Our knowledge of the process of tornado formation is limited by our ability to map the wind field close to the ground with high resolution in developing tornadoes and by our ability to simulate realistically vortices in the boundary layer. Observations are needed to guide numerical modelers and to test hypotheses for tornado formation, while numerical simulations are needed to help formulate strategies for data collection and to suggest hypotheses for tornado formation. In order to resolve the structure of the tornado and the vortex from which it develops, it is necessary to probe vortices, which are typically only on the order of 100 m across, with instruments that have a horizontal resolution of tens of meters or less. In this paper we describe an attempt to resolve the circulation of an incipient tornado; although a tornado did not actually occur, we were able to document weaker vortices on the tornado scale.
Remote sensing of tornadoes by Doppler radar over the past decade and a half has proven to be a useful and safe method of obtaining wind information in tornadoes on the 100-m scale (Bluestein and Golden 1993). More recently, a portable, low-power, 3-cm-wavelength CW (continuous wave)/FM (frequency modulated)-CW Doppler radar from the Los Alamos National Laboratory (Bluestein and Unruh 1989) has been used to measure wind spectra in tornadoes at close (visual) range (Bluestein et al. 1993, 1997). The advantages of transporting a radar close to tornadoes are as follows: High resolution can be attained, the lowest-elevation beam is close to the ground, simultaneous visual documentation is possible, and the number of tornadoes probed can be increased substantially over what is possible with a fixed-site radar.
During the VORTEX (Verification of the Origins of Rotation in Tornadoes Experiment) field program (Rasmussen et al. 1994) attempts were made to probe tornadoes and their near environment using a variety of new instruments, many of which were mobile. For example, a mobile, 3-mm wavelength (95 GHz), pulsed Doppler radar from the University of Massachusetts at Amherst (hereafter referred to as the UMass radar) (Bluestein et al. 1995) was used to probe convective storms below cloud base with unprecedented spatial resolution. A mobile, 3-cm-wavelength, pulsed Doppler radar was also used at the end of the VORTEX (Wurman et al. 1996). In addition, the ELDORA (Electra Doppler radar) was flown on an Electra aircraft near convective storms (Wakimoto et al. 1996).
On 17 May 1995, during the second year of VORTEX, data were collected in and near a supercell thunderstorm in northeastern Oklahoma by a number of instruments. The UMass radar collected Doppler data with spatial resolution on the order of 35 m. Data from ELDORA were used to map the wind field in the parent storm on the storm scale. Data from a nearby WSR-88D (NEXRAD) Doppler radar (Crum and Alberty 1993) were used to describe the life history of the parent storm. Surface data from the Oklahoma Mesonet (Brock et al. 1995) were used to identify the surface features along which the storm formed and to identify the environment into which the storm moved. Finally, CLASS (Cross-chain Loran Atmospheric Sounding System) soundings released by mobile teams (Rust et al. 1990) were available to describe further the storm environment.
The main purposes of this paper are to describe an analysis of substorm-scale vortices identified by the 3-mm-wavelength UMass radar and to relate them, using independent observing systems, to the structure of their parent storm. Although the storm did not produce a tornado, it is believed that the detection of the vortices is significant because vortices of the same scale could play a role in tornadogenesis. In the following section there is a discussion of the nature of both the ground-based mobile radar and its data, and the airborne radar and its data. Section 3 contains a description of the substorm-scale vortices. The characteristics of the parent storm are detailed in section 4. In section 5 there is summary of our findings and suggestions for future field work.
2. Description of the mobile observing systems
a. The UMass millimeter-wavelength Doppler radar system
Millimeter-wavelength radars are capable of achieving very high spatial resolution of close-range targets with an antenna small enough to be mounted, along with the rest of the radar system, on a mobile platform. For example, to achieve 0.7° half-power beamwidth, an antenna of only 0.3-m diameter is needed; at 3-km range the cross-beam resolution is 36 m. Furthermore, at millimeter wavelengths the sensitivity to tiny targets, such as cloud droplets, is higher than at longer wavelengths.
