STORM TRACK: January 31, 1980 (Volume 3 Issue 2)

On September 5, 1979, the tropical storm remnant of hurricane David moved rapidly up the east coast, accompanied by several reports of tornadoes and widely scattered wind damage (mostly to trees). One of these tornadoes occurred just prior to sunset, at about 7:20 PM EDT, in a residential area of Fairfax County. An earlier tornado occurred about 3:00 PM EDT near Crofton, Maryland -one of several reported in that area. An eyewitness reported the Maryland one as a "...huge funnel. It was just sucking everything up. ...It turned over sheds and tore down fences ...The funnel looked gray. I could see it picking up stuff."

I did not do any tornado chasing as it was a working day and, moreover, I did not expect that such an overcast, precip laden environment would afford good visibility for photography. During a morning letup in the rain (about 8:45 AM EDT), I did note heavy buildups imbedded in the pervasive overcast near the Crofton area. However, after about 10:00 AM EDT, downtown Washington was basically overcast and rainy (light to moderate) all day.

Almost two weeks after the storm, I drove over to the Fairfax damage area for the first time. I was looking for any witnesses of the tornado who might corroborate my suspicion about the lack of visibility in a tropical storm/hurricane situation. I had the good fortune to meet Mr. Frank Cervi and his son, whose home was one of two that were totally destroyed. His account of the storm is included later in this study, along with other eyewitness accounts. Mr. Cervi confirmed my expecta- tion that this kind of storm environment is poor for spotting wall clouds, flanking lines or similar tornado precursors. However, his son had a surprise observation. Without any prompting on my part, he volunteered that he saw a strange "golden glow" in the sky right after the tornado. Was he seeing refracted sunlight through a low or mid- level break in an otherwise continuous cloud cover? This report continued to haunt me over the next week, and I couldn't shake the notion that it might be significant.

What it suggested was that in a tropical storm or hurricane environment, in which pervasive radar reflectivity can make tornado detection difficult, maybe the radar operator should look for the absence of reflectivity -or holes- on his scope. When the characteristically quick and capricious hurricane tornadoes" appear, maybe they are preceded by a momentary, local clearing. I assume that such small twisters do not often present the traditional "hook" echoes of larger storms. How else could they be detected? This was what spurred my study of this storm.

This particular tornado also appealed to me in several other respects. Knowing that there are far fewer studies of hurricane tornadoes than of their more prevalent great plains/midwest cousins, this storm presented a rare opportunity to examine a tropical storm twister so far north, over a populated area, and occurring in visible late evening daylight. It seemed to fit the description I had read of such tornadoes -having a short track, occurring in the midst of heavy rain, and moving with the prevailing storm environment (from the SE). I preferred this storm to the one near the Crofton area, because the latter seemed more like the traditional land based tornado (highly visible and, therefore, probably seen in an updraft, rain-free region). Also, it occurred 4 hours earlier. When the Fairfax storm began, the remnant hurricane "eye" was much nearer (about 90 miles to the SSW at 10:00 pm according to one news report). Therefore, I assumed that the Fairfax storm occurred in an environment more similar to that of a hurricane tornado.

The study was planned to cover several areas, through eyewitness accounts, on-site damage inspection and review of available NWS data. These areas were:

l. Examination of damage patterns;

2. Survey of reported storm characteristics, including sounds;

3. Assessment of sky-glow reports after the tornado; and

4. Examination of surface weather maps.

My study can be faulted for its very late start and its small and selective sampling of eyewitnesses. It was not until I had received some encouragement from acquaintances at the National Severe Storms Forecast Center and at the Miami Hurricane Center that this project was begun in earnest. The first step was preparation of a l 1/2 page ques- tionnaire (Figure 1) and its mailing to about 22 residents in and adjacent to the area of maximum structural damage. Their selection was based on earlier newspaper interviews, as well as a random selection from mail box addresses along the streets of greatest damage. My initial mailing of 17 was not made until almost a month after the storm. I requested comments on their recollections of tornado sounds, hail, etc. On reflection, the questionnaire could probably be vastly improved. However, at the time, I didn't want it to appear excessively structured and pedantic -and certainly didn't want to "turn off" my recipients. So, the request was written with this in mind: get to the point quickly, encourage precise answers without being burdensome, allow some free range of response and don't be too long. Since my budget was limited and I lacked the professional training to properly evaluate all available information, the study was limited to a small sampling. I might add that most responses were quite excellent (Figure 2). Some candidly admitted lack of memory, depending on the question. I followed up on several written responses with phoned or personal interviews. This neighborhood is mostly middle and upper middle class, and respondees were generally either professional people or married to them.

