1. Introduction
Direct measurements of meteorological variables near tornadoes have been relatively scarce, owing to the rarity of the phenomenon and the spatial and temporal frequency of standard observations. A few fortuitous datasets have been analyzed by Tepper and Eggert (1956) and Fujita (1958). Storm intercept field programs, first organized in the 1970s, have obtained additional observations within supercell storms (e.g., Golden and Morgan 1972; Bluestein 1983; Davies-Jones 1986); however, our collection of in situ measurements within supercells remains miniscule.
A fleet of instrumented automobiles was designed by Straka et al. (1996) for use in the Verification of the Origins of Rotation in Tornadoes Experiment (VORTEX; Rasmussen et al. 1994), conducted on the U.S. Great Plains during the springs of 1994 and 1995. Coined the “mobile mesonet,” this platform collected surface data within severe storms with unprecedented spatial (100–1000 m) and temporal (10–60 s) resolution. The mobile mesonet also has been used in subsequent years following VORTEX, with the spring of 1999 being exceptionally fruitful for operations. This paper summarizes a few noteworthy observations made on 3 May 1999, during a significant outbreak of tornadoes in Oklahoma and Kansas.
During the evening of 3 May, the mobile mesonet intercepted a pair of tornadic supercells in southwestern and central Oklahoma (Fig. 1). The first supercell was intercepted approximately 25 km north of Fort Sill, Oklahoma, shortly after 2200 UTC (this storm will be referred to as storm A to be consistent with National Weather Service surveys of the event). A second supercell was intercepted farther west around 0000 UTC (this storm will be referred to as storm B). Detailed accounts of these intercepts appear in section 3, following a description of the analysis techniques in section 2. A few final comments are provided in section 4.
2. Mobile mesonet data
The mobile mesonet senses temperature, relative humidity, pressure, and wind velocity. Time and position are recorded using a global positioning system receiver. In 1999, data were recorded at 2-s intervals. A complete list of instrument specifications (including response times and errors) is available from Straka et al. (1996).
Vehicle velocities were removed from the wind velocity data when vehicle accelerations were small. In the presence of significant vehicle accelerations (>1 m s–2), accurate wind data could not be obtained. Furthermore, field operations relied on the use of radio communication. On occasion, radio frequency interference caused large errors to arise in the meteorological measurements. Data with these gross errors were removed prior to analysis. In addition, biases in the data were removed by way of intercomparisons between vehicles. The intercomparisons were made in relatively quiescent weather conditions while the vehicles were moving as a caravan (e.g., en route to a storm).
The quality-controlled observations used in the analyses were averaged over 12-s intervals. Data were plotted relative to radar echoes using time-to-space conversion, assuming that the features being analyzed did not change significantly over the time interval during which the measurements were made (the “Taylor hypothesis”). Supercells are not steady. If they were, then tornadogenesis could not occur. However, steadiness assumptions, at least for short time intervals, are virtually unavoidable in observational studies (e.g., multiple Doppler radar analysis). For the research herein, features were assumed to be steady for ±3 min, with respect to an analysis time. This is approximately the length of time that it takes for a Weather Surveillance Radar-1998 Doppler (WSR-88D) to complete a volume scan. There is some confidence that such steadiness assumptions were not too severe, for the analyzed fields tended to be free of noise [inappropriate steadiness assumptions lead to the artificial creation of warm and cold (or moist and dry) pockets in the analysis following a time-to-space conversion].
In addition to the raw thermodynamic data recorded by the mobile mesonet, several derived variables were computed. Virtual potential temperature 
The fluctuations of 
The uncertainty of perturbation variables (denoted by δ) depends on the propagation of instrument errors (e.g., temperature, relative humidity, and pressure errors), uncertainties associated with the estimation of liquid water in the case of 
The mobile mesonet comprised three vehicles on 3 May 1999, in contrast to the much larger assembly used for VORTEX. The goal of the field operations of 1999 was to sample the rear-flank downdraft (RFD) region of supercells (Lemon and Doswell 1979), rather than attempt to sample a broad inflow region, as was the case during VORTEX (Rasmussen et al. 1994). Thus, a smaller array of vehicleborne instruments could accomplish the sampling goal. Furthermore, a smaller fleet had some logistical advantages; for example, coordination among mobile mesonet crews and with the nowcaster was considerably easier.
