Doppler Radar Analysis of the Northfield, Texas, Tornado of 25 May 1994

Howard B. Bluestein School of Meteorology, University of Oklahoma, Norman, Oklahoma

Search for other papers by Howard B. Bluestein in
Current site
Google Scholar
PubMed
Close
,
Wesley P. Unruh Lawrence, Kansas

Search for other papers by Wesley P. Unruh in
Current site
Google Scholar
PubMed
Close
,
David C. Dowell School of Meteorology, University of Oklahoma, Norman, Oklahoma

Search for other papers by David C. Dowell in
Current site
Google Scholar
PubMed
Close
,
Todd A. Hutchinson School of Meteorology, University of Oklahoma, Norman, Oklahoma

Search for other papers by Todd A. Hutchinson in
Current site
Google Scholar
PubMed
Close
,
Todd M. Crawford School of Meteorology, University of Oklahoma, Norman, Oklahoma

Search for other papers by Todd M. Crawford in
Current site
Google Scholar
PubMed
Close
,
Andrew C. Wood School of Meteorology, University of Oklahoma, Norman, Oklahoma

Search for other papers by Andrew C. Wood in
Current site
Google Scholar
PubMed
Close
, and
Herbert Stein Garrettsville, Ohio

Search for other papers by Herbert Stein in
Current site
Google Scholar
PubMed
Close
Full access

Abstract

A large tornado was observed near Northfield, Texas, on 25 May 1994 during the Verification of the Origins of Rotation in Tornadoes Experiment (VORTEX). An analysis of the tornado and its parent storm is discussed. Doppler wind velocity spectra of the tornado and its parent circulation, which were computed from data collected by a low-power, portable, FM-CW (frequency-modulated continuous-wave), 3-cm-wavelength Doppler radar, are presented at increments in the range of 78 m. The FM-CW radar data from the tornado are the first ever collected of high enough quality to analyze. The CW spectra computed from data collected by the portable radar, a pseudo-dual-Doppler analysis of airborne Doppler radar data collected by a National Oceanic and Atmospheric Administration P-3 aircraft, photogrammetric analysis of a video of the tornado, and a ground-based damage survey are discussed in the context of the FM-CW spectra. This study is unique in that both ground-based and airborne Doppler radar systems probed the tornado and its environment. Wind speeds of 60 m s−1 were indicated in the tornado in a swath 300 m across, with some smaller areas of possible wind speeds up to 75 m s−1. Circumstantial evidence is presented that the tornado originated along an elliptically shaped cyclone/shear zone along the leading edge of a large hook echo in its parent supercell storm. The tornado’s parent vortex (mesocyclone) was approximately 2 km in diameter and contained tangential wind speeds of 45–50 m s−1.

 Sabbatical affiliation: National Center for Atmospheric Research, Boulder, Colorado.

Corresponding author address: Dr. Howard B. Bluestein, School of Meteorology, University of Oklahoma, 100 E. Boyd, Room 1310, Norman, OK 73019.

Email: hblue@ou.edu

Abstract

A large tornado was observed near Northfield, Texas, on 25 May 1994 during the Verification of the Origins of Rotation in Tornadoes Experiment (VORTEX). An analysis of the tornado and its parent storm is discussed. Doppler wind velocity spectra of the tornado and its parent circulation, which were computed from data collected by a low-power, portable, FM-CW (frequency-modulated continuous-wave), 3-cm-wavelength Doppler radar, are presented at increments in the range of 78 m. The FM-CW radar data from the tornado are the first ever collected of high enough quality to analyze. The CW spectra computed from data collected by the portable radar, a pseudo-dual-Doppler analysis of airborne Doppler radar data collected by a National Oceanic and Atmospheric Administration P-3 aircraft, photogrammetric analysis of a video of the tornado, and a ground-based damage survey are discussed in the context of the FM-CW spectra. This study is unique in that both ground-based and airborne Doppler radar systems probed the tornado and its environment. Wind speeds of 60 m s−1 were indicated in the tornado in a swath 300 m across, with some smaller areas of possible wind speeds up to 75 m s−1. Circumstantial evidence is presented that the tornado originated along an elliptically shaped cyclone/shear zone along the leading edge of a large hook echo in its parent supercell storm. The tornado’s parent vortex (mesocyclone) was approximately 2 km in diameter and contained tangential wind speeds of 45–50 m s−1.

 Sabbatical affiliation: National Center for Atmospheric Research, Boulder, Colorado.

Corresponding author address: Dr. Howard B. Bluestein, School of Meteorology, University of Oklahoma, 100 E. Boyd, Room 1310, Norman, OK 73019.

Email: hblue@ou.edu

1. Introduction

Real wind data are needed to verify theories for the structure and evolution of tornadoes. A review of efforts to make wind measurements and map the wind field in tornadoes is given in Bluestein and Golden (1993). Remote sensing by radar has proven to be a safer, easier, and more productive technique than attempts to make in situ measurements. Zrnić and Doviak (1975) showed how Doppler wind velocity spectra can be used to estimate the maximum horizontal wind speeds in tornadoes. Wind spectra are more meaningful than mean Doppler velocities when there is a large variation in wind speed in the volume probed by the radar (Brown et al. 1978; Wakimoto and Martner 1992). The shape of the Doppler wind spectrum can be used to infer some properties of the tornado wind field when the radar volume is approximately the same size as or slightly greater than that of the tornado vortex. A summary of Doppler wind spectra in tornadoes computed from a fixed-site pulsed Doppler radar at the National Severe Storms Laboratory (NSSL) is found in Bluestein and Golden (1993). Unfortunately, the NSSL was unable to obtain simultaneous visual documentation of the tornadoes, and since most of the tornadoes probed were relatively far from the radar site, the center of the radar volume was at or above cloud base.

Bluestein and Unruh (1989) and Bluestein et al. (1993) showed how a portable, low-power, CW (continuous wave) Doppler radar could be used during severe storm intercept field programs to increase the number of tornado datasets, provide visual documentation, and obtain measurements below cloud base. It was demonstrated, using estimates of nearby soundings, that the maximum horizontal wind speeds in tornadoes exceed the thermodynamic speed limit (Snow and Pauley 1984; Fiedler and Rotunno 1986). It was also shown that horizontal wind speeds in some tornadoes can be as high as 120 m s−1 or greater (i.e., of F-5 intensity) (Fujita 1981), as had been suggested by damage estimates and photogrammetric analyses of debris movies, and that the wind speeds in tornadoes in the shrinking (or “rope”) stage (Golden and Purcell 1978) can indeed still be very strong even though the tornado is nearing its demise.

Although efforts to obtain Doppler wind velocity spectra within range bins using the portable radar operated in the FM (frequency modulated)-CW mode (Strauch 1976) have been successful in obtaining wind data in wall clouds (Bluestein and Unruh 1993), they have not been successful in obtaining wind data in tornadoes until recently. Although FM-CW Doppler data were collected in several tornadoes in 1991, they were too noisy to be analyzed successfully. FM-CW data of high enough quality to analyze for the first time ever were collected in a tornado near Northfield, Texas, on 25 May 1994, during the Verification of the Origins of Rotation in Tornadoes Experiment (VORTEX) (Rasmussen et al. 1994). In addition, simultaneous photographs, high-quality videos showing cloud tag motion, and CW Doppler data were obtained. Furthermore, airborne Doppler data were collected from near cloud-base level aboard a National Oceanic and Atmospheric Administration (NOAA) P-3 aircraft just before the tornado touched down, and a detailed ground-based damage survey was completed after the tornado had struck. Prior to 25 May 1994, only a few tornadic storms had been probed by the airborne radar, and they were probed from well above cloud base just after the tornado had dissipated (Hane et al. 1993). Although only a “snapshot” view of the tornado and its parent storm is afforded, this study is unique in that several different observing systems were available to probe the tornado and its surroundings. The primary purposes of this paper are to discuss the FM-CW data and to integrate analyses from all the data sources and obtain a more detailed description of the tornado than possible from the portable FM-CW/CW radar alone.