In addition to the questionnaires, I conducted several on-site inspections and noted direction of tree fall. I was careful to record only tree fall that was apparently due to the storm, and which seemed undisturbed since then. I excluded trees adjacent to roads or paths, where early displacement to permit local traffic was likely. Otherwise, I looked for trees with leaves still on (recent fall) and tree stumps pulled partly from the ground (wherever found) -indicating direction of original fall. I did not record stumps or trees with any significant weathering or showing disturbance, such as stumps partially dug up (which may have been moved). I secured copies of aerial photographs, taken within the last two years, from the Fairfax County Division of Mapping and made an exact trace of residential streets and houses. The site inspections were recorded directly onto photo-copies of the aerial pictures and finally transferred to a traced copy for the newsletter (Figures 2-5). Arrows indicate location and direction of tree fall, as well as for light or power poles. Significant structural damage such as removal of roofs or walls is shown by solid black. Location of tree fall along and between residential streets is generally accurate to within 15 feet. Tree fall toward the middle of park areas is relatively less accurate as distance from structures increased. At such times -and to the extent possible- I used line of sight triangulation from two or more residences to locate tree fall locations (This was far easier to do after autumn began to set in and most of the leaves had fallen).

I secured low level (300-400 feet) aerial photos, taken by County or Park police, presumably by helicopter and on the following day. These gave additional clues to the storm's rotational structure and path.

My data sources were upper air soundings, instability charts and hourly station reports from the World Weather Buildings radar logs and radar scope photography from Asheville, North Carolina -for 10 minute intervals from 20 minutes before to 10 minutes after the tornado; and topographic (elevation contoured) maps from the Fairfax County Division of Mapping.

Figure 1

Figure 2

Newspaper accounts on this day reported at least 6 tornado sightings from northern Virginia to Maryland. The one that I studied had an intermittent damage track of at least 2.2 miles. It totally destroyed one occupied house and one unoccupied row-type apartment, removed one roof, substantially damaged half of another house, removed several attached garages, buckled several residence walls and pulled out portions of roofs -exposing supporting roof beams. Hundreds of major trees, most 15-24" in diameter, were either snapped off 15-40 feet above the ground or were pulled out at the roots. My own conservative estimate of structural damage along this track was about $500-600,000, although total Fairfax County and Fairfax City damage from this tornado was estimated at $2,000,000 to 165 structures, two days after the storm. Fortunately, there were no serious injuries from this tornado (Mr. Cervi suffered a cut arm, extricating himself from the rubble of his home). However, structural damage was sufficiently serious that several fatalities could easily have occurred. One fatality was reported in the Great Falls area, about 12-15 miles NNE of this area. However, this location was not coincident with the Fairfax tornado track, and aerial photos of the isolated tree damage responsible for the death were not convincing of local cyclonic rotation.

Figure 3

Figure 4

Figure 5

Figure 6. (Figure 2:137/46, looking NNE)

Initial damage was noted in two chimneys topped off from residences 450 and 600 feet SE of the Cervi home (Figure 2: 144/51 &, 148/52 -- read RIGHT-DOWN for all such coordinates). An apparent outdoor shed was demolished 200 feet to the SSE. The garage of the residence immediately S (portion of house closest to street in Figure 2) sustained slight suction damage when the wood frame centerpost between the two-car garage doors was pulled out at least a foot, similarly distorting the closed doors. The next damage was stunning and selective as the Frank Cervi home was essentially destroyed, while neighboring homes less than 75-150 feet away (Figure 6) were left standing with only slight (apparent) structural distortion or minor roof damage (shingles pulled off). Trees in the immediate vicinity were generally smail, young and resilient. Few were snapped or uprooted, and almost all came through in good shape. Two attached garages were blown away NNW of the Cervi home. The one next door, opening to the NE, probably failed due to debris impact from the Cervi home. The second one across the street, opening to the S, may have failed due to blow-in of the windward doors and air-compression blow-out from severe straight winds. The two homes to the NW of Cervi and on his side of the street required substantial rebuilding of exterior walls, although I am unaware of the extent of their damage. A 2,800 lb. car was set on top of another, after ramming the garage wall at the third residence from the end, NE of Cervi. All of the residences along this street appeared to have been built in the last 10 years. Figure 7 shows the damage to the Cervi home (from newspaper photos).

Figure 7

Figure 8 (Figure 2:132/44, looking NNW)

Wind damage to the homes and garages in this area appeared to have been caused by two suction vortices. Roof holes and other damage pro- bably were the result of partial vacuums around walls or along roof lines, which is the result of an airfoil effect (i.e. strong winds sweeping around a structure -like a plane's wing- and inducing a partial vacuum on the leeward side; either lifting the wing and the plane or pulling out a wall or roof section). The latter was indicated by the leeward location of the garage centerpost, most roof holes, and damaged exterior walls of nearby homes. Similar leeward location for damage was also noted for most structural failure in the Williamsburg Square/ Lyndhurst Apartments area (Figure 2: 41/13 & 21/17, respectively). The two probable suction vortices are indicated by Figures 8 and 9; also as highlighted by the illustration in Figure 10.

Unfortunately, debris fall from Williamsburg Square lay across 4-lane Route 236 and was apparently cleared quite early, removing any further indication of possible ground debris patterns. There was circumstantial evidence of one suction vortex, however, which will be discussed later.

Figure 9 (Figure 2:126/42, looking SSW)

Figure 10 (Looking NNW)

I saw no evidence of downburst effects throughout the storm's path. The highly localized nature of extreme structural damage and the pattern of tree fall did not indicate this.

