6.1 HISTORY OF RESEARCH

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

1. INTRODUCTION 1953) by radars at the State Water Survey and M. I. T., respectively. Other The focus of this presentation is on early similar observations were made at measurements and observations of A&M by Stuart Bigler. These early tornadoes, laboratory-vortex models, observations showed that hook echoes are severe-storm intercept field programs related to tornadoes. (“storm chasing”), in-situ measurements From the late 1950s onward, and made both intentionally and by chance, especially in the 1960s and 1970s, Ted Doppler radar observations of tornadoes Fujita at the University of made and their parent storms, and theory. painstaking analyses of tornado damage, Numerical modeling will be discussed very tornado photographs, and integrated these briefly, in deference to the later presentation analyses with “mesoanalysis” of surface by Morris Weisman in 9.1. data. Based on photographs of the Fargo, All of the aforementioned topics will be ND tornado of June 1957, he introduced discussed in some sense of, but not exactly, terminology such as “wall ” and “tail in chronological order. I will, however, cloud,” which persists to this day. He attempt to convey a sense of historical introduced what became known as the context. I do not have enough time here to “” in an attempt to estimate wind detail or critique each contribution in this speeds in tornadoes on the basis of the short overview. It is recognized at the outset characteristics of damage inflicted. While not that there is overlap among the sub- calibrated, his technique remains today, disciplines of tornado research and that though in a modified form. After viewing there has been much cross-fertilization of tornado damage from aircraft, he ideas. Branches of research topics are documented cycloidal ground marks and highly intertwined and as such one cannot associated them with multiple-vortex describe progress in a linear fashion. tornadoes. On the basis of aircraft flights For the sake of brevity, I will not spend over a tornadic storm in 1977 he suggested much time, if any, on tornado forecasting, that downdrafts in may play a aspects of convective storms not related to role in tornadogenesis. tornadoes, acoustic-wave research, In the late 1950s, Chester and Harriett electrical research, recent contributions from Newton suggested in a seminal paper based newer instruments such as UAVs, portable upon analysis of a squall line in 1949, that surface probes, etc., data assimilation, non- vertical shear in the environment of a and other funnel convective storm leads to propagation of its /small vortices, damage studies, and updraft as a result of interactions between exotica such as rocket probes . The period the storm updraft and the momentum in its of time covered in my review is from the environment at different heights. Their idea early 1950s until just before VORTEX2. later became the physical basis for our later understanding of supercell behavior, which 2. EARLY MEASUREMENTS AND is intimately related to tornadogenesis. OBSERVATIONS On 4 , Neil Ward, who was participating in the National Severe Storms After the Project in the late Project (NSSP), made the first scientific 1940s, serendipitous radar observations of tornado chase. He traveled with the tornadic storms were first made in 1953 in Oklahoma Highway Patrol and, receiving Illinois (9 April) and Massachusetts (9 June information by phone about echoes seen on a WSR-57 radar, observed a tornado in one knows all the variables in space as a western Oklahoma. He postulated, on the function of time without having to measure basis of his visual observations, that cold them, and one can more easily conduct outflow in the parent storm may have played controlled experiments since one can easily a role in tornadogenesis. Keith Browining vary parameters and boundary conditions. and Ralph Donaldson, in an important 1963 Most recently, Dave and Steve Lewellen at paper, noted a similarity of a radar vault to West Virginia University have been doing that observed in the famous Wokingham, LES model experiments and are able to England hailstorm of 9 July 1959. They reproduce vortex behavior with startling showed an echo hole in the storm, a feature resemblance to reality. we now recognize as being associated with Both laboratory models and numerical tornadoes and/or strong updrafts. simulations have demonstrated the Keith Browning continued his work with importance of the swirl ratio to vortex an analysis