2. Doppler radar data processing and analysis procedures

a. FM-CW Doppler wind data

The principles of operation of an FM-CW Doppler radar are described in Doviak and Zrnić (1984), and the principles of data processing are detailed in Strauch (1976). The operation of the 3-cm-wavelength FM-CW Doppler radar system we used (Fig. 1 and Table 1) and the techniques we used to analyze the data are described in detail in Bluestein and Unruh (1993). In brief, the frequency of the continuously transmitted signal is periodically increased (“chirped”) linearly and mixed with the signal from the backscattered radiation (Fig. 2). The rate at which the signal is chirped (or for the case of linear sweeps in frequency, “ramped”) is called the sweep repetition frequency, which is analogous to the pulse repetition frequency of pulsed radars. Range information is obtained from the frequency of the mixed signal, and velocity information is calculated from the rate of change of the phase shift of the mixed signal from sweep to sweep. The advantage of FM-CW operation is that relatively high range resolution may be attained with high sensitivity at very low continuously transmitted microwave power. Like the pulsed Doppler radar, the FM-CW Doppler radar has a maximum unambiguous velocity and a maximum unambiguous range. However, the velocity folding and range folding contamination produced by FM-CW Doppler radars have properties different from those associated with pulsed Doppler radars. Range folding contaminates velocity spectra, and velocity folding contaminates range information. In pulsed Doppler radar systems, on the other hand, range folding contaminates range information and velocity folding contaminates velocity spectra.

The FM-CW signals are recorded in analog form on the video track of a VHS videotape; simultaneous images of the tornado are therefore not available on videotape. However, our data collection procedure required the mode of the radar to be switched frequently between FM-CW mode and CW mode. In the CW mode the videotape records a bore-sighted view of the direction in which the portable radar is pointed, while the two separated CW audio Doppler signals are recorded on the stereo channels. Useful visual documentation is thus being recorded during the brief periods when CW data are being recorded.

The video (FM-CW) data are analyzed later on back home, not in real time out in the field. One-half of an interleaved frame from the videotape is captured and digitized as a 32 256-point time series (252 horizontal lines times 128 points per line). From this time series a segment of at least 16 384 points (usually most of or the entire 32 256-point series), which is judged subjectively to have the least contamination, is chosen for analysis. Quasi-stationary ground clutter (ground clutter is never really stationary: e.g., tree leaves may flutter in the wind) is eliminated by obtaining the average of all 128-point ramp signals and subtracting this average from each of the individual 128-point ramp segments. The subtraction operation removes signals that are stationary during the entire 8 ms of the time series record. The Doppler wind spectra are estimated by dividing each time series into three overlapping segments of 16384 points each in order to obtain optimum noise reduction (Welch 1967). A Blackman window is applied to each of these uniformly overlapping time series segments before a fast Fourier transform is computed. In the transform process the data are range normalized by a factor equal to the reciprocal of the square of the range. The spectral densities are then averaged together to produce a spectrum consisting of 8192 points. This spectral series represents 64 range bins, each having 128 Doppler velocity points. Data in the first and last halves of the first and last range bins, respectively, are contaminated by folding and are not considered.

The Doppler wind velocity spectra as a function of range that are computed from the analysis of data from an FM-CW Doppler radar are similar to the Doppler wind spectra as a function of range that could be computed from the analysis of data from a pulsed Doppler radar. However, contamination of the spectra must be interpreted in the special context of FM-CW Doppler radar operation and analysis.

b. CW Doppler wind data

In the CW mode the signals from the portable 3-cm-wavelength Doppler radar do not contain any range information. However, the maximum unambiguous velocity is limited only by the frequency response of the recording medium, which in our case is ±292 m s−1, well above what we expect the maximum horizontal wind speeds are in tornadoes. The CW Doppler wind spectra represent the weighted sum of spectra at all ranges, with most weight given to the nearby stronger scatterers. We try to get as close to tornadoes as possible in order to detect the weaker backscattering from the area of strongest winds, which may be in regions of low reflectivity, but not too close to ensure our safety. Since the data analyzed from the CW mode of the portable radar are more sensitive than those analyzed from the FM-CW mode, and it is thought that the highest wind speeds in tornadoes may occur in the vicinity of the tiniest scatterers (Bluestein et al. 1993), the CW spectra should yield better estimates of the maximum wind speeds in tornadoes than the FM-CW spectra.

Details about the operation of the radar in the CW mode and the processing of its data are found in Bluestein and Unruh (1989) and Bluestein et al. (1993). The CW signals, which are the mixture of the transmitted signal and the signal from backscattering radiation, are stored on the audio channels of the videotape, while simultaneous images of the tornado are stored on the video channel. Spectra were computed after time series of data had been digitized and captured. The spectra were subjected to a Hanning window and were averaged in time to reduce the signal-to-noise ratio. A high-pass filter was applied to reduce quasi-stationary ground-clutter contamination.

c. Airborne Doppler radar data

One of NOAA’s P-3 aircraft flew by the Northfield, Texas, storm near cloud-base level approximately 20–25 min before the FM-CW/CW radar collected data. It was not possible to collect simultaneous airborne radar data because new cell development to the south (to be discussed later) made it impossible for the aircraft to fly by the storm’s southern edge near the tornado while it was on the ground. A detailed description of the P-3 3-cm-wavelength tail radar system (Table 2) is given in Jorgensen et al. (1983).

During VORTEX-94 airborne radar data were collected using the fore–aft scanning technique (FAST) (Jorgensen et al. 1995). The radar beam is alternately scanned fore and aft approximately 20° from a plane normal to the flight track (Fig. 3). During each scan the beam rotates about the flight track axis so that the sweep traces out a cone moving along with the aircraft. Using the FAST, pseudo-dual-Doppler data (Jorgensen et al. 1983) are collected from only one aircraft, which is flying a straight flight leg. The P-3 aircraft flew approximately 20 km from the storm. Since the time difference between two intersecting fore and aft beams is 1 min for each 10 km in range from the flight track, there is a time difference of approximately 2 min between beams. Pseudo-dual-Doppler analysis is therefore valid if the storm in its reference frame is in a steady state for approximately 2 min. Hildebrand and Mueller (1985), Ray and Jorgensen (1988), and Ray and Stephenson (1990) have analyzed convective storms using data collected with the radar beam scanning normal to the flight leg and with the aircraft executing perpendicular legs (i.e., an L-shaped pattern); in this mode of operation the time difference between intersecting beams is even longer. Jorgensen and Smull (1993) were the first to use the FAST when collecting and analyzing airborne Doppler data. More recently, Blanchard (1992), Watson et al. (1993), and Dowell et al. (1997) have analyzed supercell storms using data collected using the FAST.

The components of the aircraft’s ground-relative velocity in the fore and aft directions were removed from the Doppler wind data before pseudo-dual-Doppler analysis. The position and orientation of the aircraft were recorded by the aircraft’s inertial navigation equipment, which was corrected for drift. The Doppler velocity data were contaminated by turbulent, short-term changes in aircraft orientation and by errors (biases) in the antenna pointing angle. Efforts were made to correct these errors by ensuring that the Doppler velocities of ground return corrected for aircraft motion were as close to zero as possible.

The Doppler velocity and reflectivity data were edited to correct velocity folding and other errors. Unfolding the velocity data was particularly time consuming and difficult because the maximum unambiguous velocity was only 12.9 m s−1, while actual wind velocities were as high as three times that at high levels, and gate-to-gate shear near regions of intense divergence, convergence, and vorticity was often comparable to the Nyquist interval.

Wind and reflectivity data were interpolated, using a Cressman weighting function (Cressman 1959) having a radius of influence of 750 m, to a Cartesian grid having both vertical and horizontal spacing of 500 m, which are appropriate for the resolution and spacing of the data at a range of 20–30 km (less than 1 km and 700 m along the flight track between consecutive scans, respectively), which is the range of the far end of the region of interest in the storm. Vertical air motion was calculated after having removed the vertical component of precipitation motion using the empirical relationship between radar reflectivity factor and raindrop terminal fall velocity in Foote and duToit (1969). Since the distribution of hail within the storm was unknown, we were unable to account for the terminal velocity of hailstones. However, for low elevation angles (< 45°) the effect of errors in terminal velocity on the horizontal wind field are insignificant compared to other errors in the Doppler analysis (Doviak et al. 1976).

The three-dimensional wind field was calculated iteratively using an initial first guess of zero vertical velocity and a lower (ground level) boundary condition of zero vertical velocity. Vertical velocity was computed kinematically using upward integration of the anelastic continuity equation. As a result of possible errors in our calculations of vertical velocity (Doviak et al. 1976; Hildebrand and Mueller 1989; Ray and Stephenson 1990), our subsequent discussions will include qualitative rather than quantitative estimates of vertical motion.

At each grid point the greater reflectivity value of the two different beams was used because it is likely that greater attenuation of the radar beam at the other viewing angle had reduced the reflectivity. If the radar had been properly calibrated, the greater reflectivity would probably be more accurate, even though this itself could be an underestimate owing to attenuation.

