May 12, 2000 Tornadic Storm


In a situation highly reminiscent of the Jarrell tornado of three years ago (5-27-97), a supercell storm formed along a stalled prefrontal trough at mid-afternoon in an environment of extremely high CAPE values but only weak to moderate shear. The storm moved south- southwest along the stalled boundary and produced a large tornado on the western shore of Lake Whitney in Bosque County. One person died and one was seriously injured when the storm destroyed a mobile home. Many homes were damaged or destroyed. The National Weather Service at Fort Worth surveyed the damage and rated it F-3 on the Fujita scale.

The Storm Prediction Center outlooks for Friday, May 12th were for a slight risk of severe storms in Central Texas but the primary threat was expected to be large hail and damaging winds. The tornado threat was characterized as being very low. A cold front preceded by a prefrontal through was moving southward from northwest Texas. At noon CDT, the front along an Ardmore, OK - Mineral Wells, TX - Midland, TX line. Ahead of the front and trough, a large area of relatively clear skies was permitting near-maximum surface insolation. By early afternoon, I was concerned about the surface-based instabilities for our area. The cold front and prefrontal trough were slowing in southward movement as strong solar insolation took its toll on the advancing cooler air. GOES-8 satellite estimates of CAPE were in excess of 4000 j/kg and rising, and convective inhibition (the result of warm air in the mid levels) was slowly weakening.
I wasn't certain that storms would develop, but the high CAPE level and stalling boundary (clearly visible in high-resolution satellite imagery) were suggesting that this might be an analog to the Jarrell day of three years ago.

At 2:30pm, I updated my forecast/outlook page with this information:

For this afternoon and evening:

The lower atmosphere has become extremely unstable over Central Texas this afternoon as the mid-May sun has warmed the earth and the moist layer of air trapped near it. The atmosphere remains capped by a layer of warm air (between 5,000 and 10,000 feet above ground) which has inhibited storm formation for several days. Caps can sometimes be broken by daytime heating, but one as strong as the present one will require a mechanical source of lift. A southward moving cold front will provide such a mechanism later this afternoon and tonight. The winds at the surface and aloft are forecast to weaken with time today, so the storms which do form will do so without the benefit of a wind profile which would aid organization into supercells. Therefore, the most likely storm type later this afternoon and tonight is multicell storms, and although these may become briefly severe (hail and strong straightline winds), the tornado threat (at this time) appears to be very low.

The exception to this might be storms which deviate sharply southward in motion. Southward motion would maximize inflow at lowest levels into a storm and the extreme instability (CAPE exceeds 4000 j/kg in our area and may be approaching 5,000 j/kg) might produce a storm situation where the weak winds at the surface and aloft were compensated by the extreme levels of instability.

Radar and Satellite View

Shortly after 4pm CDT on Friday, May 12, 2000, the first of several severe thunderstorm cells developed in Hill County. The initial movement was eastward. As the subsequent storms developed, eastward movement halted, and the storms began drifting and back- building to the south and southwest in a manner very similar to the storms which crossed Central Texas three years ago, on May 27, 1997. One of the storms developed a pendant (and subsequently a hook) echo as it approached Lake Whitney from the northeast. On the southwest shore of Lake Whitney (on the Brazos River, the boundary between Hill and Bosque counties) the storms produced a tornado which destroyed and damaged a number of homes and caused one fatality. Following a survey of the damage on 5-13-2000, the Nat'l. Weather Service in Fort Worth has rated the damage as F3 on the Fujita scale.

High resolution visible image from GOES-8 at 5pm CDT 5-12-2000.

In this satellite image, the storms in Hill County can be seen forming along a line of cumulus clouds which marks the leading edge of a cold front and prefrontal trough which had been moving southeastward during the day.

NWS NEXRAD reflectivity (0.5 deg elevation) at 5:32pm CDT on 5-12-2000.

This radar image was taken within a few minutes of the tornado occurrence. The storm which produced the tornado is along the Hill- Bosque county line north-northwest of Waco (ACT). Note the pendant echo on the southwest flank of the storm. Other storms can be seen developing to the southwest, in the Evant and Jonesboro areas, and storms were continuing to the northeast in Ellis and southern Dallas counties.

NWS NEXRAD velocity (0.5 deg elevation) at 5:32pm CDT on 5-12-2000.

This is the image from NEXRAD in the doppler mode. The radar detects not only how much energy was returned from the target (the particles in the storm) but also whether the frequency of the radar beam has changed slightly to a higher or lower frequency. A shift in frequency indicates that particles from which the beam bounced back were moving either toward or away from the radar site. A computer processes this information and determines the speed with which the particles were moving relative to the radar. In this image, green indicates motion toward the radar and yellow, orange and brown indicate motion away from the radar. The radar which produced this particular image is in southern Tarrant County, south of Fort Worth.