of a tornadic supercell in behavior and the phenomenon of vortex Oklahoma on 26 May 1963. Observed by breakdown. Bob Davies-Jones at NSSL has the Lab in Norman and also analyzed lab model behavior and documented visually by Ralph Donaldson, it produced a long-standing and widely was suggested in 1965 that there is an reproduced figure illustrating the effect of ordered sequence of events leading up to swirl ratio on tornado structure and behavior. tornadogenesis. Lewellen and Lewellen have recently proposed that “corner flow collapse” may be 3. LABORATORY MODELS responsible for triggering tornadoes when the rear-flank downdraft (RFD) cuts off flow In order to understand how intense around a developing vortex. vortices such as tornadoes interact with the ground, laboratory models of vortices, 4. THEORY “vortex chambers,” have been built. They have been used to elucidate tornado Theoretical work has progressed in a behavior, but do not address the question of number of areas. Bob Gall at the Univ. of how tornadoes form in storms, because the Arizona in the late 1970s and early to mid parent storms are not represented, save for 1980s did linear stability analyses of vortices an exhaust fan that simulates the updraft. and explained the multiple-vortex Neil Ward at NSSL in 1972 reported on phenomenon in terms of a dynamic his early laboratory studies conducted in the instability. Bob Walko in the early 1990s at mid 1960s. His seminal work was followed OU pioneered the use of a simple, idealized by experiments at the University of model to elucidate physical mechanisms of Oklahoma (OU) with the “Viney tornado vortex formation. Jeff Trapp and Brian simulator” by Martin Jishke and Gene Fiedler at OU in the 1990s continued studies Wilkins and students. John Snow, Chris in the same vein, while more recently Paul Church and graduate students conducted Markowski at Penn State has conducted similar experiments in Purdue beginning in more studies. These “toy” models make use the late 1970s. In the mid 1990s, Snow and of positively and/or negatively buoyant his group pioneered the use of a laser bubbles released in environments of vertical Doppler velocimeter to map the wind field shear. without disturbing the flow and without Lance Leslie and Roger Smith in 1978 having to resort to the use of sometimes introduced the “dynamic pipe effect” (DPE) smelly tracers. (Smoke was used in the as an explanation of how vortices produced simulator at OU.) aloft can propagate downward. Jeff Trapp Numerical simulations of laboratory and Bob Davies-Jones, twenty years later, vortices were first done by Rich Rotunno, at did idealized numerical simulations to NCAR, in a series of studies conducted in identify conditions amenable to the DPE and the late 1970s and early 1980s. An those not amenable to the DPE. advantage of numerical simulations is that Observations of the DPE were made using WSR-88D Doppler radar data by Jeff Trapp tornadogenesis. A statistical study of radar and collaborators. case studies by Aaron Kennedy at OU and Over the years there has been a flurry of collaborators was inconclusive. Bob Davies- studies trying to explain tornado Jones picked up on suggestions by Ted windspeeds. Doug Lilly in an influential Fujita in the early 1970s and by Das in the NCAR publication (an unrefereed report) in early 1980s that rain-filled downdrafts might 1969 explained the hydrostatic initiate tornadogenesis. Davies-Jonesʼ consequences of a vortex in cyclostrophic theory is that descending rain curtains balance. Dergagarbedian and Fendell in transport momentum downward, which can 1970 came up with a theory for the lead to vortex intensification. Most recently, maximum windspeeds in tornadoes based Byko and collaborators used mobile Doppler on hydrostatic balance. Brian Fiedler, as a radar data and idealized modeling to post doc at NCAR in the mid 1980s, worked investigate the causes and consequences of with Rich Rotunno to show how the the DRCs. “thermodynamic speed limit” may be exceeded when there is a supercritical end- 5. EARLY SEVERE-STORM INTERCEPT wall vortex. Bob Walko in 1988 investigated ACTIVITIES the causes of subsidence inside tornadoes. It had been postulated that subsidence The reader is referred to Bluestein (1999) inside tornadoes could be responsible for for a summary of severe-storm intercept warming and a hydrostatic warm core. field programs prior to the late 1990s. In Doug Lilly at OU in the mid 1980s 1963 Frank Ludlam, a prominent British introduced the idea that helicity might act to