One interesting eyewitness account should be noted at this point, as an indication of both the power of the wind and for what it may suggest about the normal venting of homes. One day after the tornado, Mary Cervi recounted her experience with the stormed "Do you know what a split second is? There was no cracking; there was no giving way of beams. Just boom. That's it." What this suggests to me as a potentially serious problem, which may or may not have been the case in this instance, is that increasing efforts to insulate homes and save energy may make explosive decompression from tornadoes more likely. The use of storm windows, caulking and other measures are on the increase. This recent national trend seems to qualify the earlier engineering assessments of the structural engineers at Texas Tech University -- that, in most circumstances, the normal venting from chimneys, exhaust pipes and weep holes around windows is sufficient to equalize rapid tornadic pressure changes. As homes become more air tight, the Texas Tech engineers may need to reassess their advice in this regard, especially in high tornado risk areas. Perhaps these residents should be encouraged to return to the advice that was prevalent several years ago; i.e. slightly opening windows on the N and E side of the house when tornadoes are likely from the southwest.

The second damage area was the Woodson High School football field where eight 60 foot wooden light poles (each with 16 lights and each light with a 3 foot reflector) were snapped and clearly laid out in a counterclockwise pattern (Figure ll). A wooden press box and a small wooden ticket booth also were destroyed, and several hundred feet of cyclone fence were ripped out along the west side of the field. The latter seemed to be the direct result of maximum winds in this area, since there appeared to be little debris impact on,the fence. Aerial photos showed that the debris trail from the nearest structure, the ticket booth (several hundred feet to the SE), was far too short to reach the fence.

There were no reliable eyewitness accounts of a condensation funnel or tornado, other than reports of extreme turbulence overhead. Most reports indicated steadily increasing, wind-driven, and heavy rain right up to the tornado and less wind with light to moderate rain thereafter. There were no hail reports throughout this storm track. Only two of eight respondents in this area reported lightning and thunder (rolling or intermittent) before the tornado. There was no consistent relationship between open and closed windows and thunder reports. It should be noted that one of these reports was from a meteorologist resident. Two of six respondents from Whitacre Road to the Lyndhurst Apartments also reported lightning and thunder before the tornado. Therefore, it appears at least possible that these events occurred, as noted. There were no reports of lightning or thunder after the tornado.

Figure 11 (Figure 2:102/39, looking SSW)

More striking were the widely varying reports of tornado sounds. Six members of the Prank Cervi family heard no warning roar prior to their home's destruction. Yet, paradoxically, a neighbor across the street (Figure 2: 132/47) - who works for the FAA and is familiar with jet engines - reported a rapidly intensifying, brief but steady "very loud jet engine" roar. This sound faded rapidly thereafter. Neither respondent on either side of the FAA employee reported more than "loud wind noise" or "in- crease in volume" of wind sound. The meteorologist about 400 feet ENE of the FAA employee (Figure 2:130/52), who reports he can normally hear street sounds through closed windows (closed at the time of the storm), reported no roar. He was standing outside the back of his home within two minutes of the tornado. Two other reports were received by adjacent residents within 200 feet S and W of the Cervi home, describing a "roaring sound" or a "roaring truck." As with the thunder reports, there was no significant relationship between sound perceptions and closed or open windows. Of notable interest is that some windows on the back side of the Cervi home were open, yet no roar was heard -and this residence sustained the most damage (possibly this destruction was due to the direct strike of a suction vortex, in which case open windows may not have helped much anyway). Mr. Cervi reported prevailing winds from the SE up to the time of the storm.

Rapid pressure changes were noted by residents immediately E and S of the Cervi home as well as by the Cervi family (ears "hurt," "clogged up," or "popped painfully").

The incongruity of tornado "roar" reports on three sides of the Cervi home but no report from the latter suggests one possible explanations The tornado sound was heard seconds after the Cervi home was damaged, perhaps as maximum wind damage touched down for the second time somewhere between the ticket booth and the west fence to the high school football field. The Cervi family could well have been in a state of shock, freeing themselves from the debris of their home, and failed to hear the subsequent roar heard by others.

A normal assumption is that a tornadic sound precedes the most severe damage. However, this experience indicates that destructive wind rotation can precede the sound. This could be due to two reasons: (a) Critical wind speeds for a tornadic "roar" had not yet been attained and occurred over the football field, or (b) destructive rotation began almost simultaneously throughout the storm cell and preceded the sound of the roar, originating in the middle of the cell where maximum wind shear was occurring. This is obviously quite speculative and generalizing from the particular. If any Storm Track readers have had similar experiences or are aware of studies in this area, please let me know.

The data thus far suggests initial touch-down of a small F-2 tornado along a 500 foot damage track from the Cervi home to the NW. At least two suction vortices were indicated in this area, which fortunately either skirted or passed between other residences NW of the Cervi's.

Before leaving the Baccarat area, two very interesting reports bear review: one on a striking incident of clockwise rotation and the second regarding possible evidence of an outflow gust-front several miles ahead of the developing tornado.