The UMass 3-mm-wavelength (95 GHz) pulsed system (Table 1) was mounted in a 12-passenger storm-intercept van at the University of Oklahoma (OU) in Norman (Fig. 1) and used successfully during the spring of 1993 and 1994 to probe severe thunderstorms in the southern Plains and nearby regions (Bluestein et al. 1995). During data collection the antenna was raised through an opening in the roof of the van by a hydraulic jack. The radar was powered by a 3-kW generator that was mounted on a platform outside the rear of the van. A video monitor for a video camera recorder mounted just underneath the antenna, a computer that did the signal processing and controlled the scanning of the antenna, a video display for the radar, and other peripherals were located in the back of the van.
Only one operator was required to run the radar system. Two front and two rear load levelers were lowered to provide stability during data collection. The generator was usually started at least 15 min before data collection. Raising the antenna through the hatch and initiating scan sequences took only a few minutes. It similarly took only a minute or so to close the hatch, raise the load levelers, and begin to move to another location.
The main problems with millimeter-wavelength pulsed systems are severe attenuation and a relatively low maximum unambiguous velocity for a reasonable maximum unambiguous range (i.e., a range out to 10 km or farther). The latter problem was alleviated substantially by using a polarization-diversity pulse-pair (PDPP) technique (Doviak and Sirmans 1973), in which both vertically and horizontally polarized pulses at different pulse repetition frequencies (Fig. 2) are processed; a maximum unambiguous velocity of ±80 m s−1 is achieved using the PDPP technique, rather than only ±12 m s−1 using the traditional pulse-pair processing technique (Table 1). In practice, the PDPP velocity field is used to unfold the ordinary pulse-pair estimated velocity field, rather than used as is, because the former is noisier. It was found that the radar could penetrate heavy precipitation 1–1.5 km at a range of 3 km or less and that scattering from targets at ranges greater than about 10 km did not affect the measurements noticeably.
The radar reflectivity factor measured by millimeter-wavelength radar systems is not interpreted the same way as that measured by conventional centimeter-wavelength systems. Radar reflectivities greater than 20 dBZe are seldom measured at ranges beyond a few kilometers owing to Mie scattering (Lhermitte 1990) and attenuation. The difference between radar reflectivity factor in dBZ and dBZe at 95 GHz of larger raindrops (i.e., those greater than 1.5 mm in diameter) can be over 10 dBZ; attenuation at 95 GHz in 25–30-dBZ precipitation (1.5 mm h−1 rainfall rate) can be several dBZ per kilometer. Since the diameter of the majority of drops in a 25–30-dBZ rain is at least 1.5 mm (Laws and Parsons 1943), the measured value of dBZe would be about 15–20 dBZe less the attenuation during the two-way propagation. Heavier rain would likely contain even larger drops, which would contribute to a larger difference between dBZ and dBZe. The result is that the radar equivalent reflectivity factor at 95 GHz in rain “compresses” or “saturates” between 20 and 30 dBZe. Since during long-range measurements the two-way attenuation further reduces the measured reflectivity, we seldom measure greater than 20 dBZe beyond a few kilometers.
The ELDORA [its full name is ELDORA/ASTRAIA (Analyese Stereoscopic par Impulsions Aeroporte)] is an airborne, 3-cm-wavelength, pulsed Doppler radar system (Table 2) developed by the National Center for Atmospheric Research (NCAR) and the Centre de Recherche en Physique de l’Environment Terrestre et Planetaire in Paris, France (Wakimoto et al. 1996). Unique aspects of ELDORA include increased sensitivity compared to that of earlier airborne radars by averaging pulses independently transmitted at four different frequencies, increased maximum unambiguous velocity compared to that of earlier airborne radars by using a dual-PRF (pulse repetition frequency) signal processing technique, and increased along-the-track resolution compared to that of earlier airborne radars due to a fast scanning rate.
The scanning strategy of the antennas, fore–aft scanning technique (Lee et al. 1994; Jorgensen et al. 1995), is identical to that employed by the airborne radar on the NOAA P-3 (Bluestein et al. 1997) (however, the ELDORA uses two antennas while the NOAA P-3 radar uses only one antenna; the ELDORA scans fore and aft simultaneously while the NOAA P-3 radar scans alternately fore and aft): Circular scans at fixed angles fore and aft from the flight track (Fig. 3) sweep out cones and create a helical spatial data sampling pattern. Data from intersecting fore and aft beams are used to synthesize the three-dimensional wind field using pseudo-dual-Doppler analysis (Jorgensen et al. 1983). The time difference between two intersecting fore and aft beams is 1 min for each 10 km in range from the flight track. Since the Electra flew approximately 10–20 km from the portions of the storm on which we focus our interest, we must assume that the storm in its reference frame was in a steady state for 1–2 min.