convection researcher at Imperial College in stabilize storms and promote longevity. He London, stated in an AMS Monograph that emphasized the importance of Beltrami flow. “…the tornado, which energetically is only a Bob Davies-Jones at NSSL, on the other detail in the severe storm. However, its hand, emphasized the importance of the importance as a hazard and the interest of related streamwise vorticity in forming the problems which it poses makes it mesocyclones. He subsequently published a desirable to indicate its probable place in the series of papers investigating the baroclinic cumulonimbus model.” Storm chasers, generation of vortices and the propagation beginning in the early 1970s set out to do of convective storms. Morris Weisman and just that. By chasing storms, one could Rich Rotunno at NCAR questioned how dramatically increase the number of case useful helicity is in describing vortex studies possible when remaining at a fixed formation. They and Bob Davies-Jones site over that by waiting for storms to come engaged in a spirited conversation on the to the site. role of curvature in environmental Joe Golden at NSSL, who had done hodographs on convective-storm behavior. pioneering work on observations of At issue was that storm-relative helicity waterspouts in South and the Keys, depends on storm motion, which in turn did similar studies at NSSL, most depends on the mean wind and on significantly for the Union City tornado on 24 propagation: it is not Galilean invariant. May 1973. He and Dan Purcell produced a The role of downdrafts in tornadogenesis composite figure illustrating the relationship was probably first proposed by Ted Fujita in between the tornado and storm cloud his “twisting downdraft” schematic and features, precipitation, and airflow. On the hinted at in his analyses of the tops of basis of many observations, Al Moller at OU . In the past ten years, Paul and subsequently at the NWS in Ft. Worth Markowski and collaborators have published a schematic of the visual features investigated the role of downdrafts in an in a tornadic supercell in 1978, the basic idealized model and Eric Rasmussen and aspects of which are still being reproduced collaborators have suggested that today. Les Lemon at NSSL and Chuck descending reflectivity cores (DRCs) in Doswell summarized early Doppler radar supercells might play a role in observations and storm-intercept observations in the late 1970s and produced made at NSSL, under the leadership of Ed a very influential schematic of the surface Kessler, its first director, to probe severe features in a tornadic supercell including the convective storms at S-band. The tornado rear-flank and forward-flank downdrafts, and vortex signature (TVS) was discovered the notion of a divided mesocyclone during the subsequent analysis of data from structure. Their conceptual model is still the Union City, OK tornadic supercell of 24 being discussed and slight modifications May 1973 and published by Don Burgess have been suggested recently. and colleagues in 1978. This study marked The first visual tornado data to be the turning point in the direction of analyzed quantitatively were movies subsequent Doppler radar studies of showing debris flying through the air. tornadoes and their parent storms. Les Hoecker, in 1960 published a seminal Lemon and collaborators demonstrated that photogrammetric analysis of airborne debris for the Union City storm there was a in a tornado that hit Dallas, TX in 1957. This temporal relationship between the collapse pioneering study was based on a of the BWER and tornadogenesis, which is serendipitously obtained movie. Subsequent suggestive of a possible dynamic similar studies were conducted by Joe relationship between collapse of the updraft Golden and Dan Purcell for movies of the and subsequent tornadogenesis. Union City tornado in 1973 and by Erik At NSSL a second S-band Doppler radar Rasmussen and collaborators for the Tulia, was installed northwest of the Norman TX tornado in 1980, both during deliberate Doppler radar to form a dual-Doppler storm-intercept missions. The results of network, discussed by Rodger Brown in a these studies showed that flow in tornadoes 1975 publication. Under the leadership of resembled that of a combined Rankine Peter Ray at NSSL, there were a number of vortex. Efforts to obtain movies of tornadoes important studies conducted in which showing flying debris continued through the tornadic supercells passed through the early 1980s. network. Peter Ray and colleagues first demonstrated