorms which exhibited strong, high-level reflectivity cores, and mid-level convergence signatures. Many of these thunderstorms produced severe downbursts.
Radial velocity convergence was first proposed as a procedure for forecasting microbursts by Roberts and Wilson (1989). Other research (Eilts, et al. 1996) indicated values of radial velocity con-vergence of greater than 44 kt at mid-altitudes in a storm may produce damaging winds at the surface. We applied this research to the National Weather Service Doppler radar (WSR-88D) at Shreveport to create a database of storms which, we hoped, would enhance our ability to issue more accurate severe thunderstorm warnings, and better airport advisories for downbursts. The evaluation of these data allowed us to determine preliminary severe thunderstorm warning guidelines for downbursts (and associated microbursts and macrobursts) using radial velocity convergence, and to determine a climatology of downburst-producing single cell thunderstorms.
In this paper, you will find a conceptual model of a downburst-producing single cell thunderstorm, a description of how the data were gathered for this study, the results of the data, including our initial recommendations for warnings based on the data to this point, and a preliminary climatology of downbursts. We plan to continue gathering data through the summer of 1998 in order to refine these preliminary results.
2. Conceptual model of a downburst-producing single cell thunderstorm
Conceptual models for downburst-producing single cell thunderstorms were developed by Roberts and Wilson (1989) for low reflectivity, moderate reflectivity, and high reflectivity downburst-producing thunderstorms. We have composed a more general conceptual model for downburst-producing single cell thunderstorms based on WSR-88D cases we have examined. The type of downburst depicted in this conceptual model and observed on the Shreveport WSR-88D is usually referred to as a wet-microburst. It is important to note that this conceptual model Fig. 1 applies to single cell (or pulse type) thunderstorms only. This conceptual model does not apply to other types of thunderstorms, such as multicell squall lines, bow echoes, or supercells.
It should be noted that these downburst-producing single cell storms can occur in the same environment, and at the same time, as an organized squall line event. The event at Waskom, Texas referred to earlier (Falk and Harrison 1996) occurred in a single cell thunderstorm located ahead (or downwind) of an advancing squall line.
Arrows in Fig. 1 depict storm scale wind flow within our conceptual model of a downburst-producing single cell thunderstorm. Storm top divergence will usually be detected in the development stage of the single cell thunderstorm. Since storm top divergence is associated with updraft strength, it is not a good predictor of downburst potential, although it may still be present as the downburst is occurring. In other words, storm top divergence may, or may not, be present in a downburst-producing single cell thunderstorm. If storm top divergence is still strong while a downburst is occurring, the storm will probably maintain its strength longer than if storm top divergence is not present at the time of the downburst. If storm top divergence is not present as the downburst occurs, then the storm will likely collapse (within 30 minutes) during the time of the downburst.
Roberts and Wilson (1989) showed that radial velocity convergence (or, more simply, velocity convergence) may be a predictor of potential microbursts from single cell storms. Doppler radars, of course, measure velocities only along radial lines. Velocity convergence is a term used to describe situations in which, extending outward from the radar along a radial, outbound velocities are adjacent to inbound velocities. For convenience, the magnitude of the velocity convergence is taken to be the sum of the absolute values of the maximum outbound and inbound velocities.
We have chosen to identify a term we call storm velocity convergence (SVC) in our conceptual model. SVC is simply the velocity convergence in the core of the storm just above the cloud base, and it is usually best shown by the WSR-88D storm relative motion (SRM) product. We believe SVC reflects drier air entraining into the storm as evaporative cooling and/or precipitation drag induce downward motion within the storm, leading to a downburst at the surface. Roberts and Wilson (1989) suggest the same causes for downbursts. The cases we examined indicate that SVC is a good predictor of a potential downburst when it is present in or near the high reflectivity core of a single cell storm, somewhere within the 5,000 to 11,000 ft AGL layer.
A low-level divergence of winds will occur as the downburst contacts the ground (Fujita 1981). This feature will be detected by the WSR-88D only if the storm is within 20-30 miles of the Doppler radar (due to beam height restrictions), but is assumed to be present in storms at ranges beyond 30 miles, when SVC is detected. SVC extends the range of detection of downburst-producing single cell thunderstorms to about 90 miles from the radar. A radar depiction (not shown) of SVC and low-level divergence in a microburst-producing single cell thunderstorm was documented by Falk and Harrison (1996).