Before the airborne radar data were subjected to pseudo-dual-Doppler analysis, the components of the Electra’s ground-relative velocity in the fore and aft directions were removed (Lee et al. 1994). The data were further edited to eliminate erroneous data. Corrections for velocity folding were relatively easy (in fact there were none) because the maximum unambiguous velocity is high compared to the maximum wind velocities in this storm (Table 2). Wind and reflectivity data were interpolated to a Cartesian grid having both horizontal and vertical spacing of 500 m, using a Cressman weighting function (Cressman 1959) having an influence radius of 600 m in both the horizontal and vertical.
The three-dimensional wind field was calculated iteratively using an initial guess of zero vertical velocity. Vertical velocity was computed using zero boundary conditions at the lowest grid level (0 km AGL) and at the highest grid level (15 km) and by integrating the anelastic equation of continuity upward. The component of vertical motion due to the terminal velocity of the scatterers was removed using an empirical relationship between radar reflectivity factor and raindrop terminal velocity. Since we are most concerned with the wind field in the storm at very low elevation angles, errors in horizontal velocity at low levels (near cloud base) close to the flight level of the Electra, which lead to vertical velocity errors, should be small enough not to invalidate the qualitative sense of the vertical velocity field calculated kinematically. At each point the greater reflectivity value of the two intersecting beams was used to reduce errors in reflectivity due to attenuation.
3. Description of the substorm-scale vortices
a. The visual appearance of the vortices
We choose to describe the substorm-scale vortices in the storm first and in the subsequent section describe their larger-scale context. The structure of the cloud features we probed with the UMass radar were documented (Fig. 4) by the video camera mounted underneath the radar antenna on top of the van. The cloud base was shaped like the letter “V” rotated in the clockwise direction by 90°, with the apex of the “V,” or notch, located on the left side of our field of view. The nearer cloud base, which curved around to our east and southeast, had been accompanied by a wind shift from southerly or southeasterly to westerly or northwesterly and by an apparent drop in temperature.
A cyclonic swirl was evident to the right of the notch, in the nearer cloud base. It was this feature that was the focus of our attention. A wall cloud had been visible minutes earlier in the general vicinity of the cyclonically rotating swirl in cloud base. No other evidence of small-scale rotation was visible. Cloud-to-ground lightning activity appeared frequently to the left of the notch.
A very small-scale billow is seen wrapping up in the downward direction along the edge of the cloud near the notch. It appeared in the video as if sinking motion were occurring along the edge of the billow.
b. The radar depiction of the vortices
A pair of nearly mirror image hook echoes is seen in the radar reflectivity field (Fig. 5a) in the field of view of the UMass radar, which covered the region from left of the cloud notch and continued to the right past the cyclonic swirl in cloud base. The hook echo on the right (0.9 km, 2 km) at approximately 1.5-km range corresponds with the cyclonic swirl in cloud base. The hook echo appears to be elliptically shaped, an artifact created because both the direction of the radar scan and the motion of the cyclonic cloud base were to the right of the field of view; the hook echo probably has a more circular shape. The difference in viewing angles of the two hook echoes and the motion of the storm apparently reduced the symmetry of the two hook echoes somewhat.
Features associated with the mirror image hook echo to the left (0.1 km, 2.6 km) at approximately 2-km range are not visible in the video. Although highest radar reflectivities of around 20 dBZe are noted to the left of the hook echoes, attenuation is probably severe enough to reduce the reflectivities significantly at longer ranges.