The clockwise (anti-cyclonic) rotation was reported in specific detail by the resident immediately W of the Cervi home. In two different reports to me, a clockwise pattern of small, fallen trees (about 100 feet in diameter) was shown in the backyard 150 feet NW of the Cervi home. The resident illustrated at least 6 trees (Figure 12 reproduces the second report) with arrows specifically noting "the direction of top of tree when down flat." Also of interest is that about 50% of the Cervi roof wound up in the middle of the tree circle. In subsequent visits, I was unable to independently verify this, since the trees were either already removed or had been propped up. However, I subsequently noted slight indication of anti-cyclonic rotation in the eastern edge of the tree damage area in Daniels Run Park (Figure 4: 53/35). Thorefore, it would appear that on the boundary areas to the more severe damage from this storm, some clockwise vorticity did briefly form and was strong enough to be damaging.

Figure 12

Evidence for an outflow-gust-front was given by the meteorologist, who reported the following: "Just prior to the onset (1 min. or less), a grey wall like fast moving fog swept east to west across the Robert Frost soccer field"(Figure 2: 120/52). "Tree tops began to bend." Subsequent discussion with this witness -in his backyard- revealed that about three seconds after sighting this "wall," extending from the ground to cloud deck overhead, a moderate-sized tree (12-15" base diameter) was pulled down across his yard. Direction of fall was to the WNW. Apparently, at about the same time and in the cul-de-sac on which his home fronts, a second, smaller tree was also bent over and the corner of a garage door pushed in. These three damage reports are not enough to characterize the nature of the wind, but all three enscribe a slightly southward curving track about 250 feet long. The meteorologist suggested the possibility of two funnels in the area, which corroborates a conclusion that I arrive at separately, with different evidence, later in this study. However, it is unlikely that the gray wall was a second tornado. The Robert Frost school was not damaged, and a 35 foot concrete light pole on the north edge of the field -with two 3 foot diameter light reflectors- was still standing (a dated installation tag on the pole read "1973"). After the larger tree fell, this witness went in the house and checked on possible damage (there was none). He left the house again 2-3 minutes later to survey other possible damage, at which time the lights went out.

At this point, it is relevant to estimate the forward speed of the tornado. Based on 7 reports, electric power was lost at about 7:20:45 PM EDT. An above-ground power line runs from Route 236 along the east side of Whitacre Road, south to the first curve, where it goes underground. To estimate the forward speed of the storm cell, I had two references: (1) At 7:35 PM EDT (2335 Z), the radar log from Patuxent River, Maryland reported average cell movement at 15 knots from 160 deg.(2)

Thirty minutes after the Whitacre power failure, a tornado was reported on teletype 14 miles to the NW in Sterling Park, Virginia, which is almost a perfect linear extension of the Fairfax damage track. I decided to go with the latter report, which results in a forward tornado speed of just under 28 MPH. Measuring backward from Whitacre Road, a two minute lead would place the developing tornado about 1/2 mile SSE of the Cervi home at the time of the "grey wall."

Figure 13

The wall, which blocked out view of the school, was apparently an outflow-gust-front from the tornadic storm cell, compressing rain ahead of it and wrapping around the cyclonic circulation (Figure 13). Since the meteorologist's view of the playing field was limited to about 10 deg of arc to the N and 2-3 seconds elapsed between this sighting and initial wind damage in his backyard, the gust-front was apparently already oriented slightly NNW-SSE over the soccer field, at this time (Figure 13). Assuming a normal arcing gust-front, the leading edge was at least two miles ahead of the beginning tornado.

----Examination of Damage Patterns

Most damage along the west side of Whitacre Road was confined to trees and structures exposed to the uninterrupted wind flow across the cleared, treeless football field. Generally, this damage extended 200-400 feet west of the street. Figure 3 shows tree damage in this area (where direction of tree fall could be determined) arid the most severe structural damage to the home of Paul Mayo (Figure 3: 82/27). Except for the area adjacent to Whitacre Road, the damage along Trapp Road was negligible. Tree damage was generally limited to older and larger trees, either uprooted or snapped off at the trunk 10-40 feet up, depending on distance westerly of Whitacre. Tree fall patterns in this area were generally typical of what was found later, with most trees laying from 300 (NW) to 245 (SW). Greater extremes of turning were noted in areas of more extreme structural or pervasive tree damage.

One interesting observation should be pointed out at this time. The Mayo's reported that a 60 foot oak tree, about 20" diameter at the base and '75 feet ESE of their house, was bending over at a 45 angle to the NW -about 10-15 seconds prior to their roof separation (After noting the tree inclination, they had time to go downstairs to the basement before the roof failed). Therefore, it appears that damaging straight winds were occurring immediately ahead of the tornado. Half of the respondents reported increasing winds prior to the tornado, which tends to confirm the Mayo's experience. The other half either reported other wind characteristics or made no report; none contradicted the previous observations. The significance of this is to damage surveys made of future tornadoes from hurricanes or tropical storms. Examination of aerial photos immediately after such storms may indicate a disproportionate amount of tree fall along the general axis of the storm's path (NW in this instance), as opposed to normal tornadic tree fall patterns. Allowance should be made for such preceding damaging winds in trying to characterize and identify specific tornadic circulation.