3. Analyses

Evidence from the conventional synoptic-scale network suggests that the Northfield storm formed in a moist, potentially unstable air mass, in an environment of vertical shear and convective available potential energy (not shown) high enough for supercell development (Weisman and Klemp 1982). Since VORTEX mobile-mesonet observations (Rasmussen et al. 1994) of the storm were not available owing to the lack of roads and soundings were not available in the undisturbed inflow region of the storm, we will focus mainly on the radar observations of the storm.

a. WSR-88D observations

The Northfield, Texas, tornado on 25 May 1994 occurred in a supercell that formed ahead of the southern edge of severe convective storms that were moving eastward through the Texas panhandle (Fig. 4a). A cell northeast of the Lubbock, Texas, WSR-88D Doppler radar site at a range of approximately 100–110 km at 2130 UTC (all times given in UTC) evolved into the parent supercell (hereafter referred to as the Northfield storm); the remains of earlier intense, but nontornadic activity is seen to its northwest. The area of the Northfield storm at low levels grew (Fig. 4b) until about 2200, when its low elevation angle radar echo developed a concave shape along its southern flank (Fig. 4c). The development of such a concave shape is considered a sign of intensification (Lemon 1977). An east–west-oriented line of weak cells just to the south of the Northfield storm evident at 2130 (Fig. 4a) had disappeared by 2145 (Fig. 4b), during the time period when the Northfield storm was getting larger. At 2215, approximately 15–20 min before the appearance of the tornado, a large, well-defined hook was apparent on the storm’s southwest side (Fig. 4d). The hook, whose area was unusually large, persisted through 2231 (Fig. 4e) when the tornado was visible, but had disappeared by 2246 when the tornado dissipated (Fig. 4f). The intercept crew drove through the hook, from west to east, approximately 13 min before the tornado appeared just to the north, and noted that a curtain of precipitation extended off to the south; sheets of precipitation moved southward, west of the leading edge of the north–south-oriented cloud base, while sheets of precipitation moved northward near the leading edge (rotating rain curtains had also been noted earlier when the crew was driving through rain and hail). By 2301 a northeast–southwest-oriented line segment of convective cells that had been off to the south (Fig. 4c) and had moved northeastward (Figs. 4d–f) had caught up with the Northfield storm (Fig. 4g), which subsequently weakened. Although a new intense storm replaced the Northfield storm to its west (Fig. 4g), it failed to develop a tornado.

At 2221, just prior to the touchdown of the tornado, the hook echo at 0.5° elevation angle (Fig. 5a) was associated with cyclonic shear (Fig. 5b), as evidenced by the transition from brown (indicating flow away from the radar at about 15 m s−1) along the southern edge of the hook to green (indicating flow toward the radar at about 10 m s−1) along the northern portion of the hook. However, at 1.5° elevation angle (Fig. 5c) above the hook echo at 0.5° elevation angle there was a bounded weak-echo region (BWER), which was associated with stronger cyclonic shear (Fig. 5d). The Doppler velocities changed from green along the northwestern edge of the BWER to purple (indicating velocity folding and flow away from the radar in excess of 25 m s−1) along the southwestern edge of the BWER. Although the storm was much too far away from the radar to resolve the tornado’s parent circulation, the intensity of cyclonic shear on the storm scale increased with height at low levels. On the other hand, it may not have if the horizontal scale of the circulation at 0.5° elevation angle was smaller than at 1.5° elevation angle, so that the shear at the lower elevation angle had been underestimated due to beam smoothing. The presence of the BWER, hook, and cyclonic shear are evidence of a strong updraft.

b. FM-CW Doppler wind velocity spectra

The FM-CW/CW Doppler radar probed the Northfield tornado beginning at approximately 2242. A tornado touched down briefly around 2230–2233 and appeared again shortly thereafter at about 2239. Just above the ground the tornado looked like a cone whose tip was touching the ground, while the portion of the tornado just below cloud base looked like a vertically oriented cylinder (Fig. 6a). Surrounding the tornado above cloud base there was a much wider cylinder of cyclonically rotating cloud material. Strong westerly winds were observed at the location of the field crew; blowing dust was noted just south of the road over a plowed field.

Doppler wind spectra were calculated every 78 m in range from the radar out to the maximum unambiguous range of 5 km. At 5-km range the half-power cross section of the radar beam was about 400 m, which is comparable to the 540-m width of the condensation funnel (Fig. 7) estimated from photogrammetric analysis (Holle 1988) of the photograph in Fig. 6a. The spectra are viewed most conveniently in mesh diagrams of spectral density viewed from the following four different vantage points (Fig. 8): (a) receding velocities looking toward the range of the tornado, (b) approaching velocities looking toward the range of the tornado, (c) receding velocities looking toward the radar, and (d) approaching velocities looking toward the radar; the vertical scale is not logarithmic as in earlier studies of tornado wind spectra (e.g., Bluestein et al. 1993) for clarity, owing to the noisier character of the FM-CW spectra. A constant-amplitude signal was added to the data in all range bins at zero Doppler velocity to guide the eye in separating the positive-velocity and negative-velocity halves of the mesh plot. The relatively high spectral densities around 4–5 km in range at about −110 and 85 m s−1 are artifacts due to range folding from backscattered radiation beyond 5 km in range. Other artifacts appear at somewhat lower spectral density at approximately −90 m s−1.

Doppler velocities not associated with ground clutter first appear about 3 km in range (Fig. 8). Highest wind velocities are found between 4 and 4.75 km in range. Most of the relatively high relative spectral densities at ranges beyond 3 km are associated with Doppler wind velocities as high as 45 m s−1. Velocities as high as 65 m s−1 are indicated by much lower spectral densities. It therefore appears as if regions of 45 m s−1 wind are relatively broad and/or are represented by highly reflective scatterers; the regions of higher wind speeds are probably much smaller in volume and/or are represented by only weakly reflective scatterers. In general, the receding parts of the spectrum have lower spectral densities spread over a wider range in velocity while the stronger, approaching parts of the spectrum are confined to a narrower range in velocity. Earlier studies of CW spectral density found higher spectral densities in the receding part of the spectrum (Bluestein et al. 1993), which may be due to higher reflectivities in heavy precipitation cores to the right of tornadoes that are viewed from the southeast or east. Unfortunately, it was not possible to correlate the precise location viewed by the radar with the tornado because the contrast of the boresighted video was poor since the sun was shining on the tornado and on precipitation, which was being wrapped around it. However, our best guess is that the radar volume extended from just above the ground up to cloud base.

Mesh diagrams of spectral plots from two vantage points representative of data collected when the radar was looking at the left side of the tornado (according to voice documentation) indicate maximum approaching velocities around 4 km in range (Fig. 9). The spectra shown in Fig. 9 are particularly “clean,” perhaps because precipitation had been carried around the tornado and increased the radar reflectivity on the left side of the tornado relative to what it more commonly is, or because the radar power had been set to a relatively high level. Spectra in eight subjectively determined representative range bins (Fig. 10) show that the maximum approaching wind speeds increase from about 32 m s−1 at 3.51 km to around 53 m s−1 at 4.14 km, possibly to as high as 60–65 m s−1 at 4.37 km, and back down to about 40 m s−1 at 4.53 km. Since the data displayed in Fig. 10 are from individual video frames the spectra are rather noisy and therefore must be viewed with some caution.

A damage survey of the Northfield tornado conducted on 4 June 1994, a week and a half after the tornado (Fig. 11), shows that the damage inflicted was relatively minor (Fig. 12). However, since most of the tornado path was over open country it is difficult to estimate the maximum wind speeds based upon the damage. Since the area traversed by the tornado had very few man-made structures, it is likely that the highest wind speeds associated with the tornado did not affect any of them. It is not known how much other damage had been removed since the tornado had struck. Figure 8 was blown up to a large size and used to determine subjectively the maximum Doppler wind speeds as a function of range, by subjectively determining the highest wind speed at which the spectra entered the noise floor. The maximum Doppler velocities in the tornado of 50–65 m s−1 were found in the receding velocities at a range of 4–4.5 km (Fig. 13). Since the southern extent of the damage (Fig. 11) was also approximately 4–4.5 km away, the FM-CW Doppler spectra are consistent with the damage. The maximum approaching Doppler velocities of 50–60 m s−1 were found at longer range, near 4.5 km. The difference in ranges of the maximum receding and approaching velocities suggests a pattern of convergence (Brown and Wood 1991). One would expect to find convergence below cloud base in a tornado as a result of turbulent friction and/or if it were being spun up through vertical stretching. The maximum receding Doppler velocities exceeded the maximum approaching velocities by approximately 3 m s−1 (Fig. 13), which is qualitatively consistent with tornado motion having a component away from the radar (Fig. 11). However, since the maximum Doppler wind velocity spectral densities resolved by the FM-CW Doppler radar (Fig. 8) were relatively low, our interpretation of the spectra must be made with caution. Wind speeds in excess of those already noted and high wind speeds at other ranges may not have been resolved because their spectral densities may have been below the noise floor.

c. CW Doppler data

Our CW Doppler wind spectra may be used to estimate more accurately the maximum wind speeds in the tornado because the signal-to-noise ratio of CW spectra recovered from our portable radar is higher than that of FM-CW spectra. The CW spectra furthermore may be used as a quantitative check on the FM-CW spectra.

