Adapted from Preprints, 20th AMS Conference on Severe Local Storms, Orlando, Sep 2000


Roger Edwards and Richard L. Thompson

Storm Prediction Center

Norman, OK

James G. LaDue

NEXRAD Operational Support Facility

Norman, OK


On 3 May 1999, the costliest tornado in U.S. history struck portions of central Oklahoma, including the Oklahoma City metropolitan area, and produced damage rated up to F5 on the Fujita scale (Fujita 1981). In addition to this violent tornado, the parent supercell (hereafter, Storm A, from NCDC 1999) spawned 13 other tornadoes, including two satellite tornadoes.

Storm A initiated in the moist sector, between a dryline and a surface confluence line. The synoptic and mesoscale setting featured increasing vertical shear with time, a weak capping inversion, large convective available potential energy (CAPE), and an absence of significant low-level convergence (Thompson and Edwards 2000). In this environment, supercells formed on small, weak and/or subtle features, some of which have not yet been identified. Storm A appears to have formed along a pronounced radar reflectivity feature, confined to the convective boundary layer (CBL), and exhibiting many characteristics of a horizontal convective roll (HCR, Wilson et al. (1994); Weckwerth et al. (1997)). [Animated imagery of this feature is available online].


Base reflectivity and velocity imagery from the Frederick, Oklahoma, WSR-88D (FDR) were examined around the time of Storm A’s initiation, both in the clear-air reflectivity mode and in standard precipitation mode.

Figure 1. “Clear air mode” radar reflectivity image (dBZ) at the 1.5o elevation angle, from Frederick, Oklahoma, at 2002 UTC, 3 May 1999. Oklahoma Mesonet observations from 2000 UTC are overlaid. The possible HCR is the enhanced reflectivity band extending south to north through eastern Tillman and western Comanche Counties.

Figure 2. Filtered “clear air mode” radar reflectivity data (dBZ) at the 1.5o elevation angle, from Frederick, Oklahoma at 2012 UTC, 3 May 1999. The signature of the possible HCR east of the radar site is annotated. Reflectivies < 0 dBZ have been removed for clarity.

Figure 3. “Precipitation mode” radar reflectivity data (dBZ) at the 1.5o elevation angle, from Frederick, Oklahoma at 2012 UTC, 3 May 1999. The signature of the possible HCR, the genesis location of Storm A, and a separate, failed cumulonimbus are each annotated.

Beginning at approximately 1830 UTC, a meridionally oriented and occasionally wavy band of enhanced reflectivity appeared in 1.5 o elevation scans over eastern Tillman County, east of the radar antenna. This feature, possibly the updraft portion of a large HCR, became more pronounced during the next 1-2 hours with increasing reflectivities (e.g., Fig.1). When clear-air mode reflectivities not characteristic of the band were filtered away (Fig. 2), the possible large HCR was even more prominent, and was superimposed on the eastern portion of a field of much smaller HCRs of low wavelength. The radar switched from clear-air to precipitation mode at 2022 UTC, in response to unrelated convective development over northwestern Texas. The apparent HCR, however, was still evident in the 5-10 dBz reflectivities (Fig. 3). The feature was weakly convergent in .5 o base velocity imagery (not shown), and was occasionally apparent in .5 o base reflectivity.

Figure 4. Norman, Oklahoma sounding from 18 UTC, 3 May 1999, modified for surface thermodynamic conditions at the mesonet site nearest the possible HCR in Tillman County. Various derived parameters are given in the table beneath the skew-T diagram, including convective available potential energy (CAPE), convective inhibition (CINH), lifted indices (LIs), lifted condensation level (LCL), equilibrium level (EL), and maximum parcel level (MPL).

The reflectivity band appeared to be confined to the CBL. It was not apparent in FDR imagery above approximately 1370 m (4500 ft) AGL. This height, corresponded well with the top of the CBL as determined from modified soundings (e.g., Fig. 4). The possible HCR was not observed by radars farther away at Oklahoma City and Fort Worth, whose beams scanned above the CBL. Also, the HCR’s reflectivity pattern disappeared farther north and south as the .5o and 1.5o elevation FDR beams scanned above the CBL.

Although the reflectivity band was quasi-stationary throughout its appearance in FDR imagery, small waves formed and moved northward along its axis. As a few of these waves forced the reflectivity axis across the Oklahoma Mesonet (Brock et al., 1995) site in eastern Tillman County, the wind direction veered from south-southeast to south-southwest with little speed change, before backing again behind the wave crests.

Figure 5. "Clear-air mode" base reflectivity from Twin Lakes, Oklahoma at 2027 UTC, 3 May 99. The first echo associated with Storm A is annotated, along the position of the possible HCR.