The Mayo's home sustained heavy roof damage on the windward side of this E-W oriented structure. The eastward-facing picture window was blown in and the eastern half of the vaulted (cathedral type) roof was lifted off. The portion of roof that failed was built with 4" X 8" X 16 foot beams and was characteristic of this well constructed, privately built home. National Weather Service surveyors who visited the area on the following day told Mr. Mayo that his home would have been destroyed had it been built like other development homes in the area. The NWS team also noted a patch of flattened grass about 20-30 yards SW of this residence, 10-15 yards in diameter, which they said resulted from touchdown of a tornado. I was skeptical of this for two reasons: (1) Two 60 foot aspens remained standing with limbs intact in the center of this grass area, and I don't believe that even the strong aspen could survive so well; and (2) As Mr. Mayo suggested -and I agree- a concentration of debris from the south carport lay in the SW corner of his yard and could have slid across this part of the lawn, flattening the grass. No other serious structural damage was seen along Whitacre.

There were some additional and novel damage reports, characteristic of tornadoes. The resident at one home (Figure 2: 67/27) reported a piece of the Cervi's interior wall -with matching wallpaper- was blown into her home. The Mayo's reported a Red Cross hook and a photograph of one of the Cervi children landing in their backyard. Also, a 100 lb. door from a dempster-dumpster at the western end of the football field was deposited on the north side of their home. Two cars were blown over and one uprighted again in front of the third from last residence at the south end of Whitacre. Also, one resident on Tramp Road (Figure 2: 51/19) reported that his small frame house was lifted about 1/2" from its foun- dation, leaving a crack around the masonry/concrete base. There may have been other similar damage, which my limited investigation did not uncover. There was no reasonably conclusive indication of suction vortex activity in this area.

From other responses to the questionnaire along this road, the resident just N of Mayo said that the wind "increased in volume to a roar." The resident in the fourth house N of Mayo reported a "freight train roar." NW of here, the resident whose home was lifted reported a jet airplane sound as it passed over. However, two other reports from the north end of Whitacre and from the third home at the south end reported a much different sounds "it hummed, the whole house buzzed" (north end); and "buzzing saw or like a million bees (south end). The previous reports approximate the central axis and outer boundaries for the storm's path, suggesting that the center roared while the edges buzzed.

Pressure changes were generally noted south of Trapp Road. The most interesting of these was from the residence where the cars were overturned. The respondent was standing outside, next to a creek about 300 feet behind (west) of the house. Respondent reported no sensible pressure change, whereas the family in the house did. Their sensation was described as like "taking off in an airplane or going up a mountain." The boundary zone for abrupt pressure fall was apparently just west of this residence.

Figure 14

The most severe structural damage was done to the Williamsburg Square Apartments, 8 small group of row-houses, where one residence was unroofed and a second demolished (Figure 14). The latter was at the leeward end of a short N-S line of structures. Just south of it, was a mansard roofed residence, whose second floor eastward-facing side (not shown) was badly battered and partially collapsed. Also, note gravel streaks on the left-endmost apartment roof, from the flat graveled roof building on the extreme left. This observation, plus a local report that the separated roof blew off to the NE and close study of the general damage pattern to the N of this location, indicated an are of destruction strongly suggesting a third suction vortex (Figure 15). Some of the most extreme angles between fallen trees were noted in this area, both with respect to adjacent trees (84 deg) and regarding a maximum angle of 157 deg (NNW-SSW) between trees 240 feet apart. Possibly the most severe tree damage was done just across Route 236 on the left of the entrance to and in front of the Lyndhurst Apartments. Half a dozen trees were snapped off at the trunk (18-24" diameter/20 to 30 feet up). One three foot diameter monarch was snapped 20 feet up, across the street and to the right of the entrance road. The remaining trees along the service road immediately in front of the apartments were most- ly uprooted, with limbs and smaller scrub growth piled up in a dense -matted and intertwined- tangle. Note the roof holes pulled out on the west side of the apartment roofs (Figure 16), leeward to the prevailing flow of damaging winds -from the SE. Scattered tree fall but lesser structural damage continued northward along Estel Road, as tornadic winds aimed at a fourth apparent "touchdown" in Daniels Bun Park. However, the most severe wind damage had already occurred along Route 236, at almost the highest point (elevation) in the immediate area (Figure 17). Survey of Storm characteristics

Figure 15

Figure 16

According to a newspaper account, one eyewitness in this area did report a small funnel cloud (or a suction vortex) touching down near the crest of a hill along Route 236 (Little River Turnpike). A Lyndhurst Apartment resident reported: "It was roaring. Trees were going around in circles, popping out of the ground." Only two questionnaires were mailed to residents in this area, with one response. No other information is available on the storm in this area.