Although wind speeds possibly as high as 75 m s−1 are indicated in a spectrum (Fig. 14) representative of the left side of the tornado and wind speeds of 75–80 m s−1 are indicated on the right side, most spectral density is found at velocities of 50 m s−1 or less. More spectral power is found for receding velocities than for approaching velocities, as found in the FM-CW spectra (Fig. 8). The slow dropoff in spectral density found on the left side of the tornado is characteristic of a combined Rankine vortex in which the maximum radar reflectivity lies outside the radius of maximum wind (Bluestein et al. 1993). The spectral density on the right side, however, drops off more precipitously.

CW spectra not focused directly on the tornado (not shown) had maximum Doppler velocities of 50 m s−1, which is consistent with the core of the FM-CW tornado wind spectra.

The consistency of the FM-CW spectra was checked by comparing the FM-CW spectra shown in Fig. 9 integrated over the region that contained most of the tornado vortex and its parent circulation with a CW spectrum taken almost at the same time (Fig. 15). Such a comparison must be qualitative because the sensitivity of the radar is not the same for each. Both show maximum approaching velocities of around 60 m s−1 and very little power at receding velocities in excess of about 15 m s−1.

d. Photogrammetric analyses

Using a video of the tornado, a simultaneous photograph of the tornado taken by a still camera having a lens of known focal length, and the damage survey, cloud tags of the cylindrically shaped cloud just above the tornado were photogrammetrically analyzed and tangential wind speeds estimated. It was assumed the cloud tags were moving along with the wind and that the motion was an arc about a circle whose diameter was the width of the cloud cylinder. Most of the tangential wind speeds were estimated to be around 40–45 m s−1 (Fig. 7), which is consistent with the FM-CW and CW Doppler spectra. However, the photogrammetric-analysis estimates are valid at cloud base and above, while the Doppler spectra are valid at and below cloud base. To the extent that both techniques measured wind speeds near cloud base, and assuming that the cross line-of-sight velocities measured using the photogrammetric technique are the same as the along line-of-sight velocities measured by the radar, there is good agreement between the two techniques.

The width of the tornado’s damage path (Fig. 11) was wider than the width of the tornado condensation funnel at cloud base (Fig. 7). However, the width of the cylindrically shaped parent cloud above was much wider than the damage path. Since the cylindrically shaped cloud above the tornado is over 2 km wide, it is probably associated with a mesocyclone, the tornado’s parent, larger-scale vortex.

There is some indication that the damage path turned to the left, as is sometimes observed, and became narrower later on in the tornado’s life (Fig. 11). The former behavior has been ascribed to rotation about a larger-scale vortex (Agee et al. 1976). Viewed from the location of the radar, the tornado disappeared behind a veil of precipitation, and a bell-shaped, flared base became visible (Fig. 6b). A ropelike tornado then appeared briefly at 2300. We suspect that the latter may have been the same tornado, but not visible owing to poor visibility. Since a detailed aerial damage survey was not conducted, the details of the path width and exact location are not known precisely.

e. Airborne Doppler data

At flight level, which was near 1 km AGL, the NOAA P-3 aircraft got as close as 7 km to the southern tip of the hook echo in the Northfield storm (Fig. 16). The aircraft was in cloud until it reached as far east as the eastern edge of the hook. It then went back into cloud about 15 km east southeast of the hook. The strongest wind speeds were encountered just southeast of the hook.

A pseudo-dual-Doppler analysis at 1 km AGL of the reflectivity and horizontal storm-relative wind fields of the parent storm for a 2-min period (Fig. 17a), approximately 10–15 min before the tornado appeared, and 20–25 min before the time valid for the FM-CW and CW data shown in Figs. 8 and 14, affords a closer analysis of the hook echo seen by the Lubbock, Texas, WSR-88D radar (Fig. 4b) at the same time at low elevation angle. The hook is located near and along an elliptically shaped cyclonic circulation/shear zone, in which the vorticity is due mainly to shear between southeasterlies ahead of the hook and north-northwesterlies behind the hook. [Unfortunately, the new convective cells that had moved in from the southwest (Figs. 4c–f) made it impossible for the P-3 aircraft to fly another pass near the storm while the tornado was on the ground. The data analyzed herein represent only a “snapshot” view of the storm; it was not possible to ascertain the evolution of the wind field.] Since the radial from the WSR-88D radar at Lubbock that passed through the hook at 0.5° elevation angle is nearly normal to the winds, it is not surprising that the cyclonic shear signature was relatively weak.

The estimate of the motion of the tornado based upon WSR-88D cell motion (Fig. 4) and the damage survey (Fig. 11) is approximately from 255° at 10 m s−1. From Figs. 11 and 16 we extrapolated the location of the tornado back to the pseudo-dual-Doppler analysis of the wind field earlier to estimate where within the parent storm the tornado had developed. This suggests that the tornado had formed within about 1–2 km of the center of the elliptically shaped storm-relative cyclonic circulation/shear zone. Some of the slight uncertainty of the extrapolated tornado position and storm-scale circulation could be due to errors in the estimate of the translational speed of the tornado (an error of translational wind speed of 2 m s−1 could account for over 2 km in distance); alternatively, some of the uncertainty could be due to subsequent cyclonic wrapping of the hook, in which the motion of the southern edge of the hook was faster than that of the main storm cell itself, or to errors in the estimate of the time interval between the pseudo-dual-Doppler analysis and the time the tornado was located at a known location along its damage path.

The storm-relative cyclonic circulation was also elliptically oriented at 2 km AGL (Fig. 17b); however, the major axis of the circulation had a northwest-to-southeast orientation, rather than a north-northwest by south-southeast orientation as at 1 km. The hook was more closed off, with a narrower reflectivity notch ahead of it. A smaller anticyclone was also noted to the northwest of the cyclone; the anticyclone was reflected only as a weak region of anticyclonic curvature at 1 km.

At 3 km AGL the storm-relative anticyclone was still evident; the cyclone opened up into an intense trough whose axis was oriented from northwest to southeast (Fig. 17c). The hook below was closed off at 3 km as a BWER. The area of the BWER was surrounded by a larger 8–10-km scale cyclonic circulation. The BWER had almost completely disappeared by 5 km (Fig. 17d). The BWER of a storm is characterized by a strong updraft (Lemon and Doswell 1979). It was indeed found that strong upward vertical velocities were located near the BWER (not shown). However, owing to the uncertainties in the Doppler velocities, we do not present and discuss details of the vertical-motion field. In addition, the small anticyclone noted below was not apparent at 5 km AGL.

The extent of the BWER is vividly seen in individual scans, which are not objectively analyzed and therefore not spatially filtered, so that the highest possible spatial resolution is retained. The conic sections may be interpreted crudely as RHIs (range–height indicators). The BWER is as narrow as only a few kilometers and extends up as high as 8–9 km AGL (Fig. 18). Above the BWER there is a small region of high reflectivity. It is apparent that the updraft was narrow and tall just 10–15 min before the tornado touched down.

4. Summary and discussion

A unique combination of ground-based and airborne Doppler radars was used in conjunction with photographs and videos and a ground-based damage survey to document a large tornado during VORTEX-94. FM-CW Doppler spectra of a tornado and its parent vortex were obtained for the first time. In addition, airborne Doppler radar data were collected just before a tornado touched down, when portable radar data were also being collected.

It was found that the parent storm evolved as had been previously documented in other supercells. However, the Northfield storm developed an unusually large hook echo that was resolvable even by WSR-88D radar at relatively long range. Pseudo-dual-Doppler analyses of airborne radar data revealed that an elliptically shaped storm-relative cyclonic circulation/shear zone was located along the leading edge of a hook echo. Analysis of the damage path of the tornado and storm cell motion suggested that the tornado may have developed within the circulation/shear zone. Because the hook echo was so large, it is likely that it represented both advection of precipitation from the north around the mesocyclone and new cell growth along the flanking line of the storm. The nature of the companion smaller storm-relative anticyclone between 2 and 4 km AGL is not known.