The mobile mesonet RFD sampling strategy was favored for several reasons. In environments initially absent of vertical vorticity (i.e., vortex lines are quasi-horizontal), a downdraft is necessary for intense vertical vorticity to arise at the ground (Davies-Jones 1982; Davies-Jones and Brooks 1993; Walko 1993). Even as early as 1975, Fujita (1975) hypothesized that the angular momentum transport by a downdraft may be critical to tornadogenesis. Burgess et al. (1977) and Barnes (1978) also made similar speculations. Many studies have found that the air parcels that supply the tornado pass through the RFD. For example, observations by Brandes (1978), Lemon and Doswell (1979), Rasmussen et al. (1982), and Jensen et al. (1983) have shown or implied a near total occlusion of the low-level mesocyclone by the RFD prior to tornadogenesis. Furthermore, Brooks et al. (1993), Wicker and Wilhelmson (1995), and Adlerman et al. (1999) have found that the trajectories entering the near-ground circulations in their numerical simulations passed through the hook echo and RFD. Given the prior emphasis on the RFD in the tornadogenesis process and the apparent consensus that RFD air parcels enter the tornado, it is believed that the thermodynamic characteristics of hook echoes and RFDs naturally assume importance. For this reason, the focus of the next section is on the observations obtained largely within the RFDs of the intercepted tornadic supercells.
3. Observations
a. Storm A
The mobile mesonet approached storm A around 2200 UTC. The supercell was associated with two brief, weak tornadoes at 2151 and 2155 UTC, before an intercept could be engaged. At 2220 UTC, the supercell produced its first significant tornado.2 The tornado persisted for 15 min, although it did not do significant damage along its 10-km track (F1 Fujita-scale rating). The RFD associated with the circulation was sampled by the mobile mesonet west-southwest through east-southeast just prior to tornadogenesis, and during the lifetime of the tornado all three vehicles trailed the tornado to the southwest, within the RFD.
At 2219 UTC, 1 min before tornadogenesis, the mobile mesonet detected relatively small 
The mobile mesonet also sampled a small pressure excess (<1 hPa) in the RFD at 2219 UTC, except within a few hundred meters of the intensifying circulation center, where a small pressure deficit (about −1 hPa) was detected (Fig. 7). A small pressure deficit also was sampled in the inflow east of the gust front. Streamlines revealed difluence at the surface within the RFD, with the strongest difluence (and divergence) apparently situated in the hook echo (Fig. 4). Anticyclonic vertical vorticity also was present within the RFD to the south and east of the incipient tornado.
The mobile mesonet was unable to sample the RFD within 1 km of the tornado at later times; however, data were collected within the RFD approximately 2–5 km southwest of the tornado until its demise. At 2229 UTC (6 min prior to tornado dissipation), only small 
The relatively long-lived tornado dissipated at 2235 UTC, and another tornado was produced by storm A at 2246 UTC. The tornado produced F3 damage and lasted until 2310 UTC, but the tornado could not be closely intercepted for logistical reasons. The road network did not allow sampling within several kilometers of the tornado, and many roads became obstructed by debris. Storm A went on to produce another significant tornado near 2323 UTC, which produced F5 damage in the suburbs of Oklahoma City. This tornado also was not intercepted for logistical reasons (e.g., increasing amounts of traffic near the metropolitan area, debris-filled roadways). Instead, the mobile mesonet abandoned storm A in pursuit of another tornadic supercell to the west (storm B).
b. Storm B
The mobile mesonet arrived at storm B at approximately 0000 UTC (4 May). Storm B produced several brief, weak tornadoes from 2236 to 2324 UTC. Another tornado was reportedly on the ground for 21 min from 2338 to 2359 UTC, but this tornado dissipated before data could be collected. From 0000 to 0100 UTC, numerous additional tornadoes (as many as eight) were reported, most of which were brief and weak (F0–F1 damage ratings), although two tornadoes persisted for longer than 5 min. Data were collected within a few kilometers of the surface circulation centers at several times from 0000 to 0100 UTC.