Figure 17

----Examination of Damage Patterns

As already noted, damage along Estel Road appeared to be light to structures. Possibly 15-25 freshly cut stumps were seen along and to the west of this road, driving N towards Anton Drive. However, since direction of fall could not be determined, they are not recorded on the damage illustrations. Nonetheless, from the number of large trees remaining along the storm's path towards Daniels Run Park, it appeared to have lifted partially until reaching Mode Street. At this point, it caused extensive damage to a small pocket of large, established trees along a small, steeply inclined drainage course. Shortly after this, it crossed Ashby Road, blowing away an attached, eastward opening, garage and a backyard porch to one residence. It then dropped an additional 55 feet (Figure 17) into a heavily wooded small residential park (with beech, poplar and oak) and caused extensive damage. Once again, extreme tree fall angles were noted, with at least 7O-80 major trees either snapped off 20-30 feet up or uprooted (some root systems up to 25 feet in diameter). At least a hundred smaller trees suffered similarly. At the approximate center of maximum destruction and almost the lowest point in the park, less than 10% of the trees remained standing. 0f these, most had slender trunks and were inclined at angles greater than 45 deg to the ground, to the NW, two months after the storm. There was some slight indication of anti-cyclonic rotation on the eastern edge of the damage area, although this tree fall was selective -with most adjacent trees remaining untouched. Also, note the interesting, isolated pocket of destruction in the NW corner of the Park (Figure 4: 48/16).

Discovering this area for the first time, five weeks after the storm, was -indeed- a sublime experience. Approaching it from the untouched and forested eastern end of the Park, a clearing slowly began to appear. First, a few fallen trees are seen, widely scattered. Then, suddenly, a clearing in the middle of the woods, almost total desolation on all sides and scores of trees covering the ground in a grotesque tangle. Already, half have been cleared away. The scene on the day after can only be guessed. The evening of the storm itself must have been frightening in the extreme, with the scrape and groan of hundreds of trees bending with the wind; and then the staccato snapping and bursting of dozens of trunks as a sudden wind surge hurled them down.

The storm continued moving to the NW, but with diminishing tree fall and little or no structural damage.

----Survey of Storm Characteristics

Four sound reports were reported. One resident (Figure 4: 74/26) reported the sound was like a 16" artillery shell flying through the air.

However, occupants of the two adjacent residences on the N and E of the one receiving the greatest structural damage (Figure 4: 84/34) reported no roaring sound. Rapid pressure fall was noted by one resident on the N side of the park (Figure 4: 32/34).

----Examination of Damage Patterns

Almost no discernible tree damage could be seen on the high ground between Lee Highway and the last apparent touchdown of tornadic winds in a heavily wooded drainage course on the north edge of the Country Club Hills residential area (Figure 5: 62/43). The damage was similar to that in Daniels Run Park, except that tree fall angles were not as ex- treme. Trees to the S of this area were on higher ground and appeared to be just as tall as the tallest trees in the drainage course, where maximum damage was done. No further damage was reported to the NW. No additional data was sought for surveying storm characteristics in this area.

----Summary Examination of Damage Patterns

From the nature of the damage, it is possible to identify five points of maximum damaging winds or possible tornadic touchdowns (small circled Xs in Figure 18). If the outer boundaries of known and significant damage are also connected by dashed lines (Figures 3-5), some interesting conclusions may be drawn:

l. Arcing damage patterns suggest that more than one tornado occurred (Figure 13) along this damage track, rotating about a "tornado cyclone" (also known as a mesocyclone) and touching down briefly along the NE quad- rant of the low. Such cyclones were discussed by Agee, Snow and Clare in a May, 1976 article in Monthly Weather Review.

2. The damage path veered to the right, indicating either a storm cell turning right or a weakening cyclonic circulation -expanding in diameter as it moved NW.

3. The five touchdown points of maximum damaging winds describe a steadily lengthening pattern. Using the distance between the first two points as a base reference of 100%, the distance between each successive point was 138%, 200% and 312%.

Figure 18 graphically portrays these and shows the plot line (straight lines between circled Xs) on which the terrain elevation analysis in Figure 17 is based.

Figure 18

It should be noted that outline of the damage area from visible damage was not always clear, as is evident by the undeveloped area NW of the high school football field or the path from the Lyndhurst Apartments to Daniels Run Park. In such areas,with little or no built up or wooded areas, I was limited to connecting damage points that were available -possibly omitting some areas where potentially damaging winds occurred. A case in point is the eastern damage boundary NW of the football field. Considerable weight was given to the separation of the cyclone fence along the west face of the field. It was completely pulled out and laid lengthwise, midway across the parking lot. The northern end of the damaged fence enscribes this side of the damage pattern. Another example of a judgement call was in the row house development immediately E of the Lyndhurst Apartments. The only significant damage here was in separated aluminum siding. Some complete wall sections were blown away. I cut the difference and only extended this end of the damage pattern halfway into that complex, since I didn't judge aluminum siding damage as comparable to fundamental structural failure. Other than these problem areas, there were sufficient trees to fairly accurately define the rest of the area. Overall, I believe this damage pattern to be fairly accurate, and that concentric patterns (whether more or less extreme than shown here) actually did occur. Assuming that each tornado retained its organization as it rotated around the mesocyclone, then two or more separate tornadoes were responsible for the damage in this area -with some completing each rotation to cause repeated damage.