that good analyses were 6. DOPPLER-RADAR STUDIES possible for data collected for a storm on 20 April 1974. Significant case studies for J. Q. Brantley and Daniel Barczys of the tornadic supercells on 6 June 1974, 8 June Cornell Aeronautical Laboratory first 1974, 22 May 1977 (the much studied and suggested in 1957 that Doppler radar could modeled Del City storm), and 17 May 1981 be used to detect tornadoes. Smith and ensued. Ed Brandes and collaborators Holmes first used a continuous wave (CW), published extensively on dual-Doppler X-band, Doppler radar acquired by the analyses, beginning in 1977. Gerry Weather Bureau from the Navy and modified Heymsfield at OU published an early study appropriately, to obtain Doppler wind in 1978 for the 8 June 1974 storm. Storm spectra in a tornado that hit El Dorado, KS structure during tornadogenesis and vorticity on 10 June 1958. Roger Lhermitte made production were described from the dual- many contributions to radar meteorology, Doppler analyses. Carl Hane at NSSL, in a especially during the 1960s when he 1981 publication, demonstrated that suggested using pulsed Doppler radar for thermodynamic variables can be “retrieved” studies of convective storms. Ralph from the three-dimensional wind field Donaldson on 9 August 1968 used a pulsed, synthesized from dual-Doppler analyses. C-band, Doppler radar from the Air Force While collection of dual-Doppler data for Cambridge Research Laboratories to detect the analysis of the storm-scale wind field in cyclonic shear in a supercell over tornadic supercells was a major objective, Marblehead, MA. His innovative display was Doppler spectra in tornadoes were collected called the “plan shear indicator” (PSI). by Dusan Zrnic and collaborators at NSSL Since severe convective storms contain for a number of cases. This important work heavy precipitation that heavily attenuates produced estimates of the maximum signals at X-band and C-band, efforts were windspeed in tornadoes that did not depend on photogrammetric analysis or in situ Subsequent ultra-high resolution model measurements. simulations at the Univ. of Illinois at NCSA and by Ming Xue and collaborators at OU 7. NON-HYDROSTATIC CLOUD produced realistic-looking simulations of MODELING AND RELATED tornado-like vortices down to scales as low THEORETICAL STUDIES as 12 m. As already noted, Lewellen and collaborators produced simulations as fine At the same time Doppler radars were as 2.5 m in the horizontal and 1.5 m in the being used for the first time to probe the vertical, but not for the whole parent storm. internal structure of tornadic supercells, 3-D non-hydrostatic models were being 8. MEASUREMENTS MADE BY developed and used to simulate them. INSTRUMENTS DURING SEVERE-STORM Based on early work by Yoshi Ogura at the MISSIONS Univ. of Illinois at Champaign-Urbana and Norm Phillips at MIT that was published in The first attempts to make in situ 1962, a model was devised that did not measurements in severe storms were in include the complicating effect of sound 1980, when Al Bedard and Carl Ramzy at waves. Bob Schlesinger at the Univ. of the Wave Propagation Lab in Boulder Wisconsin at Madison was the first to exploit collaborated with the author to design and their work and constructed a working 3-D build TOTO (Totable Tornado Observatory), anelastic model in the mid to late 1970s. At a 400 lb, instrumented package intended to the same time, however, Joe Klemp at make measurements of wind and NCAR and Bob Wilhelmson at the Univ. of thermodynamic variables when left in a Illinois at Champaign-Urbana constructed a tornadoʼs path. The author and his graduate working 3-D model that included sound students used it in 1981 – 1983, and waves (i.e., was fully compressible). This subsequently Lou Wicker at NSSL used it, model became known as the Klemp and with some success, but no great Wilhelmson model and was soon exploited observational insights followed since direct for many studies of supercell behavior. hits were rare. The seminal and most influential work Fred Brock at OU and collaborators was done at NCAR by Morris Weisman and developed the “Turtle,” in the late 1980s to Rich Rotunno. In a series of papers make in situ measurement in tornadoes, but published in the early and mid 1980s, the in a much smaller and lighter package which importance of vertical shear and CAPE were could be deployed en mass, in order to demonstrated, the consequences of increase the likelihood of a direct hit. The dynamically induced vertical perturbation Turtle has