At 0026 UTC, storm B was not producing a tornado, but brief tornadoes were reported just before and after the analysis time (at 0020 and 0034 UTC). The mobile mesonet obtained data within 0.5–2.0 km of the mesocyclone center at the surface. Within 3 km of the circulation center, the most negatively buoyant air sampled within the hook echo and RFD region was associated with a 
The mobile mesonet made additional penetrations of the hook echo and RFD near 0043 UTC. A tornado, which formed at 0037 UTC, was observed at the analysis time. The tornado dissipated at approximately 0048 UTC. Again, only small 
At 0052 UTC, the RFD and hook echo region of storm B were again sampled relatively well, and another tornado was in progress at the time (the tornado formed at 0047 UTC and dissipated at 0100 UTC). As at earlier times, only small 
Surface pressure gradients were weak at 0052 UTC at distances of more than 500 m from the tornado, similar to observations in storm A (Fig. 15). A small pressure excess (<1 hPa) was detected within the RFD, and a small pressure deficit (also <1 hPa) was measured within the region of cyclonic vertical vorticity associated with the tornado parent circulation. The relatively weak pressure gradients are somewhat consistent with observations made by the mobile mesonet crews of “light winds” within a few kilometers of the tornadoes, both in storms A and B (J. Straka 1999, personal communication).
Storm B produced at least six additional tornadoes, some strong, after dark, which arrived between 0100 and 0130 UTC. Data collection operations by the mobile mesonet were terminated near 0130 UTC. The supercell dissipated in northern Oklahoma around 0400 UTC.
4. Discussion and closing remarks
The observational data presented herein have several limitations. First, road networks do not allow continuous sampling of moving updrafts for periods longer than about 5 min before repositioning of the vehicles requires that they temporarily forfeit data collection in critical regions of the storm; therefore, the time evolution of features is difficult to document. Furthermore, steadiness reluctantly was assumed for durations as long as 6 min (±3 min from the analysis reference times) in constructing the analyses, to maximize the coverage of data (gathered by a finite number of vehicles). Second, thermodynamic fields and their gradients cannot be ascertained above the surface by direct means. At best, only the sign of the gradients can be inferred above the surface, based on assumptions of the lapse rates beneath and at a distance from the storm. Third, time histories of air parcels are important, possibly more than 30 min prior to tornadogenesis. It was not possible to compute trajectories at the surface because of inadequate observation density.
Despite the above impediments, substantial evidence was presented of the following characteristics of the 3 May 1999 tornadic supercells:
RFDs were associated with small
RFDs contained substantial amounts of CAPE and small amounts of CIN,
hook echoes lacked strong baroclinity at the surface, and
pressure fluctuations and gradients were small at distances greater than approximately 250 m from the circulation centers.
The relatively small 
Some evidence also was presented of vertical vorticity couplets straddling the divergence maxima within the RFDs (roughly collocated with the hook echo), as was presented by numerous other studies at low levels (e.g., Ray et al. 1975, 1981; Brandes 1977, 1978, 1981; Fujita and Wakimoto 1982; Wurman et al. 1996). This probably is evidence that RFDs are involved in a downward displacement of vortex lines, perhaps necessarily supplying angular momentum to the tornado, as many others have previously conjectured (section 1).
It is speculated that some of the above intriguing features of the 3 May 1999 tornadic supercells, particularly the relative warmth and moistness of the downdrafts, may be relevant to the problem of tornadogenesis. Does tornadogenesis probability, longevity, and intensity increase as RFD parcels become more buoyant? If buoyant RFDs are propitious for tornadogenesis, are there any large-scale environmental conditions from which we may anticipate warm, moist RFDs? It is generally accepted that only a relatively small percentage of supercells are tornadic; yet, on 3 May 1999, nearly all were tornadic. This fact may suggest that large-scale factors may exist, at least in some cases, from which favorable RFD thermodynamic characteristics can be inferred. Additional cases and implications and, it is hoped, some answers to the above-posed questions will be the subject of a series of companion papers.