Maximum wind speeds of 65 m s−1 were noted in FM-CW Doppler wind velocity spectra within the damage path of the tornado, which was wider than the condensation funnel at cloud base. Bluestein et al. (1993) had also found that tornado damage paths were wider than the tornado condensation funnel at cloud base. Analyses of the CW Doppler wind velocity spectra, which are more sensitive than the FM-CW spectra, indicated wind speeds as high as 50–60 m s−1 and possibly as high as 75 m s−1 or more. Since the cross section of the radar beam at the range of the tornado was about 400 m and since high wind velocities were detected for both approaching and receding motion (e.g., Figs. 8c,d) within volumes 78 m × 400 m × 400 m, one can infer that the vortex detected in the spectra was contained within a volume at most that wide (i.e., 400 m); since the visible funnel was 540 m across, it is clear that a vortex on the tornado scale was in fact being detected. Analyses of FM-CW Doppler spectra, analyses of CW Doppler spectra, and photogrammetric analysis of cloud tags in a videotape all suggested that wind speeds of 45–50 m s−1 were associated with the tornado’s parent vortex (mesocyclone), which was approximately 2 km in diameter. There was, however, no clear separation in the Doppler spectra between the region representing the tornado and the region representing its parent vortex. Airborne Doppler analyses 20–25 min prior to the latter analyses did not reveal mesocyclone winds as strong as 45–50 m s−1. It is therefore likely that significant evolution had occurred during the 20–25-min period and that much higher temporal resolution is necessary to document tornadogenesis, or that the aircraft radar lacked the resolution or sensitivity to detect the regions of higher wind speeds.

Convective storms that moved in from the southwest interacted with the Northfield storm and may have been responsible for the storm’s dissipation. Nowcasters should be aware that a cell appearing on radar whose extrapolated track intersects with that of an existing supercell storm might interfere with the storm’s warm, moist inflow and cause the storm to weaken.

More observational studies are needed in which ground-based portable and mobile (Bluestein et al. 1995) Doppler radars and/or airborne radars are used to gather information about the wind field in developing tornadoes and their parent circulations. It is important that simultaneous data on both the tornado scale and the storm scale be collected for extended periods of time. FM-CW and CW Doppler wind velocity spectra allow one to estimate maximum wind speeds, which might otherwise not be detectable, owing to resolution problems. A real-time FM-CW processor is currently being designed, which will allow us to make full use of the dynamic range of the radar and more easily obtain wind spectra. Future field programs may also involve a mobile 3-mm wavelength, 10–30-m resolution, pulsed Doppler radar developed at the University of Massachusetts (Bluestein et al. 1995) and a mobile 3-cm, pulsed Doppler radar developed at the University of Oklahoma and NCAR (the National Center for Atmospheric Research) (Wurman et al. 1995). The former could be used to map out the tornado’s Doppler velocity field, while the latter could be used to map out the parent vortex’s Doppler velocity field.

Acknowledgments

This research was funded by NSF Grant ATM-9302379. The VORTEX field coordinator from NSSL and colleagues provided timely information during the experiment. Some of this work was done while the first author was a summer visitor at the NCAR in the MMM (Mesoscale and Microscale Meteorology) Division. The Graphics Services group at NCAR assisted with some of the figures. Irv Watson at NSSL provided the airborne Doppler radar data. We are grateful for the efforts of the crew of the NOAA P-3 in collecting the Doppler data. Workstations used to analyze the data were funded in part by the School of Meteorology, the College of Geosciences, and Graduate College at the University of Oklahoma. Airborne Doppler radar data were edited, objectively analyzed, and synthesized into a three-dimensional wind field using the rdss (Research Data Support System), reorder, and cedric software packages developed at NCAR (National Center for Atmospheric Research). A group at the Los Alamos National Laboratory was responsible for designing and building the portable radar system. The storm-intercept van and its maintenance were provided by the University of Oklahoma. NCAR is supported by the National Science Foundation.

REFERENCES

  • Agee, E. M., J. T. Snow, and P. R. Clare, 1976: Multiple vortex features in the tornado cyclone and the occurrence of tornado families. Mon. Wea. Rev.,104, 552–563.

  • Blanchard, D. O., 1992: Analysis of a tornadic supercell using airborne Doppler radar. Preprints, 11th Int. Conf. on Clouds and Precipitation, Vol. 2, Montreal, PQ, Canada, Amer. Meteor. Soc., 777–780.

  • Bluestein, H. B., and W. P. Unruh, 1989: Observations of the wind field in tornadoes, funnel clouds, and wall clouds with a portable Doppler radar. Bull. Amer. Meteor. Soc.,70, 1514–1525.

  • ——, and J. H. Golden, 1993: A review of tornado observations. The Tornado: Its Structure, Dynamics, Prediction, and Hazards, C. Church, D. Burgess, C. Doswell, and R. Davies-Jones, Eds., Amer. Geophys. Union, 319–352.

  • ——, and W. P. Unruh, 1993: On the use of a portable FM-CW Doppler radar for tornado research. The Tornado: Its Structure, Dynamics, Prediction, and Hazards, C. Church, D. Burgess, C. Doswell, and R. Davies-Jones, Eds., Amer. Geophys. Union, 367–376.

  • ——, J. G. LaDue, H. Stein, D. Speheger, and W. P. Unruh, 1993: Doppler radar wind spectra of supercell tornadoes. Mon. Wea. Rev.,121, 2200–2221.

  • ——, A. L. Pazmany, J. C. Galloway, and R. E. McIntosh, 1995: Studies of the substructure of severe convective storms using a mobile 3-mm-wavelength Doppler radar. Bull. Amer. Meteor. Soc.,76, 2155–2169.

  • Brown, R. A., and V. T. Wood, 1991: On the interpretation of single-Doppler velocity patterns within severe thunderstorms. Wea. Forecasting,6, 32–48.

  • ——, L. R. Lemon, and D. W. Burgess, 1978: Tornado detection by pulsed Doppler radar. Mon. Wea. Rev.,106, 29–38.

  • Cressman, G. P., 1959: An operational objective analysis system. Mon. Wea. Rev.,87, 367–374.

  • Doviak, R. J., and D. S. Zrnić, 1984: Doppler Radar and Weather Observations. Academic Press, 458 pp.

  • ——, P. S. Ray, R. G. Strauch, and L. J. Miller, 1976: Error estimation in wind fields derived from dual-Doppler radar measurement. J. Appl. Meteor.,15, 868– 878.

  • Dowell, D. C., H. B. Bluestein, and D. P. Jorgensen, 1997: Airborne Doppler radar analysis of supercells during COPS-91. Mon. Wea. Rev., in press.

  • Fiedler, B. H., and R. Rotunno, 1986: A theory for the maximum windspeeds in tornado-like vortices. J. Atmos. Sci.,43, 2328–2340.

  • Foote, G. B., and P. S. duToit, 1969: Terminal velocity of raindrops aloft. J. Appl. Meteor.,8, 249–253.

  • Fujita, T. T., 1981: Tornadoes and downbursts in the context of generalized planetary scales. J. Atmos. Sci.,38, 1511–1534.

  • Golden, J. H., and D. Purcell, 1978: Life cycle of the Union City, Oklahoma, tornado and comparison with waterspouts. Mon. Wea. Rev.,106, 3–11.

  • Hane, C. E., C. L. Ziegler, and H. B. Bluestein, 1993: Investigation of the dryline and convective storms initiated along the dryline: Field experiments during COPS-91. Bull. Amer. Meteor. Soc.,74, 2133–2145.

  • Hildebrand, P. H., and C. K. Mueller, 1985: Evaluation of meteorological airborne Doppler radar. Part I: Dual-Doppler analyses of air motions. J. Atmos. Oceanic Technol.,2, 362–380.

  • Holle, R. L., 1988: Photogrammetry of thunderstorms. Instruments and Techniques for Thunderstorm Observation and Analysis, 2d ed. E. Kessler, Ed., University of Oklahoma Press, 51–63.

  • Jorgensen, D. P., and B. F. Smull, 1993: Mesovortex circulations seen by airborne Doppler radar within a bow-echo mesoscale convective system. Bull. Amer. Meteor. Soc.,74, 2146–2157.

  • ——, P. H. Hildebrand, and C. L. Frush, 1983: Feasibility test of an airborne pulse-Doppler meteorological radar. J. Climate Appl. Meteor.,22, 744–757.

  • ——, T. J. Matejka, and J. D. DuGranrut, 1995: Multi-beam techniques for deriving wind fields from airborne Doppler radars. J. Meteor. Atmos. Phys.,58, 83–104.

  • Lemon, L. R., 1977: New severe thunderstorm radar identification techniques and warning criteria: A preliminary report. NOAA Tech. Memo. NWS-NSSFC 1, 60 pp. [NTIS PB-273049.].

  • ——, and C. A. Doswell III, 1979: Severe thunderstorm evolution and mesocyclone structure as related to tornadogenesis. Mon. Wea. Rev.,107, 1184–1197.

