On May 18, 1995, the atmosphere over Tennessee was primed for an outbreak of tornadic supercell thunderstorms. Two primary supercells which originated in southwest and middle Tennessee tracked eastward through the state, reaching the county warning area of NWSO Knoxville/Tri-Cities (Fig. 1) by late afternoon. The Knoxville/Tri-Cities WSR-88D (KMRX) performed flawlessly during this first tornado outbreak to affect east Tennessee since the radar was accepted in July 1994. The outbreak is examined in depth in the following sections.
Observed regional upper air plots were not available for a post-analysis of this particular tornadic supercell event, but initialized meteorological fields from PCGRIDDS were available from both the Eta and NGM models. Smoothed from observed data, these initialized fields valid at the synoptic times of 0000 and 1200 UTC were assumed to be close to the observed fields at the same valid times.
Prior to the onset of severe convection at 1100 UTC on May 18, 1995, a surface plot (not shown) depicted a 998 mb low in southwest Missouri. An associated warm front extended east from the low through sections of Illinois, Indiana, Ohio, and Pennsylvania. A trailing cold front stretched southward from the low through extreme southeast Oklahoma, northeast Texas, and south-centra Texas. Later that evening at 0000 UTC the surface low had weakened slightly to 1000 mb and had moved to a position close to Evansville, Indiana. The cold front was close to the Mississippi River. For the entire day on May 18, east Tennessee remained well within the warm sector of the surface low.
Conditions at 0000 UTC were examined in depth, since most of the severe convection formed close to this time. A low-level jet (LLJ) averaged around 45 kt at 850 mb across east Tennessee. This LLJ had been strengthening throughout the day and continued strengthening even after 0000 UTC. As a result there was weak warm air advection and an increase in low-level moisture over east Tennessee, reflected at the surface by dewpoints rising into the mid to upper 60s and temperatures rising into the upper 70s to low 80s. The effects of the LLJ combined with daytime heating to destabilize the atmosphere over east Tennessee.
In the mid and upper levels at 0000 UTC, conditions were also favorable for upward vertical motion. At 500 mb a positively tilted full latitude trough extended from southeast Wisconsin to east Texas. A negatively tilted shortwave trough axis extended from south-central Kentucky through southeast Tennessee into eastern Georgia. This resulted in positive vorticity advection (PVA) at 500 mb across all of east Tennessee. In fact, PVA increased all day as both the short wave and long wave trough axes moved closer to east Tennessee. The right entrance region of a 110 kt jet streak at 300 mb was located across east Tennessee during the afternoon and evening hours on May 18. As a result, strong upper level divergence was also present.
The warm sector of the surface low provided moderately unstable air over east Tennessee. Surface-based lifted indices computed by the AFOS Data Analysis Programs (Bothwell 1988) during this event averaged around -6C. A similar lifted index was obtained from PCGRIDDS from both the Eta and NGM models valid at 0000 UTC. The air across east Tennessee remained essentially undisturbed since no convection occurred prior to the tornadic supercells.
A representative hodograph (Fig. 2) of the tornadic supercell environment was constructed using the KMRX WSR-88D VAD Wind Profile (VWP) in a SHARP environment (Hart and Korotky 1991). The wind data used for this hodograph were from approximately the same time one of the tornadic supercells passed within one mile of the KMRX radar. A mean storm motion of 260 deg at 44 kt was determined by analyzing the WSR-88D storm track product (STI) from numerous volume scans. This storm motion fit the profile for a supercell very well since it was close to 20 deg to the right of and 75 percent of the mean wind. The hodograph curvature also fit the profile for a supercell. Consequently, the rotational potential in the form of storm-relative helicity was certainly favorable for tornado development.
It is often desirable to look at the combined effects from both instability and storm-relative helicity. The energy-helicity index (EHI) combines these two into a dimensionless index which describes in a simplistic manner the optimal combination of the two values (Davies 1993). The EHI for east Tennessee for this event averaged around 4.5. Davies (1993) found that EHI values over 4, along with sufficient mid-level winds and storm-relative inflow, suggest that violent tornadoes are possible.