Another radar depiction of a downburst-producing single cell thunderstorm is shown in Color Plate 1. This thunderstorm was about 66 miles south-southwest of the Shreveport WSR-88D, yet the WSR-88D was still able to detect SVC in the high reflectivity core of this single cell thunderstorm (due to the height AGL where SVC is usually detected). This thunderstorm produced a downburst of 70-80 mph which uprooted large trees.
3. Storm velocity convergence database
Using the WSR-88D at Shreveport (and one case from the Little Rock, Arkansas WSR-88D), we recorded several parameters on storms that exhibited storm velocity convergence (SVC). The criterion for a storm to be included in the database was SVC of 32 kt or greater in or near the storm core of a single cell thunderstorm. Note that not all of the storms produced severe surface winds (50 kt or more) or damage, since these criteria were not required for a storm to be in the database. We included storms that did not produce strong winds in order to help establish the SVC threshold for severe thunderstorm warnings. Using the SVC threshold of 32 kt or more in a single cell thunderstorm, we were able to gather information on 32 events during the summers of 1996 and 1997 (all but one in Shreveport's county warning area.).
To measure SVC, we used the Doppler radar VR Shear function oriented along a radial from the maximum inbound to the maximum outbound wind in the convergent signature. SVC does not require that the maximum wind pixels within the convergence be gate-to-gate. They can be several pixels apart on the WSR-88D velocity product. We mutiplied the VR Shear by two to get an estimate of convergence (since VR Shear is actually used to calculate rotational velocity, which is an averaged number).
Convergence can also be measured directly off the WSR-88D SRM display (or base velocity), but one must remember that the displayed velocity is the lowest value in the range of velocities of that particular color. When using this method, it is best to use the mid-range value of velocity for each indicated color for an estimate of true convergence (VR Shear x 2 already takes this into account).
For each storm in the database, we recorded the date of the storm, the location of the storm, the range from the radar of the storm, the magnitude of SVC, and the height of the maximum SVC. For each storm that produced either a report of severe winds (50 kt or greater) or wind damage, we also recorded the surface wind speed and/or damage, and the lead time. Lead time was defined as the time difference between the initial detection of SVC of a least 50 kt, and the first report of surface winds of at least 50 kt, or wind damage, or radar detected divergent winds at the surface. (It will be seen below that we found a good correlation between SVC of this magnitude and surface winds of 50 kt or more.) We used real time reports of wind gusts or damage whenever possible to get the best estimate of actual time of occurrence.
4. Preliminary results - Downburst warning guidelines and climatology
Severe thunderstorm warning guidelines for downbursts
Based on the evaluation of 32 events of single cell thunderstorms that contained SVC in or near the high reflectivity core of the storm, we made a preliminary determination of the following warning guidelines for issuing a severe thunderstorm warning for a downburst:
(1) High reflectivity core of 50 dBZ to heights of 25,000 ft AGL, along with
(2) Storm velocity convergence (SVC) of 50 kt somewhere in the layer 5000 - 11,000 ft AGL in or near the high reflectivity core.
Severe thunderstorm warnings based on these guidelines should contain wording about strong damaging winds.
These guidelines produced an average lead time of 11.4 min between the time SVC of at least 50 kt was detected and the severe wind occurrence at the surface. Lead times ranged from a low of zero min to a maximum of 18 min.
We found these guidelines to be most reliable within 90 miles of the WSR-88D. This is due to beam height restrictions and less reliable velocity data at longer ranges. The detection of SVC at ranges up to 90 miles from the radar is a significant improvement over the previous low-level divergence detection technique for downbursts, which is effective only to a maximum range of 20-30 miles.
We observed that several storms collapsed soon after SVC was detected. Thus, the high reflectivity core of the storm may descend rapidly after SVC occurs, so the guideline on the height of the high reflectivity core is somewhat flexible. It was also noted that single cell storms that contained storm top divergence at the same time SVC was occurring were longer lived than the storms that did not contain strong storm top divergence. Thus, storms with strong storm top divergence and SVC produced larger areas of damaging winds (macrobursts).