The first evidence of reflectivity associated with Storm A in FDR imagery appeared 5-10 km east of the possible HCR at 2030 UTC . However, .5o base reflectivity data from Twin Lakes (TLX), near Oklahoma City, indicate there was substantial precipitation already present at higher altitudes, and several minutes earlier (Fig. 5). There is also a 1 km2 echo of reflectivity in the same location at 2026 UTC, higher in the troposphere, as seen from Ft. Worth (not shown).


Based on the <5 km proximity between first observed echo from TLX and the location of the reflectivity band as sensed from FDR, we conclude that the convective towers which evolved into Storm A initiated directly from the updraft associated with the possible HCR. The northeastward displacement of the elevated reflectivity echoes from the HCR can be explained by the likelihood of several minutes’ lag time between convective initiation and development of sufficient precipitation particles to be detected by these radars.

The possible HCR first became apparent on Frederick reflectivity imagery around 1830 UTC. This occurred as surface insolation – somewhat muted by a canopy of cirrus clouds evident in visible satellite imagery (not shown) – warmed the boundary layer. Storm A began between 2015 and 2030 UTC as the leading edge of a hole in the cirrus canopy moved northeastward over the possible HCR. Temperatures at the eastern Tillman County mesonet site warmed by 1o F as the cirrus hole moved overhead, then cooled again by 1o F as the high clouds returned. Given the virtual absence of a capping inversion as indicated by modified soundings (e.g., Fig. 4), the subtle warming may have been sufficient to strengthen the updraft portion of an HCR enough for surface parcels to reach their respective levels of free convection.

Figure 6. Motion of deep cumuli, relative to the quasi-stationary band of enhanced reflectivity near initiation of Storm A, is computed based on minimum and maximum estimates of the width of the feature. Adapted from Wilson and Megenhardt (1997).

Depending on the reflectivity thresholds used, the apparent large HCR measured 5 to10 km wide during convective initiation. Using motions of deep cumuli in the boundary layer east of the apparent HCR, as measured from Frederick reflectivity loops, a buoyant parcel had between 20 and 45 minutes residence time in the feature (Fig. 6). Although the residence time findings of Wilson and Megenhardt (1997) were obtained using warm-season Florida boundaries, it is interesting to note that the feature-relative motions (Ub) computed for this feature fit well within their range of Ub conducive to initiation of sustained deep convection. This indicates two hypotheses:

    1) The lift associated with the apparent HCR and subtle warming of the boundary layer initiated Storm A.

    2) Their findings from weakly sheared, warm season cases in Florida can apply to a much more intensely sheared, pre-storm supercell environment.

More study (beyond the scope of this investigation) is needed regarding Ub and convective sustenance in shear profiles favorable for supercells, with the boundary types expanded to include large HCRs as well as drylines, reflectivity fine lines, fronts and other potentially initiative foci.

The large reflectivity feature may also represent a hybridized HCR and convergence boundary, in a demarcation zone between CBLs with slightly different characteristics. The surface flow veered slightly to the west in the area of the multiple small HCR signatures (Fig. 3); and modified soundings in that area (not shown) indicate a shallower, drier CBL. The veering surface flow also veered the mean CBL flow and shear vector, along which HCRs are typically oriented. A shallower CBL and veered flow may account for the difference in HCR size and orientation to the west of the larger HCR. This hypothesis may be inferentially verified through numerical modeling of the CBL based on properties taken from mesonet observations, given that there were no direct measurement of vertical temperature or moisture distribution to the immediate east and west of the possible HCR.


We thank the computer support staffs of SPC and OSF for their aid in making the data and images available. We appreciate the keen discussions and feedback from our colleagues at SPC and OSF, especially Dave Imy and Bob Johns. These scientists also offered constructive and insightful comments: Nolan Atkins, Todd Crawford and Dan Miller.


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.

Fujita, T.T., 1981: Tornadoes and downbursts in the context of generalized planetary scales. J. Atmos. Sci., 38, 1511-1533.

NCDC, 1999: Storm Data, 41, 5, 139 pp.

Thompson, R.L., and R. Edwards, 2000: An overview of environmental conditions and forecast implications of the 3 May 1999 tornado outbreak. Wea. Forecasting, in review.

Weckwerth, T.M., J. W. Wilson, R.M. Wakimoto, and N.A. Cook, 1997: Horizontal convective rolls: Determining the environmental conditions supporting their existence and characteristics. Mon Wea. Rev., 125, 505-526.

Wilson, J.W. and D.L. Megenhardt, 1997: Thunderstorm initiation, organization, and lifetime associated with Florida boundary layer convergence lines. Mon Wea. Rev., 125, 1507-1525.

Wilson, J.W., T.M. Weckwerth, J. Vivekanandan, R.M. Wakimoto, and R.W. Russell, 1994: Boundary layer clear-air radar echoes: Origin of echoes and accuracy of derived winds. J. Atmos. Oceanic Technol., 11, 1184-1206.