  • Rasmussen, E. N., J. M. Straka, R. Davies-Jones, C. A. Doswell, F. H. Carr, M. D. Eilts, and D. R. MacGorman, 1994: Verification of the Origins of Rotation in Tornadoes Experiment: VORTEX. Bull. Amer. Meteor. Soc.,75, 995–1006.

  • Ray, P. S., and D. P. Jorgensen, 1988: Uncertainties associated with combining airborne and ground-based Doppler radar data. J. Atmos. Oceanic Technol.,5, 177–196.

  • ——, and M. Stephenson, 1990: Assessment of the geometric and temporal errors associated with airborne Doppler radar measurements of a convective storm. J. Atmos. Oceanic Technol.,7, 206–217.

  • ——, D. P. Jorgensen, and S. L. Wang, 1985: Airborne Doppler radar observations of a convective storm. J. Climate Appl. Meteor.,24, 687–698.

  • Snow, J. T., and R. L. Pauley, 1984: On the thermodynamic method for estimating maximum tornado windspeeds. J. Climate Appl. Meteor.,23, 1465–1468.

  • Strauch, R. S., 1976: Theory and application of the FM-CW Doppler radar. Ph.D. thesis, University of Colorado, 97 pp. [Available from Dept. of Electrical and Computer Engineering, Room 1B55, Campus Box 425, University of Colorado, Boulder, CO 80309-0425.].

  • Wakimoto, R. M., and B. E. Martner, 1992: Observations of a Colorado tornado. Part II: Combined photogrammetric and Doppler radar analysis. Mon. Wea. Rev.,120, 522–543.

  • Watson, A. I., D. O. Blanchard, D. P. Jorgensen, and D. W. Burgess, 1993: The kinematic structure of a supercell thunderstorm seen by airborne Doppler radar. Preprints, 26th Int. Conf. on Radar Meteorology, Norman, OK, Amer. Meteor. Soc., 209–211.

  • Weisman, M. L., and J. B. Klemp, 1982: The dependence of numerically simulated convective storms on vertical wind shear and buoyancy. Mon. Wea. Rev.,110, 504–520.

  • Welch, P. D., 1967: The use of the fast Fourier transform for the estimation of power spectra. IEEE Trans. Audio Electroacoustics,AU15, 70–73.

  • Wurman, J., J. M. Straka, E. N. Rasmussen, M. Randell, and A. Zahrai, 1995: Design and first results from a portable pencil-beam pulsed Doppler radar. Preprints, 27th Conf. on Radar Meteorology, Vail, CO, Amer. Meteor. Soc., 713–716.

  • Zrnić, D. S., and R. J. Doviak, 1975: Velocity spectra of vortices scanned with a pulse-Doppler radar. J. Appl. Meteor.,14, 1531–1539.

Fig. 1.
Fig. 1.

Photograph of the portable, 3-cm, FM-CW/CW Doppler radar, operated by University of Oklahoma graduate students D. Dowell (left), A. Wood (center), and T. Hutchinson (right), probing the Northfield, Texas, tornado on 25 May 1994 at approximately 2244 UTC (copyright H. Bluestein).

Citation: Monthly Weather Review 125, 2; 10.1175/1520-0493(1997)125<0212:DRAOTN>2.0.CO;2

Fig. 2.
Fig. 2.

Frequency of radiation transmitted (fT) and received (fR) as a function of time in the FM-CW mode for the portable Doppler radar shown in Fig. 1. Here, B is the “sweep width” and T is the period between sweeps (1/T is the “sweep repetition frequency”).

Citation: Monthly Weather Review 125, 2; 10.1175/1520-0493(1997)125<0212:DRAOTN>2.0.CO;2

Fig. 3.
Fig. 3.

Fore–aft scanning technique (FAST) of the NOAA P-3 airborne, 3-cm Doppler radar. (a) Here, θ is the angle swept about the aircraft track, and α is the angle (fore/aft) made by each sweep with respect to a plane normal to the flight track. (b) Network of intersecting fore and aft beams; d is the spacing between consecutive fore/aft beams along the flight track; r is the range to a point of intersecting fore/aft beams.

Citation: Monthly Weather Review 125, 2; 10.1175/1520-0493(1997)125<0212:DRAOTN>2.0.CO;2

Fig. 4.
Fig. 4.

Time series of northeastern sector of PPI (plan position indicator) scans of radar reflectivity factor (dBZ, color scale at bottom of each image) at 0.5° elevation angle from the WSR-88D radar at Lubbock, Texas, on 25 May 1994 at approximately 15-min intervals. Range markers are given in kilometers; (a) 2130, (b) 2145, (c) 2200, (d) 2215, (e) 2231, (f) 2246, and (g) 2301 UTC.

Citation: Monthly Weather Review 125, 2; 10.1175/1520-0493(1997)125<0212:DRAOTN>2.0.CO;2

Fig. 5.
Fig. 5.

Northeastern sector of PPI scans from the WSR-88D radar at Lubbock, Texas, at 2221 UTC 25 May 1994. Range markers are given in kilometers. (a) Radar reflectivity factor at 0.5° elevation angle; (b) as in (a) but for Doppler velocity (m s−1, color scale at bottom of image); (c) as in (a) but for 1.5° elevation angle; (d) as in (b) for 1.5° elevation angle.

Citation: Monthly Weather Review 125, 2; 10.1175/1520-0493(1997)125<0212:DRAOTN>2.0.CO;2

Fig. 6.
Fig. 6.

Photographs (a) of the Northfield, Texas, tornado at approximately 2242 UTC 25 May 1995, with a 50-mm lens; view is to the north. (b) As in (a) but at approximately 2244, with a 28-mm lens; view is to the northeast (copyright H. Bluestein).

Citation: Monthly Weather Review 125, 2; 10.1175/1520-0493(1997)125<0212:DRAOTN>2.0.CO;2

Fig. 7.
Fig. 7.

Outline of the Northfield, Texas, tornado and its parent cloud, from a videotape taken by H. Stein. Width of the condensation funnel at cloud base, height of the cloud base, and width of the cylindrical cloud base above the tornado are indicated. Arrows indicate direction of cloud-tag motion in the plane of video and photogrammetrically analyzed tangential wind speeds (m s−1); length of arrows denotes projection of cloud-tag motion, not the wind speed (both different time intervals for tracking the cloud tags and differential motion are responsible for the different arrow lengths).

Citation: Monthly Weather Review 125, 2; 10.1175/1520-0493(1997)125<0212:DRAOTN>2.0.CO;2

Fig. 8.
Fig. 8.

Mesh plots of FM-CW Doppler wind spectra as a function of range from the radar from 1 km out to 5 km, the maximum unambiguous range, at approximately 2244 UTC 25 May 1994. Relative spectral density is given by the vertical height. Spectral densities exceeding a certain threshold are clipped off for clarity. The range resolution is 78 m; the velocity scale is given every 1.8 m s−1. The “fence” at zero Doppler wind velocity is added to aid the reader. Viewed from above and (a) from the radar, on the receding-velocity (positive Doppler velocity) side; (b) from the radar, on the approaching-velocity (negative Doppler velocity) side; (c) from the tornado, on the receding-velocity side; and (d) from the tornado, on the approaching-velocity side.

Citation: Monthly Weather Review 125, 2; 10.1175/1520-0493(1997)125<0212:DRAOTN>2.0.CO;2

Fig. 9.
Fig. 9.

As in Figs. 8c (top) and 8b (bottom) but when the radar was probing the left side of the tornado.

Citation: Monthly Weather Review 125, 2; 10.1175/1520-0493(1997)125<0212:DRAOTN>2.0.CO;2

Fig. 10.
Fig. 10.

FM-CW Doppler wind spectra as in Fig. 9 but at selected range bins.

Citation: Monthly Weather Review 125, 2; 10.1175/1520-0493(1997)125<0212:DRAOTN>2.0.CO;2

Fig. 11.
Fig. 11.

Damage survey on 4 June 1994 of the Northfield, Texas, tornado of 25 May 1994.

Citation: Monthly Weather Review 125, 2; 10.1175/1520-0493(1997)125<0212:DRAOTN>2.0.CO;2

Fig. 12.
Fig. 12.

Photographs on 4 June 1994 of some of the damage inflicted by the Northfield, Texas, tornado on 25 May 1994. (a) A shed removed from its foundation; (b) sheet metal and other debris; (c) a downed windmill. Refer to Fig. 11 for the location of the damage depicted in the photographs (copyright D. Dowell).

Citation: Monthly Weather Review 125, 2; 10.1175/1520-0493(1997)125<0212:DRAOTN>2.0.CO;2

Fig. 12.
Fig. 13.
Fig. 13.