evolved into the HITPR pressure gradients on storm propagation (Hardened In-situ Tornado Pressure were explained, and in 1983 a seminal Recorder) by and paper by Rich Rotunno and Joe Klemp was collaborators. On 24 June 2003 in published in which the origin of low-level Manchester, SD, an HITPR recorded a rotation in supercells was explored and pressure fall of ~ 100 hPa in a tornado. explained using the Klemp and Wilhelmson While small deployable packages have model and a one-way nested grid. In the mid been used to make limited measurements 1990s, Lou Wicker and Bob Wilhelmson at directly in tornadoes, mobile the Univ. of Illinois expanded and improved (networks of instrumented vehicles), on the earlier work by Rotunno and Klemp originally developed for VORTEX in 1994 and demonstrated the importance of the and 1995, were first used by Jerry Straka at upward-directed perturbation pressure OU and collaborators to make more detailed gradient underneath the mesocyclone in measurements in selected regions of amplifying vorticity. Louie Grasso and Bill supercells. Two main scientific questions Cotton at CSU, using the RAMMS model posed for mobile deployments and three nested grids, explored similar have been as follows: Are there any issues. anomaly differences between those in the rear-flank downdraft of tornadic supercell tornado. Ernie Agee and a supercells and those in non-tornadic or collaborator at Purdue recently have weakly tornadic supercells? How strong is proposed taxonomy of tornado types. One of the baroclinic zone, if any, along the edge of the most significant types other than the the forward-flank downdraft? Paul most commonly documented supercell Markowski and collaborators have reported tornado, is the non-mesocyclone tornado: on results from VORTEX deployments and some of them the author called “landspouts,” deployments during the 3 May 1999 tornado owing to their similarity to Florida outbreak in Oklahoma. Gryzch et al. have waterspouts. Other non-mesocyclone recently reported on results in supercells in tornadoes that occur along the leading edge the Northern High Plains using a different of gust fronts are called “gustnadoes,” a mobile mesonet. term first used by storm chasers. During the 2004 storm season, we used a Many studies were conducted in the mid commercially available digital infrared and late 1980s in eastern Colorado, to photograph tornadoes and wall especially during CINDE. Jim Wilson at clouds with the intent of estimating NCAR, Roger Wakimoto at UCLA, and other temperature gradients at cloud base. Robin colleagues at NCAR (e.g., Rita Roberts), Tanamachi and collaborators reported in NOAA (Ed Szoke), and U. Wyoming (Brooks 2006 that intervening precipitation interfered Martner) published a series of papers on with efforts to make meaningful observational studies using fixed-site measurements, though it was found that the Doppler radars; most of the studies were lapse rate along a tornado condensation based on single Doppler radar, while one on funnel was approximately moist-adiabatic, 15 June 1988 in the Denver area was based as expected. on dual-Doppler measurements. Roger In situ soundings in and near tornadic Wakimoto did a detailed study combining supercells were first attempted using a photographs with Doppler radar data. It radiosonde package in 1984 by the author should be noted that Fred Bates from St. and his collaborators. The first attempt failed Louis University in the early and mid-1960s as an underinflated balloon skimmed along discussed what may have been similar the ground upstream from a tornado on 26 tornadoes observed from aircraft. Non- April. For subsequent successful ascents, mesocyclone tornadoes were successfully an optical theodolite was used to determine simulated and studied by Bruce Lee and balloon location as a function of time so Bob Wilhelmson at the Univ. of Illinois. It is winds could be computed. Several case now understood that non-mesocyclone studies based on successful launches in the tornadoes develop from vorticity along pre- mid and late 1980s, including a direct existing boundaries. release into the updraft of a tornadic Tornadoes and misocyclones have also supercell in the Texas Panhandle, were been reported in quasi-linear convective reported in the literature. In the late 1980s, systems (QLCSs). Interestingly, it was Dave Rust and colleagues at NSSL began to thought in the 1940s that squall lines were release CLASS sondes, which were responsible for tornadoes, but later studies developed at NCAR. These portable pointed to the importance of isolated cells or radiosondes made use of LORAN navigation cells embedded in lines prior to their signals to determine their location as a evolution into squall lines. Weʼve come full function of time. More recently, NSSL has circle: Tornadoes were documented by Rit used GPS sondes developed at NCAR to Carbone in a study of a California rainband obtain soundings in and near severe in the early 1980s and by Greg Forbes and convective storms. Roger Wakimoto, then at the Univ. of Chicago, in a study of a bow echo in Illinois. 