Acknowledgments
I am most grateful to all of the volunteers whose countless personal sacrifices made data collection possible. I am indebted to Drs. Jerry Straka and Erik Rasmussen for their support during field operations and for stimulating discussions related to supercells and tornadogenesis during the last several years. Data-quality checks were performed by Mr. Al Pietrycha, and indispensable assistance in obtaining digital elevation model data was provided by Mr. Rob Carver. Dr. Joshua Wurman provided data from the Doppler on Wheels mobile radars, which were used to help to identify the position of the gust front in storm A. Mr. Mark Shafer provided Oklahoma Mesonet data. I also thank Dr. Chuck Doswell and two anonymous reviewers for helping to improve the manuscript. NSF Grant ATM-9617318 partially supported this research.
REFERENCES
Adlerman, E. J., Droegemeier K. K. , and Davies-Jones R. P. , 1999: A numerical simulation of cyclic mesocyclogenesis. J. Atmos. Sci., 56 , 2045–2069.
Barnes, S. L., 1978: Oklahoma thunderstorms on 29–30 April 1970. Part I: Morphology of a tornadic storm. Mon. Wea. Rev., 106 , 673–684.
Bluestein, H. B., 1983: Surface meteorological observations in severe thunderstorms. Part II: Field experiments with TOTO. J. Climate Appl. Meteor., 22 , 919–930.
Brandes, E. A., 1977: Gust front evolution and tornado genesis as viewed by Doppler radar. J. Appl. Meteor., 16 , 333–338.
Brandes, E. A., . 1978: Mesocyclone evolution and tornadogenesis: Some observations. Mon. Wea. Rev., 106 , 995–1011.
Brandes, E. A., . 1981: Finestructure of the Del City-Edmond tornadic mesocirculation. Mon. Wea. Rev., 109 , 635–647.
Brock, F. V., Crawford K. , Elliot R. , Cupernus G. , Stadler S. , Johnson H. , and Eilts M. , 1995: The Oklahoma Mesonet: A technical overview. J. Atmos. Oceanic Technol., 12 , 5–19.
Brooks, H. E., Doswell III C. A. , and Davies-Jones R. P. , 1993: Environmental helicity and the maintenance and evolution of low-level mesocyclones. The Tornado: Its Structure, Dynamics, Prediction, and Hazards, Geophys. Monogr., No. 79, Amer. Geophys. Union, 97–104.
Browning, K. A., and Ludlam F. H. , 1962: Airflow in convective storms. Quart. J. Roy. Meteor. Soc., 88 , 117–135.
Browning, K. A., and Donaldson R. J. , 1963: Airflow and structure of a tornadic storm. J. Atmos. Sci., 20 , 533–545.
Burgess, D. W., Brown R. A. , Lemon L. R. , and Safford C. R. , 1977: Evolution of a tornadic thunderstorm. Preprints, 10th Conf. on Severe Local Storms, Omaha, NE, Amer. Meteor. Soc., 84–89.
Charba, J., and Sasaki Y. , 1971: Structure and movement of the severe thunderstorms of 3 April 1964 as revealed from radar and surface mesonetwork data analysis. J. Meteor. Soc. Japan, 49 , 191–214.
Davies-Jones, R. P., 1982: A new look at the vorticity equation with application to tornadogenesis. Preprints, 12th Conf. on Severe Local Storms, San Antonio, TX, Amer. Meteor. Soc., 249–252.
Davies-Jones, R. P., . 1986: Tornado dynamics. Thunderstorm Morphology and Dynamics, E. Kessler, Ed., 2d ed., University of Oklahoma Press, 197–236.
Davies-Jones, R. P., and Brooks H. E. , 1993: Mesocyclogenesis from a theoretical perspective. The Tornado: Its Structure, Dynamics, Prediction, and Hazards, Geophys. Monogr., No. 79, Amer. Geophys. Union, 105–114.
Fujita, T. T., 1955: Results of detailed synoptic studies of squall lines. Tellus, 4 , 405–436.
Fujita, T. T., . 1958: Tornado cyclone: Bearing system of tornadoes. Proc. Seventh Conf. on Radar Meteorology, Miami Beach, FL, Amer. Meteor. Soc., K31–K38.
Fujita, T. T., . 1975: New evidence from the April 3–4, 1974 tornadoes. Preprints, Ninth Conf. on Severe Local Storms, Norman, OK, Amer. Meteor. Soc., 248–255.