Each hook echo is approximately 500 m across; the centers of the hook echoes are roughly 1 km apart. At the range of the hook echo on the right (left), the height of the radar beam is approximately 150 m (200 m) AGL, which is below cloud base. The hook echo on the right is associated with cyclonic shear in the Doppler velocity field of approximately 25 m s−1 (500 m)−1, that is, 0.05 s−1; the hook echo on the left is associated with anticyclonic shear of approximately 10 m s−1 (200 m)−1, that is, also around 0.05 s−1 (Fig. 5b). The hook echoes were therefore apparently associated with counterrotating vortices on a scale much smaller than that of their parent storm. The location of an echo-free notch in between the hook echoes and the clockwise–counterclockwise curl of the hook echoes are consistent with the anticyclonic–cyclonic couplet in the wind field implied by the Doppler radial velocity data. If the Doppler velocity shear vorticity associated with each hook were representative of the actual vorticity of an axisymmetric vortex, then the vorticity is twice the Doppler velocity shear vorticity (i.e., 0.1 s−1), which is characteristic of the weakest tornadoes; tornadoes have vorticity on the order of from 10 m s−1 (100 m)−1 to 10 m s−1 (10 m)−1, that is, 0.1–1 s−1. Wicker and Wilhelmson (1995) have simulated tornadolike vortices having vorticity of almost 0.3 s−1 within supercell storms.
Since most of the horizontal shear in the Doppler velocity field is apparent mainly at constant range (Fig. 5b), there is no strong evidence of convergence (Brown and Wood 1991). Either the vortices were not being stretched, or there may have been convergence below 150 and 200 m in each vortex, respectively, which was not detected.
4. Analysis of the parent storm
The storm that spawned the aforementioned vortices formed shortly before 2100 UTC (all times hereafter given in UTC; CDT is 5 h earlier) (Fig. 6) over 100 km southwest of Tulsa, Oklahoma, near a dryline (not shown). It grew rapidly over the next half-hour and by 2200 had split into parts that moved to the left and right of the mean wind (Fig. 7). At the time the storms had split, the OU–UMass storm-intercept crew with the UMass radar was located just ahead of a mesoscale convective system seen in Fig. 6 well to the north-northwest of the splitting storm. A time lag of approximately 1–1.5 h between storm formation and storm split is common in splitting supercells observed by radar (Bluestein and Sohl 1979) and in numerically simulated supercells (Weisman and Klemp 1982).
By 2300 the left-moving member of the splitting storm collided with the mesoscale convective system, a squall line, to the north (Fig. 6). The right-moving member of the splitting storm, however, remained isolated as it approached the area just south of Tulsa. The UMass radar collected the data shown in Fig. 5 at 2350 in the right-moving member, east of Tulsa, almost 2 h after the storm had split. By this time the right-moving member was located well ahead of the dryline (Fig. 8) in a relatively homogeneous air mass. Shortly thereafter, the right-moving member evolved into a line segment (not shown).
Level II data (Crum and Alberty 1993) from the WSR-88D radar at Tulsa (KINX) showed a low-level mesocyclone signature in the storm as it moved by within 30-km range (Fig. 9). The mesocylone signature tracked to the northeast, weakened, and reformed to the right between 2353 and 2359, just minutes after the UMass radar had collected data. At the time the UMass radar collected data, the WSR-88D radar indicated that the mesocyclone signature at 3.3° elevation angle (approximately 1.7 km AGL) was located approximately 8 km north-northwest of the UMass radar, well beyond its range.
Several minutes before the UMass radar detected the counterrotating vortices, the WSR-88D radar did not indicate a mesocyclone signature at the lowest elevation angle (not shown), 0.5°, which corresponds to approximately 250 m AGL at the range of the mesocyclone signature aloft. However, a mesocylone signature did appear at 2.4° elevation angle, which corresponds to approximately 1.3 km AGL (Fig. 10a). The core of the storm, whose reflectivity was in excess of 55 dBZe, was located 5 km to the west of the mesocyclone signature (Fig. 10b). A narrow band of high reflectivity extended to the southeast of the core; a weak-echo notch was sandwiched in between the core to the north and the band to the south. It therefore appears as if the vortices detected by the UMass radar below 200 m AGL were not associated with the mesocyclone signature aloft, to the north (Fig. 9).
The reflectivity field detected by ELDORA looked qualitatively similar to that detected by the WSR-88D. South of a 50 dBZe core at 1.5 km AGL there was a band of higher reflectivity extending to the southeast (Fig. 11a). The ELDORA shows that the echo core was elongated in the northeast–southwest direction. The weak-echo notch shown in the WSR-88D reflectivity analysis shows up clearly in the ELDORA reflectivity analysis.