9. TORNADO CLASSIFICATION Jeff Trapp at Purdue and Morris Weisman demonstrated in the early 2000s using Storm chasers have identified a number numerical simulation experiments how of other types of tornadoes, in addition to the strong vortices could be produced in QLCSs. The former published a climatology of QLCS up to the first year of VORTEX to acquire tornadoes in 2005. close-range Doppler wind spectra in It has been recognized for many years tornadoes. In a series of papers in the early that tornadoes can also appear in tropical - mid 1990s, we verified F-5 wind speeds in cyclones, especially when they make a tornado near Red Rock, OK on 26 April landfall. Novlan and Gray in 1974 produced 1991 and determined wind spectra as a a seminal climatology of these tornadoes. function of range with 78 m range resolution, Bill McCaul at OU studied tornadoes in near Northfield, TX on 25 May 1994. This Hurricane Danny in 1985 and in 1991 radar, however, had a relatively broad published a study of the composite beamwidth of 50 and was steerable only by environment of hurricane tornadoes. More hand, and not automatically scanned. recently, Spratt et al. in 1997 discussed While Bob Crane at OU suggested WSR-88D observations of mounting an old, Ka-band radar on a truck tornadoes in Florida and Baker and and having automatic scanning, I chose to colleagues studied tornadoes in Hurricane use a scanning W-band radar from the Ivan over the Gulf in 2004. It appears that Microwave Remote Sensing Laboratory at the parent storms of many tornadoes both the Univ. of Massachusetts in Amherst, over the ocean and after landfall are shallow under the supervision of Bob McIntosh, supercells, while others are not associated because a very narrow beam antenna could with mesocyclones. easily be mounted on a small truck. The first Perhaps similar to shallow supercells in system was developed by Andy Pazmany tropical cyclones are “low-top” or “mini- and first used in 1993 with a 0.60 (half- supercells,” sometimes observed near power) beam. We collected fine-scale data upper-level cyclones, where the tropopause along gust fronts and wall clouds, but not in is low, but vertical shear is strong. Cooley in a tornado until VORTEX during year 1 on 25 1978 described “cold-air funnels” that also May 1994. Alas, while scanning the nearby occur near upper-level cyclones. tornado the system failed owing to an Tornadoes have been classified not only incorrectly made electrical connection. It according to how and where they form, but was not until the 1999 storm season that an by their sense of rotation. It was recognized updated system with a larger antenna long ago that anticyclonic rotation in having a 0.180 beam was used and tornadoes is rare. Based on visual tornadoes on 3 May near Verden, OK, 15 documentation of one in Iowa on 13 June May near Stockton, KS, and 5 June near 1976 by John Brown and Kevin Knupp, one Bassett, NE, at close range were probed, at near Grand Island, NE on 3 June 1980 by one elevation angle and data successfully Ted Fujita, one near Geary, OK on 29 May recorded. Significant datasets were also 2004 and one near El Reno, OK on 24 April collected on 5 May 2002 near Happy, TX 2006 by the author and his students, it is and 12 May 2004 near Attica, KS, when now recognized that when they are high-resolution, vertical cross sections observed they often occur in conjunction through and surrounding the weak-echo hole with another nearby cyclonic tornado. were obtained. In addition, measurements showing the vertical variation of Doppler 10. DOPPLER RADARS USED DURING velocity near the ground were made near SEVERE-STORM INTERCEPT the tornado core. Wen-Chau Leeʼs GBVTD OPERATIONS algorithm was implemented. In addition, horizontal-vortex shear signatures were In 1987 Wes Unruh at the Los Alamos found along the edge of the tornado. National Laboratory and I repeated the The first proposal to use airborne Doppler Smith and Holmes 1958 experiment, this radars to