Fujita, T. T., and Wakimoto R. M. , 1982: Anticyclonic tornadoes in 1980 and 1981. Preprints, 12th Conf. on Severe Local Storms, San Antonio, TX, Amer. Meteor. Soc., 213–216.
Golden, J. H., and Morgan B. J. , 1972: The NSSL/Notre Dame tornado intercept program, spring 1972. Bull. Amer. Meteor. Soc., 53 , 1178–1179.
Jensen, B., Marshall T. P. , Mabey M. A. , and Rasmussen E. N. , 1983: Storm scale structure of the Pampa storm. Preprints, 13th Conf. on Severe Local Storms, Tulsa, OK, Amer. Meteor. Soc., 85–88.
Lemon, L. R., and Doswell C. A. , 1979: Severe thunderstorm evolution and mesocyclone structure as related to tornadogenesis. Mon. Wea. Rev., 107 , 1184–1197.
Markowski, P. M., 2000: Surface thermodynamic characteristics of rear-flank downdrafts, with implications for tornado genesis and maintenance. Ph.D. dissertation, University of Oklahoma, 217 pp. [Available from Bell and Howell Information and Learning, Ann Arbor, MI 48106-1346.].
Rasmussen, E. N., Peterson R. E. , Minor J. E. , and Campbell B. D. , 1982: Evolutionary characteristics and photogrammetric determination of windspeeds within the Tulia outbreak tornadoes 28 May 1980. Preprints, 12th Conf. on Severe Local Storms, San Antonio, TX, Amer. Meteor. Soc., 301–304.
Rasmussen, E. N., Straka J. M. , Davies-Jones R. P. , Doswell C. A. , Carr F. H. , Eilts M. D. , and MacGorman D. R. , 1994: Verification of the Origins of Rotation in Tornadoes Experiment: VORTEX. Bull. Amer. Meteor. Soc., 75 , 995–1006.
Ray, P. S., Doviak R. J. , Walker G. B. , Sirmans D. , Carter J. , and Bumgarner B. , 1975: Dual-Doppler observation of a tornadic storm. J. Appl. Meteor., 14 , 1521–1530.
Ray, P. S., Johnson B. C. , Johnson K. W. , Bradberry J. S. , Stephens J. J. , Wagner K. K. , Wilhelmson R. B. , and Klemp J. B. , 1981: The morphology of several tornadic storms on 20 May 1977. J. Atmos. Sci., 38 , 1643–1663.
Rutledge, S. A., and Hobbs P. V. , 1984: The mesoscale and microscale structure and organization of clouds and precipitation in midlatitude cyclones. Part XII: A diagnostic modeling study of precipitation development in narrow cold-frontal rainbands. J. Atmos. Sci., 41 , 2949–2972.
Straka, J. M., Rasmussen E. N. , and Fredrickson S. E. , 1996: A mobile mesonet for finescale meteorological observations. J. Atmos. Oceanic Technol., 13 , 921–936.
Tepper, M., and Eggert W. E. , 1956: Tornado proximity traces. Bull. Amer. Meteor. Soc., 37 , 152–159.
Walko, R. L., 1993: Tornado spin-up beneath a convective cell: Required basic structure of the near-field boundary layer winds. The Tornado: Its Structure, Dynamics, Prediction, and Hazards, Geophys. Monogr., No. 79, Amer. Geophys. Union, 89–95.
Wicker, L. J., and Wilhelmson R. B. , 1995: Simulation and analysis of tornado development and decay within a three-dimensional supercell thunderstorm. J. Atmos. Sci., 52 , 2675–2703.
Wurman, J., Straka J. M. , and Rasmussen E. N. , 1996: Fine-scale Doppler radar observations of tornadic storms. Science, 272 , 1774–1777.
The tracks of storms A and B, with the regions indicated where mobile mesonet data were collected. The radar echoes outline the 30-dBZ base reflectivity values at the 0.5° elevation angle of the Twin Lakes, OK, KTLX WSR-88D; times (UTC) also are included for each radar echo. Tornado paths are gray (storm A) and black (storm B). The surface stations SWO, CSM, OKC, and FSI are Stillwater, Clinton, Oklahoma City, and Fort Sill, respectively
Citation: Weather and Forecasting 17, 3; 10.1175/1520-0434(2002)017<0430:MMOOM>2.0.CO;2
(left) Skew T–logp diagram from Norman, OK, at 0000 UTC 4 May 1999. The hodograph is shown in the inset, with 10 m s–1 speed rings and numerals along the hodograph indicating heights AGL (km). (right) Vertical profile of θe derived from the Norman sounding.