A pseudo-dual-Doppler analysis of the wind field 5 min after appearance of the vortices in the UMass dataset reveals that the apex of the weak-echo notch was collocated with a kink in a boundary separating two different airflow regimes. Northeast of the kink the storm-relative winds shifted from southeasterly ahead of the boundary to north and northwesterly behind the boundary. The echo core was located just behind the boundary. Southeast of the kink the winds shifted from southeasterly ahead of the boundary to northerly, west of the boundary. Owing to the difference in orientation of the boundaries, the boundary northeast of the kink was characterized mainly by convergence on the order of 10−2 s−1 (Fig. 12), while southeast of the kink the boundary was characterized mainly by vorticity on the order of 10−2 s−1 (Fig. 13). A maximum in cyclonic vorticity was associated with a circulation along the boundary at (32.5 km, 25 km). A broad anticyclonic circulation was apparent west of the boundary.
The circulation, weak-echo notch, and echo core depicted in the ELDORA analysis were extrapolated back in time for a 5-min period using the motion of the storm estimated from the movement of the storm’s echo mass (Fig. 6) as depicted by the WSR-88D radar. (ELDORA data collection did not cover the area of interest in the previous pass to the west northwest; airborne data from the NOAA P-3 were also not available.) It appears that the vortices detected by the UMass radar were associated with the circulation resolved by ELDORA along the south-southeast–north-northwest-oriented boundary southeast of the kink. However, the pseudo-dual-Doppler analysis of the ELDORA data showed a cyclonic circulation only, on the 1-km scale, while the UMass radar resolved an anticyclonic–cyclonic couplet on the 500-m scale.
Since the mesocyclone signature detected by the WSR-88D radar was not evident at 2355, the time of the pseudo-dual-Doppler analysis of the ELDORA data (Fig. 9), it is difficult to determine exactly what it represented. Eight minutes earlier, at 2347, the mesocyclone signature had been located east of the echo core and was probably associated with the wall cloud we had observed earlier. The track of the mesocyclone signature after 2359 suggests that it was not associated with the circulation resolved by the ELDORA data. Instead, it was probably associated with the flow near the kink in the boundary close to the core. The strong inbound velocities seen by the WSR-88D are probably associated with the northeast relative flow near the core and kink (Fig. 11a).
At 3 km AGL the pseudo-dual-Doppler analysis showed a southeasterly jet feeding the radar echo core in excess of 50 dBZe (Fig. 11b) directly above the echo notch seen below at 1.5 km AGL. A cyclone was located just southwest of the jet and an anticyclone was located southwest of the cyclone; the cyclone and anticyclone formed a couplet in vorticity (not shown).
Vertical cross sections of radar reflectivity and three-dimensional wind through the boundaries north and south of the kink exhibit different characteristics. South of the kink there are an erect updraft and radar echo overhang near the cyclonic circulation at 1.5 and 3 km AGL (Fig. 14a). Sinking motion is found below 3.5 km AGL behind the boundary, and rising motion is found above 3.5 km AGL. The sinking motion analyzed behind the boundary is consistent with the sinking motion seen on the video on the western side of the cloud base. On the other hand, north of the kink there is a highly tilted updraft flowing over a wedge of sinking motion at most 2 km deep (Fig. 14b). The airflow looks like that associated with that of a density current (Simpson 1969). The radar echo core is confined to the low altitudes and is collocated with the deepest portion of the wedge of sinking motion.
5. Summary and future work
The parent storm exhibited supercell characteristics: It was long lived, split into right- and left-moving sections, isolated, and had a mesocyclone for periods of time. It formed in an environment of strong vertical shear, which is conducive to supercell formation (Weisman and Klemp 1982). Almost 2 h after the storm had split, 500-m scale, counterrotating vortices and hook echoes were detected by the UMass 3-mm-wavelength Doppler radar system below cloud base, near the surface, in the right-moving storm. The vortices were located along a rear-flank downdraft boundary, south of a kink and an echo notch, near a larger-scale cyclonic vortex seen in a pseudo-dual-Doppler analysis of ELDORA data. The cyclonic vortex was near an echo overhang and erect updraft, and south of a southeasterly front-to-rear “inflow” jet at 3 km. The counterrotating 500-m-scale vortices were observed at a time when a mesocyclone signature associated with the kink in the boundary was undergoing reorganization.