map the wind field in convective time using a battery-powered, portable, storms was made by Roger Lhermitte in CW/FM-CW, X-band, Doppler adapted also 1971. He proposed using two aircraft, each for applications not originally intended for having its own radar, flying at perpendicular the original instrument. We used this radar flight paths to collect dual-Doppler data. Airborne, X-band Doppler radars were used his collaborators, and Erik Rasmussen and by NOAA beginning in the 1980s to map the Jerry Straka. winds in hurricanes. The first supercell During year 2 of VORTEX, Josh Wurman probed by one of the NOAA P-3 airborne at OU and collaborators at OU, NSSL, and radars was on 27 May 1985, near Oklahoma NCAR, developed the first truck-mounted X- City, during PRE-STORM, through the band Doppler radar, the “Doppler on collaboration of Dave Jorgensen of NOAA Wheels” (DOW) and successfully collected and Peter Ray of NSSL. For this case, dual- data in a tornado in the Garden City storm Doppler analyses were synthesized using on 17 May 1995. Subsequently data were data from the airborne radar and data collected in tornadoes on 2 June. Since then separately from each of the two NSSL data collection efforts using the DOW and ground-based, fixed-site, S-band Doppler future generations of DOWS have been very radars. The radar-equipped aircraft flew by successful. supercells again, under the leadership of The first mobile, dual-Doppler dataset for Dave Jorgensen from NOAA, in the spring of a tornadic supercell was collected by two 1991, this time using “FAST” (fore-aft DOWs in eastern Oklahoma, near Kiefer scanning technique), for which only one and Glenpool, on 27 May 1997, but at low aircraft was needed to collect “pseudo”-dual- elevation angle only. Volumetric mobile dual- Doppler data. The first chance to collect Doppler data were first collected data in a tornadic supercell came during successfully near Bridgeport, NE on 20 May COPS-91, on 26 April, but the aircraft could 1998. Since then, Mike Biggerstaff at OU not fly owing to a mechanical problem. Later has used two mobile C-band radars, the in the season, successful data were SMART-Rs, to collect mobile dual-Doppler collected in supercells for the first time. data on a tornadic supercell near Geary, OK David Dowell at OU and collaborators on 29 May 2004, an effort which was published the first analyses of these reported by MacGorman et al. in 2008. supercells in 1997. A significant single-Doppler dataset was During VORTEX, late in the season of collected by a DOW in Spencer, SD on 30 1994 and on many occasions in 1995, May 1998; the tornado damage path was excellent datasets were collected. In the correlated with the Doppler wind data by latter year the NOAA P-3 was joined by Josh Wurman and his student, Curtis NCARʼs ELDORA, which made use of Alexander. Data from the Spencer storm frequency hopping to increase the number exhibited multiple vortices, as did DOW data of independent samples and reduce the time from the Mulhall, OK tornado on 3 May for getting Doppler velocity measurements 1999. The highest wind speeds in a tornado having acceptably small errors as the ever recorded (135 m s-1) were made by a aircraft flew rapidly by the targeted storm. DOW in Bridgecreek, OK on 3 May 1999. The design of ELDORA was a collaboration Double gust front structure was documented between Peter Hildebrand at NCAR and near Crowell, TX on 30 April 2000. Wen- colleagues and collaborators from . Chau Lee and Josh Wurman deduced 3-D The airborne field experiment using tornado structure in the Mulhall tornado ELDORA was led by Roger Wakimoto. In using the formerʼs GBVTD analysis 1994 the Newcastle, TX tornadic storm was technique. To date, successive generations probed on 29 May, and in 1995 the Garden of DOWs have been used to probe City, KS tornadic storm was documented on hundreds of tornadoes and have facilitated a 17 May, the Dimmitt and Friona, TX tornadic climatology of various tornado parameters. storms were probed on 2 June, and the A poor-personʼs version of the DOW was McLean, TX cyclic tornadic storm was built at the Univ. of Mass. at Amherst and probed on 8 June. A number of significant first used in 2001 for surveillance of papers by Roger Wakimoto and his students reflectivity only. The radar system was were published, along with others by David based on a commercially available marine Dowell and me, Conrad Ziegler at NSSL and radar. In 2002, both Doppler and polarimetric capabilities became available. During IHOP, in 2002, another DOW