1. INTRODUCTION

Twenty-five northern Alabama tornadic thunderstorms were analyzed using archive level IV data sets from Weather Surveillance Radar - 1988 Doppler (WSR-88D) systems at Maxwell AFB, AL (KMXX), NWSFO Birmingham, AL (KBMX), and Columbus AFB, MS (KGWX). The tornadoes occurred during the period from February 1993 to May 1995, and their approximate tracks are shown in Figure 1. The data set covered a spectrum of storms from a weak multicell that produced an F0 tornado, to a classic supercell with a deep Bounded Weak Echo Region (BWER) that produced softball size hail and an F4 tornado.

The 25 events were divided into supercell and nonsupercell categories. Nine supercells were responsible for 15 of the tornadoes. The supercells produced 7 strong tornadoes and 2 violent tornadoes, while 10 nonsupercells were responsible for 2 strong tornadoes and 8 weak tornadoes (see Figure 2).

Ninety percent of the nonsupercell cases were associated with squall lines. The average range from the Radar Data Acquisition (RDA) site to the supercells, at an initial tornado touchdown, was 84 n mi, varying from 26 to 115 n mi. Five of the supercells were at a range at or beyond 100 n mi. The 10 nonsupercell events had an average range of 76 n mi, and varied from 41 to 102 n mi. Nine data sets came from KMXX in east-central Alabama, before KBMX and KGWX WSR-88Ds were installed, reflecting the long range bias.

2. REFLECTIVITY DATA

2.1 Supercells

Of the 15 tornadoes associated with supercells and the 17 WSR-88D data sets associated with those storms, only 24 percent of the data sets indicated hook echoes at any time during the life cycle of the storms (see Figure 3). The maximum range of detection of a hook was 78 n mi, while 3 supercells less than 79 n mi from the RDA did not exhibit hooks. Two supercells, each from two different sets of radar data, displayed hook echoes from the RDA closest to the storm and not from the RDA farther away. One of the supercells without a detectable hook echo did produce a pendant echo.

Seventy-six percent of the supercells displayed an inflow notch or concavity in the tight reflectivity gradient on the updraft flank of the storm. This included two supercells that were 110 to 115 n mi from the RDA. BWERs were detectable in 65 percent of these storms, with a maximum range of 115 n mi. The BWERs were detectable on average 2 volume scans prior to the tornado-bearing volume scan. Volume scans varied from 5 to 6 minutes depending on whether the WSR-88D was in Volume Coverage Pattern (VCP) 11 or 21. The BWER times varied from 6 volume scans prior to touchdown, to 2 volume scans after touchdown. Weak Echo Regions (WERs) were noted in 94 percent of the supercells.

Severe Weather Probability (SWP) values were available for 53 percent of the supercell events. The maximum SWP before the occurrence of a tornado averaged 52 percent, and ranged from 23 to 77 percent.

The maximum reflectivity before tornado touchdown averaged 65 dBZ (see Figure 4), at a height of 21,600 ft, and occurred 3 volume scans before touchdown. The maximum reflectivity height decreased in 41 percent of the supercell cases right before tornado touchdown. Eighteen percent of the time the maximum reflectivity height was increasing right before tornado touchdown and in the remaining cases no trend was apparent.

The maximum Vertically Integrated Liquid (VIL) water averaged 63 prior to tornado touchdown, and ranged from 33 to 80 . These maximum VILs occurred on average 3 volume scans prior to touchdown. In 53 percent of the cases, the VILs were decreasing just prior to touchdown (Peters and Kilduff 1993, Murphy et al. 1995), indicative of the collapsing phase of the supercell (Lemon and Doswell 1979).

The maximum echo top averaged 53,000 ft (see Figure 5), ranging from 47,000 to 61,000 ft.

2.2 Nonsupercells

None of the nonsupercell tornadic thunderstorms exhibited a hook echo. Thirty percent of the cases displayed pendants. Inflow notches were noted in 40 percent of the events. None of the nonsupercell thunderstorms displayed a BWER, but 60 percent did show a WER. SWPs were too low to calculate.

The maximum reflectivity averaged 60 dBZ before tornado touchdown, ranging from 52 to 67 dBZ, at an average height of 17,800 ft. The maximum reflectivity occurred 4 volume scans before the tornado-bearing volume scan, and consequently 70 percent of the time was decreasing right before touchdown. Forty percent of the time the height of the maximum reflectivity was also lowering.