Small-scale vortices along gust front boundaries have been documented (e.g., McCaul et al. 1987) elsewhere. Brandes (1978) suggested that vortices could be created as unstable vortex sheets roll up. However, the sense of the rotation of the vortices studied elsewhere was generally in the same direction. The symmetry of the two vortices discussed here suggests that they may have been formed as a result of local tilting of low-level horizontal vorticity associated with vertical shear, by an updraft whose diameter was less than 1 km. If, on the other hand, the vortices had been produced through a shearing instability (Carbone 1983), we would expect both of the vortices to be cyclonic, owing to the cyclonic vorticity along the leading edge of the rear-flank downdraft.
The flow pattern at 1.5 km AGL bears similarities to that in a classic tornadic supercell at the surface (Lemon and Doswell 1979; cf. Fig. 7). According to the Lemon–Doswell conceptual model, the boundary south of the kink could be interpreted as the portion of the gust front at the leading edge of the rear-flank downdraft; the boundary north of the kink could be interpreted as the portion of the gust front at the leading edge of the forward-flank downdraft. If there were a tornado, it would most likely be located near the kink in the boundary. However, in Fig. 11a we find that a nontornadic cyclonic circulation is actually located south-southeast of the kink; such a location has been previously observed to be a secondary favored region for tornado formation (Davies-Jones 1986).
Wakimoto and Atkins (1996) have documented the formation of a tornado in a low-level shear feature at the leading edge of the rear-flank downdraft in a supercell in the similar position to that of the 500-m-scale vortices we have documented in the 17 May 1995 storm (cf. Fig. 22 in Wakimoto and Atkins and Fig. 11a in this paper). Wilson (1986) has documented, along wind shift boundaries, tornadoes whose parent circulations were narrower than 1 km. We therefore believe that vortices of the type we have documented could sometimes intensify to tornado intensity, even though they failed to do so in this case. The 500-m-scale vortices detected by the UMass radar are not resolved in either the ELDORA pseudo-dual-Doppler wind analyses or the nearby WSR-88D level II Doppler velocity data: We therefore suggest that rapidly scanning ground-based mobile radars, which can be positioned at close range so that very high resolution at low elevation can be obtained and scans can be repeated at short time intervals, be an integral part of any field experiment whose objective is to document from close range and understand tornadogenesis near the ground.
Other questions that need to be addressed in the future include the following. (a) Is there anything special about a supercell gust front in producing small-scale boundary layer vortices, or do most gust fronts produce them? (b) Are small-scale vortices like the ones we documented found in clear air?
This research was funded by NSF Grant ATM-9302379. S. Gaddy was funded by an American Meteorological Society Graduate Fellowship from Loral Defense Systems—East. The ELDORA data were obtained from ATD at NCAR. Wen-Chau Lee and Peter Hildebrand at ATD were instrumental in the planning and development of ELDORA for its use in VORTEX. Roger Wakimoto, Nolan Atkins, and the rest of the Electra crew made data collection possible. Erik Rasmussen, Jerry Straka, and their assistants provided timely VORTEX information in the field. NSSL is acknowledged for its contributions to VORTEX. Peter Lamb and Brian Skinner at OU provided the WSR-88D data. Ken Crawford and his colleagues at OU provided data from the Oklahoma Mesonet. Morris Weisman and his crew from NCAR collected the mobile CLASS data. The OU Motor Pool maintained our chase van. The OU Graduate College funded a portion of our workstation, which we used to process the ELDORA data. Part of this research was done while the first author was a visitor at the MMM Division at NCAR. Morris Weisman provided a useful informal review. Some of the figures were prepared by the NCAR Graphics group.
Bluestein, H. B., and C. J. Sohl, 1979: Some observations of a splitting severe thunderstorm. Mon. Wea. Rev.,107, 861–873.
——, 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, Geophys. Monogr. No. 79, Amer. Geophys. Union, 319–352.
——, 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.