Over the past few years authors such as Davies-Jones et al. (1990) and Johns et al. (1993) have shown that storm-relative helicities over 400 m2/s2, along with adequate instability, can significantly enhance the potential for strong and even violent tornadoes. The strongest tornado to affect east Tennessee on the evening of May 18, 1995, was rated F3 on the Fujita scale. Although this was not a violent tornado, it was close enough to give validity to environmental parameters such as EHI and storm-relative helicity.
First Supercell Track: Cumberland County to Hamblen County
The first tornadic supercell to affect east Tennessee on May 18 had its origin in southwest Tennessee, several hundred miles away. It was first tracked by the KMRX Storm Track (STI) product near Columbia, Tennessee, approximately 180 nm from the east Tennessee counties affected by the tornado. The track of this supercell cut through the heart of the county warning area (CWA) of the NWSO Knoxville/Tri-Cities office (Fig. 1).
At approximately 2307 UTC (707 P.M. EDT) on May 18, the storm moved into western Cumberland County, the westernmost county in the CWA. A moderate mesocyclone in the mid-levels of the storm, a well defined V-notch, and a weak echo region (WER) were already being detected. Five minutes later, the mesocyclone was identified as strong, since rotational velocities exceeded 40 kt at a range of more than 80 nm from the radar. At this time the supercell was approximately 5 nm west of Crossville, Tennessee (CSV), in central Cumberland County. This was also the first time that a low-level appendage was observed by the KMRX radar. It is believed the low-level appendage was present well before it tracked into Cumberland County, however, since this supercell had a history of producing tornadoes earlier in middle Tennessee.
As the supercell moved through the city of Crossville between 2317 and 2322 UTC, an F3 tornado accompanied it. Approximately 30 minutes prior to this, a tornado warning was issued for Cumberland County. The storm damaged or destroyed more than 30 homes and injured 20 people. The mesocyclone strengthened, but still remained a strong mesocyclone. Its vertical depth increased from 10,000 to 16,000 ft. A marked increase was also noted in the echo tops and VIL values as they increased to 51,000 ft MSL and 72 kg/m2, respectively.
After the tornadic supercell moved through Crossville, its radar reflectivity characteristics began to resemble a classic supercell. Approximately 3 nm east of Crossville, a hook echo was first detected. Earlier when the supercell was moving through Crossville, its WER became bounded. Thus, by 2327 UTC the storm had the radar characteristics of a classic supercell (hook echo, BWER and V-notch). A deep, strong and persistent mesocyclone was also present. Figures 3 a-d illustrate these radar features, which became even more well-defined as the supercell moved from Cumberland County through the counties of Morgan and Anderson.
Tornado warnings were issued for Morgan and Anderson Counties with lead times between 15 and 20 minutes. This storm was tracked for about one hour, during which time rotational velocities averaged between 45 and 50 kt. The average depth of the strong mesocyclone was 15,000 ft, although some fluctuations in the mesocyclone depth occurred during this period. This may have been one reason why tornado damage was cyclical. Numerous reports of golf ball to softball size hail also accompanied the supercell. The corresponding VIL values ranged from the middle 60s to middle 70s (kg/m2).
The supercell entered Knox County from Anderson County around 0030 UTC and moved across the northern sections of the county before exiting the northeast corner around 0100 UTC. The classic hook echo signature that had been so notable in Anderson County was noted on the 0.50 deg elevation base reflectivity product. The 0028 UTC Storm Relative Velocity Map product (SRM) at 1.50 deg showed 50 kt of rotational velocity about 4 nm southeast of the city of Clinton in extreme eastern Anderson County. This rotation was around 7800 ft MSL and met the criteria for a strong mesocyclone. Based on past history and this strong rotation, a tornado warning was issued for northern Knox County at 0029 UTC.