Maximum VIL values averaged 45 , and occurred 3 volume scans before the scan during touchdown. Seventy percent of the time the VILs were decreasing just prior to tornado touchdown. Maximum echo tops averaged 40,000 ft.

3. VELOCITY DATA

3.1 Supercells

Rotational velocity defined as

averaged a maximum of 45 kt before tornado touchdown, ranging from 31 to 55+ kt, and occurred on average 3 volume scans before touchdown. Forty-one percent of the supercells displayed rotational velocity of 55+ kt sometime during their life cycle. The average height of the maximum rotational velocity was 15,100 ft, ranging from 8,400 to 28,300 ft.

A mid-level mesocyclone was defined for this study as a mesocyclone with at least 30 kt of rotational velocity at or above 15,000 ft. All supercells displayed mid-level mesocyclones on average 4 volume scans before tornado touchdown. The mid-level mesocyclone duration averaged 9 volume scans, ranging from 4 to more than 19 volume scans. The average depth of rotational velocity of at least 30 kt was 29,700 ft, ranging from 16,900 to 38,800 ft.

A qualitative evaluation of an operator's defined mesocyclone intensity was determined using the WSR-88D Operational Support Facility's (OSF) criteria (NWS 1995), which is based on the rotational velocity and range of a velocity couplet. Ninety-four percent of the mesocyclones attained strong intensity before tornado touchdown. The other 6 percent of the mesocyclones were at least moderate intensity.

The WSR-88D's mesocyclone algorithm detected a mesocyclone 80 percent of the time, on average 3 volume scans before touchdown. When a mesocyclone was not detected by the algorithm, in 1 case a 3-D correlated shear was detected, and in the other 2 cases the supercells were 105 to 111 n mi from the RDA. The algorithm does not detect mesocyclones beyond 110 n mi due to height considerations (U. S. Department of Commerce 1991).

Shear defined as

averaged a maximum of 18 x , ranging from 8 x to 31 x .

Velocity couplets associated with mesocyclones had an average azimuthal core diameter of 3.6 n mi. Thirty-one percent of the time the core diameter appeared to decrease shortly before tornado touchdown. However, in 63 percent of the cases, no trend was evident in the data.

3.2 Nonsupercells

The maximum rotational velocity before tornado touchdown averaged 34 kt, ranging from 21 to 46 kt, and occurred on average 2 volume scans before the tornado-bearing volume scan. Fifty percent of the nonsupercells displayed maximum rotational velocity of at least 35 kt. The average height of the maximum rotational velocity was 11,200 ft, and ranged from 2,900 to 20,400 ft.

Forty percent of the nonsupercells displayed mid-level mesocyclones on average 4 volume scans before tornado touchdown. These mid-level mesocyclones averaged 4 volume scans in duration, ranging from 1 to at least 6 volume scans.

Twenty percent of the nonsupercells did not produce a rotational velocity of at least 30 kt. Of the other 80 percent, the average depth of 30 kt rotation was 19,300 ft, ranging from 7,200 to 33,300 ft.

When the operator's mesocyclone intensity rating was applied before tornado touchdown, 40 percent of the nonsupercells displayed strong mesocyclones, 30 percent were moderate mesocyclones, 20 percent were minimal mesocyclones, and 10 percent produced only weak shear.

The WSR-88D's mesocyclone algorithm detected a mesocyclone 50 percent of the time, on average between 4 and 5 volume scans before the tornado-bearing volume scan. One nonmesocyclone case displayed a 3-D correlated shear. Maximum shear averaged 11 x , ranging from 7 x to 18 x .

The velocity couplets displayed an average azimuthal core diameter of 3.0 n mi. In 30 percent of the cases the core diameter appeared to decrease shortly before touchdown, but no trend was apparent in 50 percent of the cases.

4. CONCLUSIONS

4.1 Supercells

Inflow notches and BWERs in reflectivity data were strong signatures associated with the tornadic supercells. Storms with BWERs were often topped by reflectivities in excess of 60 dBZ at higher elevations. Maximum reflectivities were usually above 20,000 ft. Although VILs were usually exceptionally high, the trend of decreasing VILs continued to be a strong clue in more than half of the supercells when they were becoming tornadic.