Arcing damage patterns are highlighted by black arrows, which locate the approximate center of each damage area. They do not necessarily indicate the center track of the tornado(es), which was probably closer to the western edge of each curved area. Suction vortices are shown by smail., open-lined arrows.

The apparent turning of the damage track can be plotted exactly by joining the three Figures (#3-5) corner to corner, where they overlap, and sighting down the track from end to end (as in Figure 18).

Selection of touchdown points/maximum wind sites is somewhat arbitrary. These are shown by small circled Xs (Figures 3-5). The first is the Cervi residence. The second was selected at the west end of the football field, about halfway between the destroyed ticket booth and the Kayo home. This location does correspond well with the location of the separated fence (N-5 line immediately west of it). The third location is in Williamsburg Square, 'between the two most damaged structures. The fourth is about midway between the tree damage boundaries along Mode Street and the north end of the Park, and it is biased slightly toward the resi- dence that lost the attached garage and porch (Figure 4: 83/34). The fifth location is simply the center of the tree damage in the Country Club Hills subdivision (Figure 5).

Before leaving these damage pattern conclusions, the possibility should be noted that heavily timbered park or drainage areas were especially susceptible to severe wind damage, and that this may have been over- weighed in my analysis. I tend to think not, based partly on my analysis in Figure 17 and partly on several drive-through inspections of the area. The substantial drop in elevation for the last two tree damage sites suggests a descent of tornadic winds in these areas. Moreover, the survival of numerous residential shade trees (many old and established) along the high ground between these points does not indicate a constant altitude, damaging rotation.

As mentioned at the outset of this study, the principal motivation to do it was an initial report of a "golden glow" by one of the Cervi family, right after the tornado. Additional reports were also received. A resident NE of Cervi saw a "daylight (orange)" color. Other reports in this area ranged from "silver gray" to "very pink." One of the Mayo sons reported seeing a "purplish haze," and the resident four houses N (Figure 2: 67/27) also reported a golden glow. I questioned this last witness carefully, since she lived closest to a large shopping center to the NE, where electricity remained on until about 7:30 PM EDT. She was definite that the glow she saw was from the sky overhead and not from the shopping center. Having lived several years in that area, she had seen the reflection of those lights many times on overcast days.

More significantly, these observations are confirmed by photocopies Of the radar imagery, at this time, from the Patuxent River station in Maryland. Despite somewhat poor quality of detail in these pictures, clear evidence of holes, or absence of reflectivity, in the tornado area are indicated. Two holes are revealed 10 minutes before the tornado. At the time of the tornado, one hole has filled, but the second remains barely discernible. This is also the one closest to and just S of the damage area. Figure 19 reproduces the two photos as faithfully as possible, with the two levels of reflectivity shown by either dotted or solid lines. It should be noted that these holes are surrounded by at least 10 miles of heavy cloud mass (more than 3/4 of the respondees confirmed that light to moderate rain continued after the tornado). The meteorologist respondee reported heavy rain returning 20 minutes after the tornado. Consequently, these openings were not evidence of a general clearing behind a squall line but were isolated occurrences. As such, I continue to believe them to be significant for radar detection of tornadoes in hurricane/tropical storm environments. Further research appears warranted in this area.

Figure 20 reproduces the radar log for this day. Note that storm tops at about this time (7:35 pm EDT or 2335Z) were only reported at 30,000

Figure 19

Surface weather maps (Figures 21 and 22) were interesting for the similarities they indicated between this tropical storm and midwestern severe storms. Applying a forecasting technique that I developed for midwestern storms and using only surface weather data, the basic para- meters of wind convergence and of low level moisture, temperature and maximum wind axes were found to be equally reliable for forecasting tornadoes in this storm situation. Beginning with map data from 10:00 to 12:00 AM EDT, I was able to consistently forecast a tornado watch area (after the fact), centered within 26 miles and westerly of the Fairfax tornado. The National Weather Service issued a tornado watch at 6:55 PM EDT for an area 70 miles either side of a line 70 miles WNW of Harrisburg, Pennsylvania to 20 miles E of Norfolk, Virginia -between 8:00 PM EDT (060000 Z) and 2:00 AM EDT (060600 Z). The forecast called for tornadoes and a few severe thunderstorms with hail surface and aloft to 1" diameter, strong gusty surface winds to 55 knots, and storm tops to 38,000 feet. The center of this area was 16 miles easterly from the Fairfax tornado. The point here is not a pejorative comparison between different forecasting systems but rather to point out the highly reliable predictability of tornado forecasting for this storm. In fact, several residents were critical of the National Weather service for their failure to issue a public watch, especially when they issued one for aviation interests. Mr. Mayo said he would have had his family in their basement much earlier had he known of a tornado possibility (rather than rushing downstairs with seconds to spare).