We also looked for single cell thunderstorms that produced severe downburst winds, but did not contain SVC. We found none! However, this is not to say that all downburst-producing single cell thunderstorms contain SVC. Further evaluation of this will be made after more data are acquired.
It was noted as the data were being gathered that storms producing severe downburst winds had reflectivity cores of 50 dBZ to at least 25,000 ft AGL. Although this feature was not checked in the earlier part of the database, during the latter half of the study all of the severe downburst events had these high reflectivity cores. The WSR-88D mid-level LRM product can be used to quickly determine which storms have high reflectivity cores, and thus should be further examined for the SVC signature. It should be noted that the magnitude and height of the high reflectivity core (50 dBZ to at least 25,000 ft AGL) may vary somewhat over other parts of the country, and should be adjusted for local climatology of downburst storms.
As noted in the above guidelines, the best layer to detect SVC appears to be 5000 - 11,000 ft AGL. SVC detected above 11,000 ft AGL did not correlate well with damaging winds at the surface, and SVC was usually not detected at all below 5000 ft AGL. However, there were three events that occurred close enough to the WSR-88D (within 32 miles) to see the low-level divergent signature. This feature is thought to be detectable only below 2500 ft AGL.
Out of the 32 events recorded, SVC ranged from a low of 32 kt (our lower end to be included in database) to a high of 110 kt. Figure 2 shows magnitude of SVC versus range of the storm from the WSR-88D for the 32 events. The figure also shows which SVC events produced severe weather. In the database, 50% of the events had SVC of at least 50 kt. Of these events, 81% produced severe weather (75% produced severe downburst winds and 6% produced severe hail). The severe hail report was somewhat of an anomaly since only one event out of 32 produced severe hail. Of the events that contained SVC under 50 kt (the other 50%), 13% produced severe downburst winds. These two events were at ranges beyond 90 miles from the WSR-88D.
It can be seen in Fig. 2 that SVC of 50 kt or greater appears to be a reliable guideline to issue a severe thunderstorm warning for downbursts, at least in the area of our study. The SVC must occur in or near the high reflectivity core of a single cell thunderstorm. As previously mentioned, the convergence "couplet" does not have to be gate-to-gate, but can be separated by a nominal distance, depending on the diameter of the storm.
Radar interpretation of SVC should not be confused with convergence that occurs along outflow boundaries, along squall lines, near the leading edge of bow echo storms, or in supercells. These types of thunderstorms usually occur in stronger vertically sheared environments than single cell storms, and are more "organized" than single cell storms. The radar meteorologist must have a firm grasp of the physical processes that are taking place within thunderstorms to properly interpret the radar display. Thus it is important to be knowledgable of the conceptual models (or storm structure) of various types of storms (Falk 1997, Ray 1986, Doswell 1985).
We hope in the future to correlate the magnitude of SVC with the magnitude of surface winds, but at this point, we will limit our discussion to saying 50 kt or more of SVC in the 5000 - 11,000 ft AGL layer correlated well with surface winds of 50 kt or greater.
Local airport advisory guidelines for downbursts
Local airport advisories are issued for many airports by National Weather Service offices. Downburst-producing single cell storms are one of the most dangerous weather phenomena for aircraft. Local airport advisory criteria suggest that the thunderstorm be within 5 mi of the airport sometime during the valid time of the advisory. Wind speed criteria (35 kt or more) for local airport advisories are lower than that for severe thunderstorms warnings, thus the guidelines for issuing a local airport advisory for a downburst should have lower thresholds.
Preliminary indications from the database are that a local airport advisory should be issued for a downburst-producing single cell thunderstorm based on the following guidelines:
(1) High reflectivity core of 45 dBZ to heights of 25,000 ft AGL, along with
(2) Storm velocity convergence (SVC) of 35 kt somewhere in the layer 5000 - 11,000 ft AGL in or near the high reflectivity core.
The local airport advisory should specifically mention either the word "downburst" or the word "microburst" because these words convey a message to the air traffic controllers and pilots that this thunderstorm poses a significant threat to aircraft operations.