Look at the same location in this image as the pendant echo in the preceding image and you'll see an area where the green and yellow-orange echoes are directly against each other. From that indication, we are able to infer a strong inbound and outbound flow that is "gate-to-gate", and research has shown that such a signature indicates the presence of a mesocyclone and sometimes, a tornado circulation. By the way, we are forced to infer the presence of a circulation because a doppler radar is "blind" to particle motion perpendicular to the radar beam, so we can't "see" the remainder of the circulation directly with radar. Nevertheless, research using dual-doppler radars has shown this to be a reliable indicator of storm-scale rotation.

We aren't actually seeing the tornado here. Although some people claim that the radar can detect the actual tornado, it is unlikely except when a storm is very close to the radar site. Because of the curvature of the earth's surface, the radar beam which leaves the radar at 0.5 degrees elevation above the horizon will be about a mile above the earth at 75 miles from the radar site, and almost two miles above the surface at 125 miles. So, the radar is limited by its "horizon" which is a result of the very subtle but constant curvature of the earth's surface. In addition to the "horizon" limitation, the radar beam gradually becomes broader as its travels away from the radar, and motion entirely within the beam may not be detected. Thus, features which are relatively small (like tornadoes) are not normally detected but may be inferred from the data returned by the radar.

Weather Data and Analysis

Here's my 4pm CDT analysis of the surface weather chart:

At this point of Friday afternoon, locating the cold front was not easy because the winds were turning to the northwest behind the prefrontal trough but ahead of the cold front. However, close examination revealed that the cold front was between Abilene and San Angelo (temperature of 88F at Abilene versus 100F at San Angelo). I analyzed the front to be south of Mineral Wells but north of Fort Worth (note that the temperature and dewpoint at Mineral Wells was significantly lower than either at Fort Worth. Between the cold front and the trough, the airmass was much drier so that prefrontal trough was acting like a dryline feature by this time, at least in the area south of Fort Worth.

Another important feature was the apparent mesoscale low near Junction, where the lowest pressure on the map was found (1002.6mb). This feature was partly in response to intense insolation (note Junction was at 102F) but had been apparent at least 3 hours earlier, so it was not entirely a temperature-related feature. The Waco area was in that critical zone northeast of the mesoscale low, and the impact of the mesolow appears to have been the backing of the surface flow in the Waco to Austin area over the 3-hour period between 1pm and 4pm CDT. In fact, the winds continued backing through 6pm at Waco. Because of the backing of the surface flow, the wind profile in the area south of Fort Worth to Waco and Austin remained somewhat favorable for rotating storm updrafts, as can be seen in the VAD Wind Profile from the NEXRAD site at Granger (near Fort Hood):

VAD Wind Profile from KGRK (Granger-Central Texas) at 2:57pm CDT on 5-12-2000

Because the wind speeds are relatively weak up through 10,000 ft, this would certainly not qualify as a classic tornadic profile. However, there was good veering of the wind with height above ground and for a storm deviating well south of the mean wind vector, the storm- relative helicity would be pronounced. Coupled with the other critical piece of information which could be gleaned from a comparison of the 1pm (not shown) and 4pm analyses (that moisture was "pooling" in the Waco area and that CAPE values were reaching extreme levels), one could make a case for some tornadic threat near Waco if storms could develop.

The pooling of moisture at Waco was obvious when one considers that despite strong insolation and winds sufficient for turbulent mixing, the dewpoint at Waco actually remained unchanged (75F) while the temperature climbed 3 degrees (F) between 1pm and 4pm CDT. In fact, if you simply modify the 7am CDT sounding from Fort Worth for the temperature and dewpoint found at Waco at 4pm CDT (a crude way of estimating conditions because it disregards changes in temperature and humidity at other levels), you get the following indicators of potential for severe storms: CAPE = 6100 j/kg and lifted index (LI) = -12.5 deg. However, this simple modification also revealed that the atmosphere would still be strongly capped (4.0 deg) and that convective inhibition would be 157 j/kg.

The 4pm CDT analyses of CAPE and moisture convergence from the CAPS program at the Univ. of Oklahoma also indicated that instability was approaching extreme levels in the Central Texas area. Here are zoomed sections of two maps, one showing CAPE and the other showing moisture convergence:

CAPE at 4pm CDT on 5-12-2000
The area in bright orange shows CAPE exceeding 5000 j/kg.

Surface Moisture Convergence at 4pm CDT on 5-12-2000

Three areas of MOCON are identified. The bright yellow area coincides with severe storm development in the Collin County area which occurred more or less contemporaneously with the Hill County evolution. The Collin County storm produced one confirmed tornado. The first orange area to the southwest coincidcs with the storms which evolved in Ellis County and which became severe but did not produce any confirmed tornadoes. The second orange area precisely overlies the zone where the tornadic storm occurred less than two hours later (southwest Hill and southeast Bosque counties).

It isn't entirely clear why the Lake Witney storm moved toward the south-southwest while the other storms were moving mostly east-northeast. Perhaps in time we can explain this aspect of storm behavior and more accurately anticipate and predict when it will occur.

Update in progress ...

  Back to the main homepage!

This page was last updated on 5-26-2000.

If you have comments or suggestions, email me at

This page created with Netscape Navigator Gold