DAMAGE AND RADAR ANALYSIS OF THE NASHVILLE, TN TORNADO
by Timothy P. Marshall, Haag Engineering Co., Dallas, Texas and
Tim Troutman, National Weather Service, Melbourne, Florida
On April 16, 1998, three supercells produced a total of ten tornadoes in central Tennessee. One of the tornadoes went through downtown Nashville. Within days after the event, the authors conducted ground and aerial damage surveys at the request of NOAA headquarters. The purpose of the damage survey was to: 1) identify the tornado tracks, and 2) to determine the F-scale damage ratings by studying building performance.
Several findings were made with regard to our damage survey of the Nashville tornado: 1) the damage path was quite wide, reaching 1200 meters and extending 24 km in length, 2) the tornado produced mostly F-0 and F-1 damage consistently throughout its path, and 3) a few areas of F-2 and F-3 damage were attributed to deficiencies in building construction rather than a sharp increase in wind velocity. It was apparent that the tornadic circulation remained quite broad and relatively weak throughout itís life. There was no indication of subvortices in the damage.
A study of the radar imagery was made to determine the characteristics of the supercell which spawned the Nashville tornado. Radar images were obtained from the Nashville WSR-88D located 5 km northeast of the end of the damage path. Several findings were made with regard to the analysis of the radar data: 1) the supercell responsible for producing the Nashville tornado developed about 160 km west of the city during the early afternoon, well over an hour before the city was struck, 2) radar reflectivity in the lowest elevations show a "classic" supercell structure with "flying-eagle" echo configuration and well defined hook echo, and 3) there were relatively strong gate-to-gate shears averaging 50 knots during the time of the tornado.
2. SYNOPTIC ENVIRONMENT
The synoptic situation was classic for severe weather over Tennessee on April 16, 1998. At 1500 UTC, a surface cold front had moved into northwest Tennessee. Ahead of the front, southerly surface winds transported gulf moisture northward; dewpoint temperatures reached 19 C in Nashville. Skies over central Tennessee remained partly cloudy throughout mid-day, allowing surface heating to occur.
At 1200 UTC on April 16, 1998, a positive tilt height trough was approaching the area from the west. At 850 mb, a 40 knot jet was located just east of Nashville; the axis of the jet extended southwestward to Jackson, MS. The 850 mb thermal ridge extended over Nashville. At 700 mb, there was a 60 knot jet over Nashville with the axis of the jet extending westward toward Little Rock, AR. Dry air had moved into western Tennessee with relative humidities less than 50%. However, the Nashville sounding remained saturated up to 700 mb.
Analysis of the 500 mb map at 1200 UTC revealed a broad jet extending southwestward from Tennessee to New Mexico. Wind velocities up to 80 knots were located upstream of Tennessee, about 30 knots higher than at Nashville. The 500 mb air temperature at Nashville was -11 C with colder air located to the west. The 500 mb air temperature at Amarillo was -20 C. At 300 mb, a 110 knot polar jet extended from Oklahoma to Michigan whereas a 110 knot subtropical jet extended from Texas to southern Alabama. This split flow resulted in a very diffluent area over central Tennessee. A composite map showing significant weather features on this day is shown in Figure 1.
Various stability indices were analyzed for this event. A modified sounding at 2200 UTC revealed the SWEAT (Severe Weather Threat) index was near 500. The total totals index was 55 and the lifted index was -7 C. CAPE (Convective Available Potential Energy) was around 1600 J kg-1. The EHI (Energy-Helicity Index) was 2.7 which more than doubled from its 1.3 value at 1800 UTC. Winds in the lowest 3 km veered 60 degrees and increased from 14 knots at the surface to 54 knots at 700 mb. However, surface winds backed and became more easterly with the approaching storm resulting in a 160 degree change in wind direction in the lowest 3 km.
At 1200 UTC, the SPC (Storm Prediction Center) issued a moderate risk for central and eastern Tennessee and upgraded the area to a high risk by 1530 UTC. Forecasters specifically cited that, although unidirectional flow existed above Tennessee, right moving storms would have enhanced tornado potential.