From this data set, it appeared that 80 n mi was about the maximum range to detect a hook. Looking for a hook beyond 80 n mi should be considered a fruitless endeavor. The hook echo was rarely detected in this data set where the average range to supercells was 84 n mi.

The typical supercell in this study contained a strong mesocyclone, with a mid-level mesocyclone extending to nearly 30,000 ft. The WSR-88D mesocyclone algorithm usually detected the supercell's mesocyclone, but it must be kept in mind that the present algorithm does not work beyond 110 n mi.

4.2 Nonsupercells

Ninety percent of the nonsupercell tornadic events were associated with squall lines. Eighty-nine percent of these squall-line tornadoes were associated with cells that were either at the northern or southern end of the squall line, or were associated with bowing segments along the squall line.

The reflectivity signatures were much more subtle than those associated with supercells. Furthermore, seventy percent of the time reflectivity was decreasing prior to tornado touchdown.

Velocity signatures showed more promise, with 50 percent of the nonsupercells displaying a rotational velocity of at least 35 kt. Forty percent produced mid-level mesocyclones, and 40 percent produced strong operator-defined mesocyclones.

4.3 Closing Comments

This study clearly showed that the WSR-88D excels in depicting the three-dimensional structure of severe storms. All supercells that traversed northern Alabama in this study were depicted with clear signatures and clues to their tornado potential, often at extended ranges of 80 to 115 n mi. It is fortunate that severe thunderstorms representing the greatest threat to life and property are so well detected by the WSR-88D.

Nonsupercell tornadic thunderstorms represent a much greater warning challenge than supercells. For these cases, velocity data provided major assistance in drawing out clues to potential tornadic nonsupercells. Caution must be used, however, since there are usually several velocity couplets associated with squall lines making it difficult to differentiate between severe and nonsevere storms.

This study concentrated on the quantification of specific radar features associated with 25 tornadic thunderstorms set in the southeast United States. It was abundantly clear in the study that no one radar-detected element - whether base product or derived product - was a universal identifier in the identification of a tornadic event. Often the strongest signatures were not evident at the lowest elevation scan but at higher elevations. Jumping from the age of nearly one-dimensional radar detection, with the WSR-57 and WSR-74, to a volume scanning strategy has proven invaluable in uncovering storm structure detail. It is imperative to evaluate the three-dimensional structure of thunderstorms by analysis of WSR-88D data from multiple elevations.

NWS meteorologists have high-resolution, three-dimensional reflectivity and velocity data at their disposal for warning purposes. How this data is used will determine the effectiveness of the WSR-88D as a warning tool. Warning is an integrated process. It starts with a meteorologist who must possess a full understanding of the current atmospheric environment and its potential evolution as well as a strong conceptual model of storm structure. This combined knowledge is background necessary to anticipate the development of severe thunderstorms. Once thunderstorms develop, the warning meteorologist has to quickly and efficiently use the WSR-88D to its fullest capability to effectively warn for severe thunderstorms.

References

Lemon, L. R., and C. A. Doswell III, 1979: Severe thunderstorm evolution and mesocyclone structure as related to tornadogenesis. Mon. Wea. Rev., 107, 1184-1197.

Murphy, R. A., K. J. Pence, J. A. Westland, and R. E. Kilduff, 1995: A comparison study of VIL vs rotational velocity associated with tornadic thunderstorms. Postprints, The First WSR-88D Users' Conference, Norman, Oklahoma, OSF/JSPO, 259-265.

National Weather Service, 1995: The WSR-88D operator's guide to mesocyclone recognition and diagnosis. WSR-88D Operational Support Facility, Norman, OK., 111 pp.

Peters, B. E., and R. E. Kilduff, 1993: Early impressions of the east Alabama WSR-88D. NWS Southern Region Technical Attachment, SR/SSD 93-23, Fort Worth, TX., 5 pp.

U.S. Department of Commerce 1991: Federal Meteorological Handbook No. 11, Part C-WSR-88D products and algorithms. Office of the Federal Coordinator for Meteorological Services and Supporting Research, Washington, D.C.

Back to home page

Weather Hazards | Graphical Forecast | Present Weather
Past Weather | Tropical Weather | Contact Us