Citation: Weather and Forecasting 17, 3; 10.1175/1520-0434(2002)017<0430:MMOOM>2.0.CO;2
Meso-α-scale hourly subjective analyses of 
Citation: Weather and Forecasting 17, 3; 10.1175/1520-0434(2002)017<0430:MMOOM>2.0.CO;2
(Continued)
Citation: Weather and Forecasting 17, 3; 10.1175/1520-0434(2002)017<0430:MMOOM>2.0.CO;2
Subjective analysis of 
Citation: Weather and Forecasting 17, 3; 10.1175/1520-0434(2002)017<0430:MMOOM>2.0.CO;2
As in Fig. 4 but
Citation: Weather and Forecasting 17, 3; 10.1175/1520-0434(2002)017<0430:MMOOM>2.0.CO;2
As in Fig. 4 but CIN (J kg–1) is analyzed. CAPE (below 500 hPa) and CIN values appear to the left and right, respectively, in each station model
Citation: Weather and Forecasting 17, 3; 10.1175/1520-0434(2002)017<0430:MMOOM>2.0.CO;2
As in Fig. 4 but p′ (hPa) is analyzed (
Citation: Weather and Forecasting 17, 3; 10.1175/1520-0434(2002)017<0430:MMOOM>2.0.CO;2
As in Fig. 4 but at 2229 UTC 3 May 1999. The analysis time is 9 min after tornadogenesis and 6 min prior to tornado dissipation. The T indicates the tornado location
Citation: Weather and Forecasting 17, 3; 10.1175/1520-0434(2002)017<0430:MMOOM>2.0.CO;2
As in Fig. 4 but at 2234 UTC 3 May 1999. The analysis time is 14 min after tornadogenesis and 1 min prior to tornado dissipation
Citation: Weather and Forecasting 17, 3; 10.1175/1520-0434(2002)017<0430:MMOOM>2.0.CO;2
As in Fig. 4 but storm B is analyzed at 0026 UTC 4 May 1999 (
Citation: Weather and Forecasting 17, 3; 10.1175/1520-0434(2002)017<0430:MMOOM>2.0.CO;2
As in Fig. 10 but
Citation: Weather and Forecasting 17, 3; 10.1175/1520-0434(2002)017<0430:MMOOM>2.0.CO;2
As in Fig. 10 but at 0043 UTC 4 May 1999. The analysis time is 6 min after tornadogenesis and 5 min prior to tornado dissipation
Citation: Weather and Forecasting 17, 3; 10.1175/1520-0434(2002)017<0430:MMOOM>2.0.CO;2
As in Fig. 10 but at 0052 UTC 4 May 1999. The analysis time is 5 min after tornadogenesis and 8 min prior to tornado dissipation (this is a different tornado than that which appears in Fig. 12). A few streamlines also have been drawn
Citation: Weather and Forecasting 17, 3; 10.1175/1520-0434(2002)017<0430:MMOOM>2.0.CO;2
As in Fig. 13 but CIN (J kg–1) is analyzed. Station models are as in Fig. 6
Citation: Weather and Forecasting 17, 3; 10.1175/1520-0434(2002)017<0430:MMOOM>2.0.CO;2
As in Fig. 13 but p′ (hPa) is analyzed (
Citation: Weather and Forecasting 17, 3; 10.1175/1520-0434(2002)017<0430:MMOOM>2.0.CO;2
It always is possible to debate the representativeness of a sounding. The 0000 UTC Norman sounding was obtained during the lifetimes of the tornadic supercells and from within 75–100 km of the storms that were intercepted; however, it will be evident later that the environment sampled by the Norman sounding, at least near the surface, differed from the environment to the west in which the supercells were occurring (Fig. 3).
The tornado is referred to as significant because of its longevity.
The adjective “weak” is used to describe the baroclinity because a value of |∇h