The 0040 UTC SRM product at the 1.50 deg elevation angle showed 45 kt of rotational velocity about 8 nm north of Knoxville at roughly 6200 ft MSL. No reports of funnel clouds or tornado sightings were relayed to NWSO Knoxville/Tri-Cities, but 1.75-in diameter hail was reported in northern Knox County at 0030 UTC. The VIL products indicated values as high as 70 kg/m2, and the Layer Reflectivity Maximum (LRM) products for the layer from 24,000 to 33,000 ft reached 75 dBZ as the storm moved through northern Knox County.
The counties next affected were Grainger and Jefferson, with the storm entering the southern part of Grainger County around 0045 UTC and the extreme northern part of Jefferson County around 0106 UTC. The storm was moving inside a 20 nm radius of the KMRX radar, and a switch to VCP 11 had been done well before the storm entered Knox County, roughly 20 to 40 nm west of NWSO. As the storm moved into Grainger and Jefferson Counties, the radar analyst examined higher slices of reflectivity and velocity. At NWSO Knoxville/Tri-Cities routine product lists and predefined user functions have been created for use when storms move within 30 nm of the radar. These product lists and user functions have reflectivity and SRM products for 6.00, 10.00, and 14.00 deg elevation angles.
As the storm entered Grainger County, the 0051 UTC SRM product at the 4.30 deg angle (Fig. 4a) showed a strong mesocyclone with 55 kt rotational velocity about 17 nm west of the KMRX radar at 9500 ft MSL. Again, based on past history and continued detection of a strong mesocyclone, tornado warnings were issued for Grainger County and also southern Union County at 0052 UTC. The storm continued on an almost due east track, but no severe weather was observed in Union County with this storm. At 0107 UTC, tornado warnings were issued for Jefferson and Hamblen Counties, as radar continued to show up to 55 kt of rotational velocity about 10 nm west of the KMRX radar, along the Grainger-Jefferson County line. A funnel cloud was sighted in Jefferson County around 0115 UTC, and some trees were downed around 0145 UTC. Wind damage was reported in Grainger county around 0100 UTC.
As the storm continued to track rapidly east across Hamblen County, the 1.50 deg angle base reflectivity product continued to show a distinct hook echo, and the 10.00 deg base reflectivity product at 0126 UTC (Fig. 4b) showed a BWER about 10-12 nm northeast of the KMRX radar. The 0126 UTC SRM product indicated 50 kt of rotational velocity about 10 nm northeast of the radar near 13,000 ft MSL. Golf ball sized hail and a funnel cloud were reported in Hamblen County around 0123 UTC, with a possible tornado around 0145 UTC. Due to the "cone of silence" effect, VIL and LRM products were of little use until the storm moved farther east of the radar, over eastern Hamblen County. After passing out of the county the supercell began to weaken.
Second Supercell Track: Fountain City Tornado
The second classic supercell originated over the southern counties of middle Tennessee then moved rapidly east-northeast at 41 kt. By 2352 UTC (7:52 p.m. EDT), the KMRX radar indicated a mesocyclone with this storm over Bledsoe County, which is just to the west of Rhea County (Fig. 1). The rotational velocity was around 35 kt. At the same time, a strong thunderstorm was moving rapidly northeast at about 45 kt across Hamilton County in southeast Tennessee. The SRM indicated only weak shear with a rotational velocity around 16 kt and a VIL of around 30 kg/m2 with this storm. The KMRX radar indicated a merger of the two cells was possible within the next hour about 35 nm southwest of Knox County.
The supercell had a history of high VILs (in the 50s) and moderately strong rotation, so a severe thunderstorm warning was issued for Roane and Loudon Counties at 0002 UTC. It was extended for Anderson County at 0033 UTC. Trees and power lines were reported down as well as golf ball size hail. As the southeast Tennessee non-severe thunderstorm approached the supercell over Roane County from the southwest, a disruption of the inflow to the supercell likely occurred, reducing storm-relative helicity and weakening the updraft in the storm. When the storms merged around 0056 UTC, the rotational velocity of the supercell decreased to 26 kt with only uncorrelated shear depicted. From 0101 to 0131 UTC, no mesocyclone, 3-D correlated or uncorrelated shear were detected; and rotational velocities were only 16 to 20 kt. VILs also decreased from the 50s to the 30s.