Maximum receding (solid line) and approaching (dashed line) Doppler wind velocities indicated in Fig. 8, as functions of range.

Citation: Monthly Weather Review 125, 2; 10.1175/1520-0493(1997)125<0212:DRAOTN>2.0.CO;2

Fig. 14.
Fig. 14.

The CW relative spectral density as a function of Doppler wind velocity in the Northfield, Texas, tornado when the radar was probing the left (top) and right (bottom) sides of the tornado.

Citation: Monthly Weather Review 125, 2; 10.1175/1520-0493(1997)125<0212:DRAOTN>2.0.CO;2

Fig. 15.
Fig. 15.

Relative spectral density as in Fig. 14 but (top) FM-CW spectrum shown in Fig. 9 integrated from 3.6-km to 4.4-km range and (bottom) CW spectrum from approximately the same time, when the radar was probing approximately the same area to the left of the tornado.

Citation: Monthly Weather Review 125, 2; 10.1175/1520-0493(1997)125<0212:DRAOTN>2.0.CO;2

Fig. 16.
Fig. 16.

The flight track of the NOAA P-3 as a function of time (UTC); radar reflectivity factor (dBZ) contours from fore scans at 1 km AGL given for 2219 UTC. Also plotted along flight track are flight levels (m MSL) at 2217:30 and 2222, wind vectors (see key in lower right), temperature (°C), dewpoint (°C), approximate damage path (dashed line), location of the storm-relative circulation in the hook echo at 1 km AGL at 2219 shown in Fig. 17a (marked by the letter “C”), location of the portable CW/FM-CW Doppler radar at 2242 (marked by the letter “O”), and the approximate direction in which the portable radar was pointing (arrow). Scale east (abscissa in kilometers) and north (ordinate in kilometers) of an arbitrary point southwest of the storm.

Citation: Monthly Weather Review 125, 2; 10.1175/1520-0493(1997)125<0212:DRAOTN>2.0.CO;2

Fig. 17.
Fig. 17.

Storm-relative horizontal wind field (vectors, scale shown at the top right) and radar reflectivity (dBZ, solid contours) of the Northfield, Texas, storm at 2219 UTC 25 May 1994, as determined by pseudo-dual-Doppler analysis of data from the NOAA P-3 aircraft. Distance (x) east and (y) north of an arbitrary point southwest of the storm, in kilometers. Elevations are (a) 1, (b) 2, (c) 3, and (d) 5 km AGL.

Citation: Monthly Weather Review 125, 2; 10.1175/1520-0493(1997)125<0212:DRAOTN>2.0.CO;2

Fig. 18.
Fig. 18.

Conic sections of radar reflectivity factor from the aft-looking beam of the NOAA P-3 airborne Doppler radar on 25 May 1994 at (a) 2218:56 and (b) 2219:14 UTC. The cross section depicted in (b) is approximately 2 km east of the cross section depicted in (a).

Citation: Monthly Weather Review 125, 2; 10.1175/1520-0493(1997)125<0212:DRAOTN>2.0.CO;2

Table 1.

Characteristics of the FM-CW/CW portable Doppler radar.

Table 1.
Table 2.

Characteristics of the NOAA P-3 airborne Doppler radar.

Table 2.
Save
  • Agee, E. M., J. T. Snow, and P. R. Clare, 1976: Multiple vortex features in the tornado cyclone and the occurrence of tornado families. Mon. Wea. Rev.,104, 552–563.

  • Blanchard, D. O., 1992: Analysis of a tornadic supercell using airborne Doppler radar. Preprints, 11th Int. Conf. on Clouds and Precipitation, Vol. 2, Montreal, PQ, Canada, Amer. Meteor. Soc., 777–780.

  • Bluestein, H. B., and W. P. Unruh, 1989: Observations of the wind field in tornadoes, funnel clouds, and wall clouds with a portable Doppler radar. Bull. Amer. Meteor. Soc.,70, 1514–1525.

  • ——, and J. H. Golden, 1993: A review of tornado observations. The Tornado: Its Structure, Dynamics, Prediction, and Hazards, C. Church, D. Burgess, C. Doswell, and R. Davies-Jones, Eds., Amer. Geophys. Union, 319–352.

  • ——, and W. P. Unruh, 1993: On the use of a portable FM-CW Doppler radar for tornado research. The Tornado: Its Structure, Dynamics, Prediction, and Hazards, C. Church, D. Burgess, C. Doswell, and R. Davies-Jones, Eds., Amer. Geophys. Union, 367–376.

  • ——, J. G. LaDue, H. Stein, D. Speheger, and W. P. Unruh, 1993: Doppler radar wind spectra of supercell tornadoes. Mon. Wea. Rev.,121, 2200–2221.

  • ——, A. L. Pazmany, J. C. Galloway, and R. E. McIntosh, 1995: Studies of the substructure of severe convective storms using a mobile 3-mm-wavelength Doppler radar. Bull. Amer. Meteor. Soc.,76, 2155–2169.

  • Brown, R. A., and V. T. Wood, 1991: On the interpretation of single-Doppler velocity patterns within severe thunderstorms. Wea. Forecasting,6, 32–48.

  • ——, L. R. Lemon, and D. W. Burgess, 1978: Tornado detection by pulsed Doppler radar. Mon. Wea. Rev.,106, 29–38.

  • Cressman, G. P., 1959: An operational objective analysis system. Mon. Wea. Rev.,87, 367–374.

  • Doviak, R. J., and D. S. Zrnić, 1984: Doppler Radar and Weather Observations. Academic Press, 458 pp.

  • ——, P. S. Ray, R. G. Strauch, and L. J. Miller, 1976: Error estimation in wind fields derived from dual-Doppler radar measurement. J. Appl. Meteor.,15, 868– 878.

  • Dowell, D. C., H. B. Bluestein, and D. P. Jorgensen, 1997: Airborne Doppler radar analysis of supercells during COPS-91. Mon. Wea. Rev., in press.

  • Fiedler, B. H., and R. Rotunno, 1986: A theory for the maximum windspeeds in tornado-like vortices. J. Atmos. Sci.,43, 2328–2340.

  • Foote, G. B., and P. S. duToit, 1969: Terminal velocity of raindrops aloft. J. Appl. Meteor.,8, 249–253.

  • Fujita, T. T., 1981: Tornadoes and downbursts in the context of generalized planetary scales. J. Atmos. Sci.,38, 1511–1534.

  • Golden, J. H., and D. Purcell, 1978: Life cycle of the Union City, Oklahoma, tornado and comparison with waterspouts. Mon. Wea. Rev.,106, 3–11.

  • Hane, C. E., C. L. Ziegler, and H. B. Bluestein, 1993: Investigation of the dryline and convective storms initiated along the dryline: Field experiments during COPS-91. Bull. Amer. Meteor. Soc.,74, 2133–2145.

  • Hildebrand, P. H., and C. K. Mueller, 1985: Evaluation of meteorological airborne Doppler radar. Part I: Dual-Doppler analyses of air motions. J. Atmos. Oceanic Technol.,2, 362–380.

  • Holle, R. L., 1988: Photogrammetry of thunderstorms. Instruments and Techniques for Thunderstorm Observation and Analysis, 2d ed. E. Kessler, Ed., University of Oklahoma Press, 51–63.

  • Jorgensen, D. P., and B. F. Smull, 1993: Mesovortex circulations seen by airborne Doppler radar within a bow-echo mesoscale convective system. Bull. Amer. Meteor. Soc.,74, 2146–2157.

  • ——, P. H. Hildebrand, and C. L. Frush, 1983: Feasibility test of an airborne pulse-Doppler meteorological radar. J. Climate Appl. Meteor.,22, 744–757.

  • ——, T. J. Matejka, and J. D. DuGranrut, 1995: Multi-beam techniques for deriving wind fields from airborne Doppler radars. J. Meteor. Atmos. Phys.,58, 83–104.

  • Lemon, L. R., 1977: New severe thunderstorm radar identification techniques and warning criteria: A preliminary report. NOAA Tech. Memo. NWS-NSSFC 1, 60 pp. [NTIS PB-273049.].

  • ——, and C. A. Doswell III, 1979: Severe thunderstorm evolution and mesocyclone structure as related to tornadogenesis. Mon. Wea. Rev.,107, 1184–1197.

  • Rasmussen, E. N., J. M. Straka, R. Davies-Jones, C. A. Doswell, F. H. Carr, M. D. Eilts, and D. R. MacGorman, 1994: Verification of the Origins of Rotation in Tornadoes Experiment: VORTEX. Bull. Amer. Meteor. Soc.,75, 995–1006.

  • Ray, P. S., and D. P. Jorgensen, 1988: Uncertainties associated with combining airborne and ground-based Doppler radar data. J. Atmos. Oceanic Technol.,5, 177–196.