The magnitude and height of the high reflectivity core will vary somewhat over other parts of the country due to local climatology of downburst-producing storms. It appears that SVC of 35 kt is sufficient to produce 35 kt surface winds. Certainly the guidelines for severe thunderstorm warnings are sufficient for a local airport advisory to be issued.
Climatology of downbursts
From the database we were able to make some preliminary observations of the climatology of downburst-producing single cell thunderstorms. Downbursts occurred in the summer season from early June through the end of September, and were much more common than we previously suspected.
Although we recorded only one severe downburst in 1996, we recorded 13 in 1997. We also recorded 18 SVC events in 1997 that likely produced downburst winds, but were not severe. This brings the total of downbursts in 1997 (both severe and non-severe) to 31. This wide difference in number from one year to the next is likely due to the fact that we were not as aware of the radar signatures of downbursts in 1996 as we were in 1997. Of course, it could also be argued that 1997 was a "good" year for downbursts. We were also more diligent in 1997 in getting ground truth wind reports from storms we suspected of producing downburst winds. At any rate, it is apparent that downbursts can be common in the summer months, and may be the most prevalent severe weather phenomena during the summer (at least in the south central and southeast U.S.).
Almost all of the downbursts occurred during the mid-afternoon to early evening hours during the hottest time of the day. However, the single cell thunderstorm that exhibited the highest SVC (110 kt) occurred around 1330 UTC (830 AM CDT)! This thunderstorm produced a macroburst which uprooted trees and downed power lines near Hope, Arkansas. Any single cell storm that has a high reflectivity core should be monitored for SVC at any time of the day during the summer.
We noted that on a given day several storms could produce downbursts, so they can and do occur in "clusters" on favorable days. But there were also some days when only one storm (out of many storms) produced a severe downburst, such as the storm mentioned in the preceeding paragraph. The severe downburst at Waskom, Texas (Falk and Harrison 1996) was another such isolated event.
Additional events will help us further determine a climatology of downbursts.
A database of downburst-producing single cell thunderstorms was collected during the summers of 1996 and 1997. By analyzing these data, a conceptual model of a downburst-producing single cell storm was developed, based on previous single cell microburst research and on the radar signatures we have seen.
Information was gathered on 32 storms that exhibited a signature we called storm velocity convergence (SVC), and from these data, guidelines were developed for issuing severe thunderstorm warnings and local airport advisories for downbursts. We also were able to develop a preliminary climatology of downburst-producing single cell thunderstorms.
As we continue to gather additional events through the summer of 1998, we will review and refine our guidelines for warnings and climatology of downbursts.
Doswell III, C. A., 1985: The operational meteorology of convective weather, Volume II: Storm Scale Analysis. NOAA Technical Memorandum ERL ESG-15, NOAA National Severe Storms Lab, Norman, OK.
Eilts, M. D., J. T. Johnson, E. D. Mitchell, R. J. Lynn, P. Spencer, S. Cobb, and T. M. Smith, 1996: Damaging downburst prediction and detection algorithm for the WSR-88D. 18th Conference on Severe Local Storms, American Meteorological Society, 541-545.
Falk, K. W., and M. L. Harrison, 1996: Waskom, Texas, single cell microburst detected on National Weather Service Doppler radar - 23 July 1996. Technical Attachment SR/SSD 96-43, National Weather Service, Southern Region, Fort Worth, TX.
Falk, K. W. 1997: Techniques for issuing severe thunderstorm and tornado warnings with the WSR-88D Doppler radar. NOAA Technical Memorandum NWS SR-185, NOAA National Weather Service, Southern Region, Fort Worth, TX.
Fujita, T. T., 1981: Tornadoes and downbursts in the context of generalized planetary scales. Journal of the Atmospheric Sciences, 38, 1511-1534.
Ray, P. S., 1986: Mesoscale Meteorology and Forecasting, American Meteorological Society, 793 pp.
Roberts, R. D., and J. W. Wilson, 1989: A proposed microburst nowcasting procedure using Single-Doppler radar. Journal of Applied Meteorology, 28, 285-303.