having planned observations and measurements, polarimetric capability, which had been built numerical studies with idealized models, by Josh Wurman for the government of laboratory-vortex experiments, and Greece, but not used to detect tornadoes, numerical simulations. Observations have was used in the field. The UMass X-band improved in step with advances in radar became known as the UMass X-Pol to technology, most prominently with distinguish itself from the Greek-govʼt radar. increasingly more sophisticated radar Early work with a fixed-site, S-band, systems. Numerical simulations have polarimetric Doppler radar at NSSL (KOUN) improved with increased computer power, by Alexander Ryzhkov, and collaborators speed, and storage capabilities. demonstrated that a tornado debris Soon we will be able to analyze signature could be detected (most clearly convective storms simulated with ultra-fine evident as a region of low ρhv). The first case spatial resolution so that the tornado and its reported was in a tornadic storm near parent storm are both resolved adequately Oklahoma City on 3 May 1999. (on scales < 10 m). When we finally are able Since IHOP, the most significant datasets to do controlled experiments and resolve all collected by the UMass X-Pol were on 12 features adequately, then what? Will the May 2004 near Attica, KS, when a tornado simulations be too complex to advance our debris signature was clearly detected at understanding easily? close range along with visual Soon we will have observations that documentation, and on 7 May 2007, when document tornadoes in the act of forming the formation of the Greensburg, KS along with all that goes on in the parent tornadic supercell was documented. storm, using rapid-scan (via mechanically Because tornadoes evolve on very fast scanning, phased-array, or imaging time scales (~ 10 s or less), rapid-scan techniques), and fine-scale, polarimetric radars have been developed. Josh Wurman Doppler radar measurements. Will these and colleagues at NCAR developed the observations be adequate to advance our rapid-DOW, which first scanned a tornado in physical understanding, or will complexity 9 June 2005 in Kansas. The rapid-DOW is again hinder us? We cannot do controlled an X-band, Doppler radar that scans experiments in the atmosphere. mechanically in azimuth, but in electronically Finally, can we address what is cause in elevation at six different angles, by and what is effect when processes are changing frequency. detected nearly simultaneously? This and The MWR-05XP is a mobile, X-band, the aforementioned questions are our phased-array, Doppler radar that challenges for the future. ProSensing, Inc. in Amherst, MA adapted for meteorological use from a military phased- 12. POSTSCRIPT array radar acquired by the Naval Postgraduate School in Monterey, CA. Both This manuscript is a draft based on the the rapid DOW and the MWR-05XP can actual presentation given at the conference. scan sector volumes of storms in 10 s or I did not have time to include in this less; the latter is faster and can scan more manuscript the dozens of figures and the elevation angles nearly simulatneously, but hundreds of references used. As a draft, the former has finer azimuthal resolution. there are probably some factual errors and The MWR-05XP first successfully scanned a some important events or studies were tornado on 23 May 2008 in Kansas. inadvertently not mentioned. I would appreciate receiving corrections in fact and suggestions for other significant studies not 11. CURRENT AND FUTURE TORNADO cited. RESEARCH 13. ACKNOWLEDGMENTS The history of tornado research consists of a series of both serendipitous and NSF grant AGS-0934307 supported the author at the time of this review.

14. SELECTED REVIEW REFERENCES

Severe Local Storms, AMS Monogr. 5 (27), 1963, D. Atlas, ed.

Thunderstorm Morphology and Dynamics, 1986, E. Kessler, ed., Univ. of Oklahoma Press, Ch. 10.

Mesoscale Meteorology and Forecasting, 1986, P. Ray, ed., AMS, Ch. 18.

The Tornado: Its Structure, Dynamics, Prediction, and Hazards. AGU Geophy. Monogr. 79, 1990, C. Church et al., eds.

Radar in Meteorology: Battan Memorial and 40th Anniversary Radar Meteorology Conf., AMS, 1990, D. Atlas, ed., Ch. 14, 15, 16, 24.

Bluestein, H. B., 1999: A history of storm- intercept field programs. Wea. Forecasting, 14, 558 – 577.

Severe Convective Storms, AMS Monogr. 28 (50), 2001, C. Doswell, ed., Ch. 1, 4, 5.

Radar and Atmospheric : A Collection of Essays in Honor of David Atlas, AMS Monogr. 30 (52), 2003, Ch. 5.