——, W. P. Unruh, D. C. Dowell, T. A. Hutchinson, T. M. Crawford, A. C. Wood, and H. Stein, 1997: Doppler radar analysis of the Northfield, Texas, tornado of 25 May 1994. Mon. Wea. Rev.,125, 212–230.
Brandes, E. A., 1978: Mesocyclone evolution and tornadogenesis: Some observations. Mon. Wea. Rev.,106, 995–1011.
Brock, F. V., K. C. Crawford, R. L. Elliott, G. W. Cuperus, S. J. Stadler, H. L. Johnson, and M. D. Eilts, 1995: The Oklahoma mesonet: A technical overview. J. Atmos. Oceanic Technol.,12, 5–19.
Brown, R. A., and V. T. Wood, 1991: On the interpretation of single-Doppler velocity patterns within severe thunderstorms. Wea. Forecasting,6, 32–48.
Carbone, R. E., 1983: A severe frontal rainband. Part II: Tornado parent vortex circulation. J. Atmos. Sci.,40, 2639–2654.
Cressman, G. P., 1959: An operational objective analysis system. Mon. Wea. Rev.,87, 367–374.
Crum, T. D., and R. L. Alberty, 1993: The WSR-88D and WSR-88D Operational Support Facility. Bull. Amer. Meteor. Soc.,74, 1669–1687.
Davies-Jones, R. P., 1986: Tornado dynamics. Thunderstorm Morphology and Dynamics, 2d ed., E. Kessler, Ed., University of Oklahoma Press, 197–236.
Doviak, R. J., and D. Sirmans, 1973: Doppler radar with polarization diversity. J. Atmos. Sci.,30, 737–738.
Jorgensen, D. P., 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.
Laws, J. O., and D. A. Parsons, 1943: The relation of raindrop size to intensity. Trans. Amer. Geophys. Union,24, 432–460.
Lee, W.-C., P. Dodge, F. Marks, and P. Hildebrand, 1994: Mapping of airborne Doppler radar data. J. Atmos. Oceanic Technol.,11, 572–578.
Lemon, L. R., and C. A. Doswell III, 1979: Severe thunderstorm evolution and mesocyclone structure as related to tornadogenesis. Mon. Wea. Rev.,107, 1184–1197.
Lhermitte, R., 1990: Attenuation and scattering of millimeter wavelength radiation by clouds and precipitation. J. Atmos. Oceanic Technol.,7, 464–479.
McCaul, E. W., Jr., H. B. Bluestein, and R. J. Doviak, 1987: Airborne Doppler lidar observations of convective phenomena in Oklahoma. J. Atmos. Oceanic Technol.,4, 479–497.
Rasmussen, E. N., J. M. Straka, R. Davies-Jones, C. A. Doswell III, 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.
Rust, W. D., R. Davies-Jones, D. W. Burgess, R. A. Maddox, L. C. Showell, T. C. Marshall, and D. K. Lauritsen, 1990: Testing of a mobile version of a Cross-chain Loran Atmospheric Sounding System (M-CLASS). Bull. Amer. Meteor. Soc.,71, 173–180.
Simpson, J. E., 1969: A comparison between laboratory and atmospheric density currents. Quart. J. Roy. Meteor. Soc.,95, 758–765.
Wakimoto, R. M., and N. T. Atkins, 1996: Observations on the origins of rotation: The Newcastle tornado during VORTEX 94. Mon. Wea. Rev.,124, 384–407.
——, W.-C. Lee, H. B. Bluestein, C.-H. Liu, and P. H. Hildebrand, 1996: ELDORA observations during VORTEX 95. Bull. Amer. Meteor. Soc.,77, 1465–1481.
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.
Wicker, L. J., and R. B. Wilhelmson, 1995: Simulation and analysis of tornado development and decay within a three-dimensional supercell thunderstorm. J. Atmos. Sci.,52, 2675–2703.
Wilson, J. W., 1986: Tornadogenesis by nonprecipitation induced wind shear lines. Mon. Wea. Rev.,114, 270–284.
Wurman, J., J. M. Straka, and E. N. Rasmussen, 1996: Fine-scale Doppler radar observations of tornadoes. Science,272, 1774–1777.
Characteristics of the UMass 3-mm-wavelength radar system.
Characteristics of ELDORA operating in convective mode.