  • ——, and M. Stephenson, 1990: Assessment of the geometric and temporal errors associated with airborne Doppler radar measurements of a convective storm. J. Atmos. Oceanic Technol.,7, 206–217.

  • ——, D. P. Jorgensen, and S. L. Wang, 1985: Airborne Doppler radar observations of a convective storm. J. Climate Appl. Meteor.,24, 687–698.

  • Snow, J. T., and R. L. Pauley, 1984: On the thermodynamic method for estimating maximum tornado windspeeds. J. Climate Appl. Meteor.,23, 1465–1468.

  • Strauch, R. S., 1976: Theory and application of the FM-CW Doppler radar. Ph.D. thesis, University of Colorado, 97 pp. [Available from Dept. of Electrical and Computer Engineering, Room 1B55, Campus Box 425, University of Colorado, Boulder, CO 80309-0425.].

  • Wakimoto, R. M., and B. E. Martner, 1992: Observations of a Colorado tornado. Part II: Combined photogrammetric and Doppler radar analysis. Mon. Wea. Rev.,120, 522–543.

  • Watson, A. I., D. O. Blanchard, D. P. Jorgensen, and D. W. Burgess, 1993: The kinematic structure of a supercell thunderstorm seen by airborne Doppler radar. Preprints, 26th Int. Conf. on Radar Meteorology, Norman, OK, Amer. Meteor. Soc., 209–211.

  • Weisman, M. L., and J. B. Klemp, 1982: The dependence of numerically simulated convective storms on vertical wind shear and buoyancy. Mon. Wea. Rev.,110, 504–520.

  • Welch, P. D., 1967: The use of the fast Fourier transform for the estimation of power spectra. IEEE Trans. Audio Electroacoustics,AU15, 70–73.

  • Wurman, J., J. M. Straka, E. N. Rasmussen, M. Randell, and A. Zahrai, 1995: Design and first results from a portable pencil-beam pulsed Doppler radar. Preprints, 27th Conf. on Radar Meteorology, Vail, CO, Amer. Meteor. Soc., 713–716.

  • Zrnić, D. S., and R. J. Doviak, 1975: Velocity spectra of vortices scanned with a pulse-Doppler radar. J. Appl. Meteor.,14, 1531–1539.

  • Fig. 1.

    Photograph of the portable, 3-cm, FM-CW/CW Doppler radar, operated by University of Oklahoma graduate students D. Dowell (left), A. Wood (center), and T. Hutchinson (right), probing the Northfield, Texas, tornado on 25 May 1994 at approximately 2244 UTC (copyright H. Bluestein).

  • Fig. 2.

    Frequency of radiation transmitted (fT) and received (fR) as a function of time in the FM-CW mode for the portable Doppler radar shown in Fig. 1. Here, B is the “sweep width” and T is the period between sweeps (1/T is the “sweep repetition frequency”).

  • Fig. 3.

    Fore–aft scanning technique (FAST) of the NOAA P-3 airborne, 3-cm Doppler radar. (a) Here, θ is the angle swept about the aircraft track, and α is the angle (fore/aft) made by each sweep with respect to a plane normal to the flight track. (b) Network of intersecting fore and aft beams; d is the spacing between consecutive fore/aft beams along the flight track; r is the range to a point of intersecting fore/aft beams.

  • Fig. 4.

    Time series of northeastern sector of PPI (plan position indicator) scans of radar reflectivity factor (dBZ, color scale at bottom of each image) at 0.5° elevation angle from the WSR-88D radar at Lubbock, Texas, on 25 May 1994 at approximately 15-min intervals. Range markers are given in kilometers; (a) 2130, (b) 2145, (c) 2200, (d) 2215, (e) 2231, (f) 2246, and (g) 2301 UTC.

  • Fig. 5.

    Northeastern sector of PPI scans from the WSR-88D radar at Lubbock, Texas, at 2221 UTC 25 May 1994. Range markers are given in kilometers. (a) Radar reflectivity factor at 0.5° elevation angle; (b) as in (a) but for Doppler velocity (m s−1, color scale at bottom of image); (c) as in (a) but for 1.5° elevation angle; (d) as in (b) for 1.5° elevation angle.

  • Fig. 6.

    Photographs (a) of the Northfield, Texas, tornado at approximately 2242 UTC 25 May 1995, with a 50-mm lens; view is to the north. (b) As in (a) but at approximately 2244, with a 28-mm lens; view is to the northeast (copyright H. Bluestein).

  • Fig. 7.

    Outline of the Northfield, Texas, tornado and its parent cloud, from a videotape taken by H. Stein. Width of the condensation funnel at cloud base, height of the cloud base, and width of the cylindrical cloud base above the tornado are indicated. Arrows indicate direction of cloud-tag motion in the plane of video and photogrammetrically analyzed tangential wind speeds (m s−1); length of arrows denotes projection of cloud-tag motion, not the wind speed (both different time intervals for tracking the cloud tags and differential motion are responsible for the different arrow lengths).

  • Fig. 8.

    Mesh plots of FM-CW Doppler wind spectra as a function of range from the radar from 1 km out to 5 km, the maximum unambiguous range, at approximately 2244 UTC 25 May 1994. Relative spectral density is given by the vertical height. Spectral densities exceeding a certain threshold are clipped off for clarity. The range resolution is 78 m; the velocity scale is given every 1.8 m s−1. The “fence” at zero Doppler wind velocity is added to aid the reader. Viewed from above and (a) from the radar, on the receding-velocity (positive Doppler velocity) side; (b) from the radar, on the approaching-velocity (negative Doppler velocity) side; (c) from the tornado, on the receding-velocity side; and (d) from the tornado, on the approaching-velocity side.

  • Fig. 9.

    As in Figs. 8c (top) and 8b (bottom) but when the radar was probing the left side of the tornado.

  • Fig. 10.

    FM-CW Doppler wind spectra as in Fig. 9 but at selected range bins.

  • Fig. 11.

    Damage survey on 4 June 1994 of the Northfield, Texas, tornado of 25 May 1994.

  • Fig. 12.

    Photographs on 4 June 1994 of some of the damage inflicted by the Northfield, Texas, tornado on 25 May 1994. (a) A shed removed from its foundation; (b) sheet metal and other debris; (c) a downed windmill. Refer to Fig. 11 for the location of the damage depicted in the photographs (copyright D. Dowell).

  • Fig. 12.

    (Continued)

  • Fig. 13.

    Maximum receding (solid line) and approaching (dashed line) Doppler wind velocities indicated in Fig. 8, as functions of range.

  • Fig. 14.

    The CW relative spectral density as a function of Doppler wind velocity in the Northfield, Texas, tornado when the radar was probing the left (top) and right (bottom) sides of the tornado.

  • Fig. 15.

    Relative spectral density as in Fig. 14 but (top) FM-CW spectrum shown in Fig. 9 integrated from 3.6-km to 4.4-km range and (bottom) CW spectrum from approximately the same time, when the radar was probing approximately the same area to the left of the tornado.

  • Fig. 16.

    The flight track of the NOAA P-3 as a function of time (UTC); radar reflectivity factor (dBZ) contours from fore scans at 1 km AGL given for 2219 UTC. Also plotted along flight track are flight levels (m MSL) at 2217:30 and 2222, wind vectors (see key in lower right), temperature (°C), dewpoint (°C), approximate damage path (dashed line), location of the storm-relative circulation in the hook echo at 1 km AGL at 2219 shown in Fig. 17a (marked by the letter “C”), location of the portable CW/FM-CW Doppler radar at 2242 (marked by the letter “O”), and the approximate direction in which the portable radar was pointing (arrow). Scale east (abscissa in kilometers) and north (ordinate in kilometers) of an arbitrary point southwest of the storm.

  • Fig. 17.

    Storm-relative horizontal wind field (vectors, scale shown at the top right) and radar reflectivity (dBZ, solid contours) of the Northfield, Texas, storm at 2219 UTC 25 May 1994, as determined by pseudo-dual-Doppler analysis of data from the NOAA P-3 aircraft. Distance (x) east and (y) north of an arbitrary point southwest of the storm, in kilometers. Elevations are (a) 1, (b) 2, (c) 3, and (d) 5 km AGL.

  • Fig. 18.

    Conic sections of radar reflectivity factor from the aft-looking beam of the NOAA P-3 airborne Doppler radar on 25 May 1994 at (a) 2218:56 and (b) 2219:14 UTC. The cross section depicted in (b) is approximately 2 km east of the cross section depicted in (a).

All Time Past Year Past 30 Days
Abstract Views 0 0 0
Full Text Views 227 90 27
PDF Downloads 130 36 17