3. ANALYSIS OF RADAR DATA
The WSR-88D radar was located in Old Hickory (KOHX), just east of Nashville. Reflectivity and radial velocity images were obtained at six minute intervals for the event. Analysis of these radar data revealed that the supercell which produced the Nashville tornado developed about 160 km to the west of Nashville. Base reflectivity scans in the lowest elevations revealed a "classic" supercell structure with "flying eagle" echo configuration and well defined hook echo (Fig. 2). The mean storm motion was 245 degrees at 45 knots which was slightly slower and to the right of the 500 mb environmental winds. A strong mesocyclone with TVS (Tornado Vortex Signature) was detected at 3:29 p.m. just as the Nashville tornado formed (Fig 3).
Within five minutes, the rotational velocity increased from 45 to 55 knots and the rotational shears increased from 2.5 x 10-2 to 7.5 x 10-2 per second. (Fig 4). Rotational velocity and shear values dropped significantly as the mesocyclone approached the radar station and entered the stationís ground clutter. Relatively strong gate-to-gate shears averaged 50 knots during the tornado. The hook echo eventually moved into the forward flank area around 4 p.m. just after the tornado had lifted and passed south of the station.
4. DAMAGE SURVEY METHODOLOGY
The authors realized it was essential that the damage survey begin as quickly as possible after the disaster in order to preserve evidence. A meeting was held at the Nashville Weather Service the morning after the disaster in order to establish logistics of the survey. Hard copy radar images and newspaper accounts were gathered to determine the locations of potential damage paths. An aerial damage survey was scheduled to be done first in order to quickly establish the locations and number of tornado damage paths. However, inclement weather conditions prevented us from conducting the aerial survey until three days after the event. Therefore, the ground survey was conducted first by driving through the damage area. Since numerous downed trees and power lines prevented us from driving down each road, the authors proceeded in walking portions of the damage path in the rain. Similar methodologies have been described by McDonald and Marshall (1984) and Bunting and Smith (1990) for conducting damage surveys.
It was important to have proper equipment in conducting the damage survey. Detailed road maps were obtained before the authors began surveying the damage. Still cameras with both print and slide film were used to photograph the damage. A wide angle lens on one camera captured the overall damage scene whereas a zoom lens on another camera captured specific details. Also, bringing a second camera along was also a good idea in case one of the cameras malfunctioned. Notebook paper, a clipboard, and pens were brought along for documentation purposes. House-by-house F-scale ratings were plotted on paper for each block. A tape recorder was utilized to record the locations of the photographs as well as record pertinent observations. A tape measure was helpful to determine the distances between objects and obtain dimensions of building components. Proper identification also was needed to enter the disaster areas. Hard hats provided some safety from falling objects.
A high-winged, single-engine aircraft was employed for the aerial survey with pilot from the Civil Air Patrol. The aircraft was flown between 1 and 2 km above the ground in overlapping circles which paralleled the damage path. Clearance of the air space had to be obtained from air traffic control due to our close proximity to the Nashville airport. Numerous photographs were taken of the damage path. The best perspective was obtained when photographing directly above the damaged buildings. In many cases, specific buildings were be identified and served as landmarks. This was especially true for churches, schools, and hospitals. Numerous trees were downed as a result of the tornado which made identifying roads easier. The aircraft also had a GPS (global positioning system) digital display onboard which helped confirm our location especially when roads or other landmarks were not available.
5. GROUND SURVEY RESULTS
The Nashville tornado first touched down west of downtown around 3:30 p.m. about one mile west of Charlotte Pike and I-440. Minor damage was noted to roof coverings and a number of trees were uprooted. One person died when struck by a falling tree at Centennial park. The tornado entered downtown Nashville around 3:40 p.m. causing extensive window glass breakage to several high rise buildings and collapsing several older masonry buildings. The Nations Bank Office Towers were hardest hit. Over 100 windows were blown out of the Tennessee Performing Arts Center.