I understand the argument put forward by the NWS that they wanted to emphasize the principal hazards to the greatest number of people, which were flooding and general, high winds. It was thought that addition of a tornado watch would have been confusing and would have detracted from the likelier threats of straight winds and flooding. However, I tend to share the view of residents who survived this tornado. The NWS should reconsider the desirability of issuing public tornado watches in such instances and in areas such as this, where tornadoes are comparatively rare. We experience many more flash flood situations here than tornadoes, and people are more familiar with flood precautions. I suspect that NWS does not sufficiently consider variations between local climatic experiences of different communities, and that this should come more into play. If the Cervi family weren't as lucky and the demolished Williamsburg Square residence had been occupied by another family, this storm would easily be remembered now as one of the dead- liest to strike the metropolitan area.

Figure 20

In general, I believe it is riskier to withhold information from the public and decide what is in their best interest, rather than to keep them fully informed.

Figures 23 and 24 show the upper air soundings and stability indices on the morning and evening of the 5th, along with duplicate, typed re- productions of the data printouts for those wishing to do their own charts (copy of psuedo-adiabatic charts did not duplicate very well). Figure 25 is the surface weather observation log from the Washington, D.C. National Airport.

The Fairfax storm of September 5 presented a unique opportunity to study the effects of one or more tropical storm tornadoes, comparatively rare this far north, over a residential area with numerous witnesses, and in daylight. It was a small storm, doing F-2 to F-3 damage of a very localized nature. Damage patterns revealed a rotating mesocyclone with two or more satellite tornadoes moving about the overall circula- tion core and briefly intensifying along the NE quadrant. This apparently indicated a symbiotic relationship between each tornado and the environ- mental winds as its forward movement became coincident with the prevail- ing SE winds of 30-35 MPH. In fact, it is probably more accurate to speak of tornadic winds at such times of maximum ground damage than of tornadoes, which -generally- were not visible. Suction vortices were present and were probably responsible for all of the most extreme damage. However, the mesocyclone was not sufficiently organized to produce the highly visible condensation column of classic midwest tor- nadoes. This storm probably looked more like the Xenia, Ohio tornado in its early stages, based on published pictures of that storm; i.e. a lowering in the storm base, with ragged cloud fragments forming and blowing apart, and suction vortices rapidly spiralling up from the ground (for a few seconds) and dissipating. Sky glow reports, confirmed by radar, indicate that an opening in the clouds -and a hole on the radar scope- did occur just south of the tornado damage, 10 minutes before it occurred, and moved over the area immediately after the tor- nado. This may be significant for future radar detection of tornadoes in tropcial storm/hurricane situations. An interesting report of an apparent outflow gust-front (a cloud to ground "grey wall" ) was received. Widely varying tornado sound experiences were reported, and posed many questions about the reliability of this phenomenon as a warning. The NWS forecast 1" hail and cloud tops to 38,000 feet. However, no hail was reported and tops did not exceed 30,000 feet at that time. As the storm tracked to the NW, the distance between touchdown of damaging winds extended consistently, and the track veered slightly to the right. Most tree fall indicated damaging winds from the SE and NE, with more extreme angles presumably due to local suction vortices. There was indication that damaging tree fall can occur ahead of the tornadic circulation, and that this may pose a problem for future aerial surveys of ground damage from such storms (i.e. what is tornadic wind damage as opposed to other kinds of wind damage?). Finally, the surface para- meters indicated a highly predictable storm, with many similarities to a midwest tornadic storm.

I trust that this study will be a small contribution to tornadic research and will be of interest to meteorologists and friends of Storm Track around the country. Let me know what you think of it, where you agree or disagree, etc. Along with normal subscribers, I have mailed courtesy copies of this study to respondees to the questionnaire and to others who assisted in its development. I am greatly indebted to those listed on the following page for their assistance, including residents for their responses, the Fairfax County Division of Mapping for its invaluable aerial photos and contour maps, and meteorologists at the World Weather Building for observational data. However, along with these credits, I want to emphasize that any mischief taken with the data is strictly my own. With the exception of Les Lemon's suggestions regarding relation- ship between the gray wall and an outflow gust-front, all other conclusions and speculations are my own. Once again, my sincere thanks to:

Tom Merritt, Faiirfax County Division of Mapping Charles Archambault, Deputy MIC, World Weather Building

Frank Cervi Anne Griffith W. W. Hacker Roland R. Kessler Paul Mayo Robert E. Phipps, Principal, W. T. Woodson High School Resident, 9509 Spode Court 9510 Baccarat Drive 9512 Baccarat Drive 9514 Baccarat Drive 4104 Whitacre Road 4112 Whitacre Road 4024 Whitacre Road Nancy Hoadley (wife) -- for editing/proof-reading

Along with other subscribers, copies of this newsletter were mailed to: Ronald Holle, National Hurricane Center, Miami Les Lemon, National Severe Storms Forecast Center, Kansas City Richard E. Peterson, Atmospheric Science group, Texas Tech University Library, National Severe Storms Laboratory, Norman, Oklahoma

Figure 21

Figure 22

Figure 23

Figure 24

Figure 25