Between 0136 and 0141 UTC, however, a marked strengthening of the merged cells occurred over extreme southern Knox County, with rotational velocities increasing from 15 to 30 kt and a WER becoming noticeable. A severe thunderstorm warning for Knox County was issued at 0140 UTC. Between 0141 and 0151 UTC, the storm top increased from 26,000 to 41,000 ft MSL. Also, a distinctive BWER became apparent having a "donut hole" appearance on both the 3.4 deg (12,000 ft) and 4.3 deg (17,000 ft) reflectivity products. This is shown in Fig. 4d. An appendage was noted at the 0.5 deg reflectivity slice, and rotational velocities increased to 40 kt, indicating a moderate to strong mesocyclone (Fig. 4c).
A tornado warning was issued for Knox County at 0152 UTC. By 0156 UTC, a small hook was noted over Knoxville in central Knox County on the 1.5 deg (7,000 ft) reflectivity product. Rotational velocities were between 50 and 55 kt. From 0156 to 0206 UTC, the storm top collapsed from 39,000 to 29,000 ft MSL, with the BWER becoming less distinct and finally disappearing. Rotational velocities remained between 40 and 50 kt, however, until after 0212 UTC. A tornado was sighted between 0205 and 0230 UTC over Fountain City, which is a northern suburb of Knoxville. The F1 tornado uprooted numerous trees, downed power lines, and damaged many homes in the Fountain City area.
Since the tornadic storm moved northeast at 40 kt, a tornado warning was issued for both Grainger and Jefferson Counties at 0210 UTC, and for Hamblen County at 0231 UTC. Even though the storm reflectivity lost its BWER/WER characteristics, VILs remained in the 50s. Base velocity at the 0.5 deg elevation indicated 40 to 50 kt winds across southern Grainger and Jefferson Counties around 0235 UTC. Numerous reports of trees and power lines down were reported from the area. The severe thunderstorm then dissipated over northern Hamblen County.
Overall, the KMRX WSR-88D radar performed flawlessly during the first tornado outbreak to affect east Tennessee since the radar was accepted in July 1994. Radar algorithms such as the STI, meso and VIL were extremely valuable in warning decisions. The STI was particularly accurate and helpful because the tornadic supercells were isolated and their tracks deviated little from volume scan to volume scan.
Another obvious benefit of the WSR-88D was that identifiable supercell characteristics, such as a bounded weak echo region and hook echo, could be recognized easily. Ultimately, all of this radar information translated into lead times which averaged over 20 minutes for each affected county. Since at the time of the event NWSO Knoxville/Tri-Cities did not yet have warning responsibility for all of the counties that comprise the MAR county warning area (Fig. 1), the radar information was passed on to area WSOs which issued tornado and severe thunderstorm warnings based upon the information. This resulted in warning lead times which no doubt assisted in saving lives and minimizing injuries.
Bothwell, P. D., 1988: Forecasting convection with the AFOS data analysis programs (ADAP-version 2.0). NOAA Tech. Memo. NWS SR-122, NWS Southern Region, Fort Worth, TX, 92 pp.
Davies, J. M., 1993: Hourly helicity, instability, and EHI in forecasting supercell tornadoes. Preprints, 17th Conf. Severe Local Storms, Amer. Meteor. Soc., 107-111.
Davies-Jones, R., D. Burgess, and M. Foster, 1990: Test of helicity as a tornado forecast parameter. Preprints, 16th Conf. Severe Local Storms (Amer. Meteor. Soc., 588-592.
Hart, J. A., and W. D. Korotky, 1991: The SHARP workstation -- v1.50. A skew T/hodograph analysis and research program for the IBM and compatible PC. User's manual. NOAA/NWS Forecast Office, Charleston, WV, 62 pp.
Johns, R. H., J. M. Davies, and P. W. Leftwich, 1993: Some wind and instability parameters associated with strong and violent tornadoes. Part II: Variations in the combinations of wind and instability parameters. The Tornado: Its Structure, Dynamics, Prediction, and Hazards. Geophys. Mono. 79, Amer. Geophys. Union, 583-590.