The tornado crossed the Cumberland river striking the new Tennessee Oilerís Football stadium that was under construction. Three of the six tower cranes were toppled at the construction site. A video shot by local police showed a large rotating wall cloud/tornado with no condensation at the ground. The tornado continued eastward traveling through a light commercial and residential area. Numerous trees were uprooted and several trees fell onto buildings. F-scale damage ratings were assigned to 1500 residences in east Nashville. Most of the homes sustained F-0 to F-1 damage. Only five homes had their roofs removed and were rated F-2. A site examination of these locations revealed minimal attachment of the rafters to the wall top plates. The tornado was rated F-3 due to the old collapsed masonry buildings in the downtown area. These buildings had roofs that were uplifted and dropped suddenly causing the collapse of the masonry walls. Such buildings were inherently weak and had minimal resistance to uplift loads. Metal buildings sustained damage to end walls and overhead doors. Two steel transmission towers were downed. The authors concluded the damage was consistent with peak wind velocities around 40 to 50 m/s. Such wind speed/damage correlations have been done previously by Mehta et. al (1981) and Mehta (1986).
The tornado continued east striking the Cornelia Fort Airport. Around 30 private planes were destroyed amounting to about three million dollars in damage. Some of the planes were tossed around including one which impacted a metal hangar and another plane that was deposited in the bed of a pickup truck. The tornado continued eastward toppling trees as it traveled over a golf course at the Opryland Amusement Center before lifting around 3:54 p.m. The Nashville tornado traveled about 24 km during the 24 minutes it was on the ground, averaging 1 km per minute.
6. AERIAL SURVEY RESULTS
Fujita and Smith (1993) have demonstrated that near ground wind fields can be inferred by looking at damage patterns from the air. They have shown that aerial surveys were essential in order to determine details like cycloidal marks and suction vortices. The authors were able to fly over the damage path and quickly establish the details of the path. Hundreds of photographs were taken during the aerial survey; the locations of which were identified later by using road maps and landmarks. Since housing damage was relatively minimal, the aerial photographs were employed to conduct a tree fall analysis in order to deduce the near ground wind field. The directions of 489 trees were plotted on a street map (Fig 5). We found that the trees fell in line with the general large-scale circulation with strong convergence near the center of the damage path. No evidence of subvortices was found in the tree damage.
This event was unique from several aspects. The tornado struck a downtown area. The tornado had a broad circulation that reached 1200 meters in width and did not increase or decrease in intensity throughout most of it 24 km path. It was not known why the tornadic circulation did not tighten given the classic appearance of the hook echo with gate-to-gate shears approaching 50 knots. It was interesting to note that this storm only deviated slightly to the right and traveled slightly slower than that of the 500mb winds.
We would like to thank NOAA Headquarters for opportunity to study this event as well as personnel at the National Weather Service in Nashville, especially Derrel Martin the Meteorologist in Charge who provided the weather and radar data. The Civil Air Patrol volunteered their aircraft and pilot for us to conduct the aerial survey.
Bunting, W. F., and B. E. Smith, 1990: A Guide for Conducting Damage Surveys, unpublished manuscript. 44 pp.
Fujita, T.T. and B. E. Smith, 1993: Aerial Survey and Photography of Tornado and Microburst Damage, The Tornado: Itís Structure, Prediction, and Hazard, American Geophysical Union, p. 479-493.
McDonald, J. R., and T. P. Marshall, 1984: Tornado Damage Documentation, IDR Publication, Texas Tech University, 27 pp.
Mehta, K.C., R. D. Marshall, and T.A. Reinhold, 1981: Wind Speed-Damage Correlation in Hurricane Frederic, American Society of Civil Engineers Conference, St. Louis, MO, 14 pp.
Mehta, K.C., 1986: Windspeed estimates: Engineering Analysis. Proceedings of the Symposium on Tornadoes, Texas Tech University, Lubbock, pp. 89-103.