| For many years, forecasters have analyzed the evolution of surface meteorological fields to gain insight into severe thunderstorm and tornado outbreaks (e.g., Miller 1972; Moller 2000). Numerous studies have shown relationships between surface features and convective initiation and organization. Structures of interest have included spatial and temporal patterns of dry line bulges (Tegtmeier 1974), low pressure systems (Tegtmeier 1974; Moller 2000), thermal and moisture ridges (Miller 1972; Moller 1979), and outflow boundaries (Markowski et al. 1998a). Outflow boundaries may be determined directly by the means of mesoscale surface analysis, or inferred by noting where the WSR-88D accumulated precipitation algorithm indicates that previous storms have laid down a swath of soil-wetting precipitation (Markowski et al. 1998a). Frequently more than one of the favored features is present and it becomes an enormous challenge for the forecaster to decide which features will trigger convection and which will influence the evolution of the thunderstorms. As one might expect May 3, 1999 was no exception. A moist, unstable boundary layer, a dry line and other surface boundaries, and vertical wind shear favorable for supercells covered a large geographic area. Were there clues in the surface fields that might have suggested the location of convective initiation and convective mode? The surface features in this event were well sampled in both the traditional synoptic scale data sets and in the Oklahoma Mesonet. Four analyses of traditional fields were conducted; a manual analysis of the synoptic data, a manual analysis supplemented by mesonet data, an objective analysis of the synoptic data, and an objective analysis supplemented by mesonet data. Aspects of the four analyses are compared. |
The evolution of the surface features prior to and during the early stages of the Oklahoma/Kansas tornado outbreak of May 3, 1999 likely was influenced by antecedent rainfall. On May 3 the dry line moving out of the Texas panhandle was retarded in its eastward progress relative to the advance farther south out of southwest Texas. Also, the airmass west of the dry line was not as dry over the panhandle as farther south. It is surmised that the soil moisture available in the panhandle from the previous days' rainfall slowed the vertical mixing process, whereas farther south the much drier soil permitted much greater mixing of moisture and momentum. Thus, a surge in the dry line out of southwest Texas became the focusing mechanism for convective initiation on May 3, 1999.
From April 29 to May 1, 1999 several periods of thunderstorms occurred over the Texas panhandle and west Texas north of Midland. The twenty four-hour rainfall estimates for April 30 (figure 1) and May 1 (figure 2) illustrate the distribution of rainfall over the panhandle and west Texas a few days before the outbreak. The dark areas in the Texas panhandle represent rainfall estimates of two to three inches.
The effect of the May 2 north Texas MCS is seen in the analysis of potential temperature shown in figures 4 and 5. The cool air over north Texas at 22 UTC on May 2, 1999 was evident when the MCS was in its mature stage. Twenty-four hours later, at 22 UTC May 3, 1999, the cool air was still evident over eastern Oklahoma. The southwest edge of the cool air provided a boundary that extended from north central Texas to central Oklahoma.
The analyses strongly suggest that four air masses were present across the southern plains on May 3, 1999, consisting of two air masses in the relatively dry sector across west Texas and extreme western Oklahoma, and two additional air masses in the moist sector further east. The relatively cool air mass produced by the heavy rainfall in north Texas on May 2 shows clearly in the potential temperature field minimum at 2200 UTC (figure 4). During the next 24 hours, this cool, moist air mass moved north with the flow into central Oklahoma (figure 5). The progress of this boundary into Oklahoma may be important in explaining the strength and longevity of the tornadoes that struck the Oklahoma City area on May 3.
However, at 2200 UTC, the surface pattern had become more complex, with an apparent double dry line pattern (figure 7). During the day, dew points rose west of the original dry line across the Panhandle, likely because of extensive evapotranspiration and evaporation from the rainfall that occurred on previous days and throughout much of the spring. Another dry line segment had appeared west of Amarillo by 2200 UTC, apparently marking the eastern extent of deep mixing. It is likely that ground moisture slowed the deep mixing process from near Amarillo eastward, resulting in the double dry line, and subsequent weak convergence along each of the dry line segments.
Additional manual analyses using mesonet data were useful for adding more detail. See, for example, figure 8. The analysis revealed a mesoscale low near the dry line bulge, and a low pressure trough extending east of the low. This trough may represent the southern edge of the relatively cool, but moist air mass that entered central Oklahoma from North Texas. The temperature analysis revealed a pool of cool air over central Oklahoma and the dew point analysis showed pooling of very moist air along the pressure trough. The resulting solenoid likely contributed to enhanced streamwise vorticity over central Oklahoma. Additionally, the low and trough likely resulted in backing of surface winds, increasing the 0-1 km vertical wind shear that appears to be critical for supercell tornado formation (Markowski 1998b).
Objective analyses that included mesonet data also revealed additional detail in the surface features. One example is the surface moisture flux divergence field. An analysis using synoptic observations (figure 9) revealed a large area of moisture convergence over southwest Oklahoma.
The terrain may also have contributed to storm initialization, as the Oklahoma City storm formed near the Wichita Mountains northwest of Lawton. The other early storms formed further west, near the Quartz Mountains in southwest Oklahoma. These storms, and several others that developed during the evening, moved north-northeast into central Oklahoma, producing over 70 tornadoes. Meanwhile, the complex events in the Texas Panhandle and western Oklahoma may have minimized the possibility for more widespread thunderstorm and cold pool formation, that could have ended the tornado outbreak through stabilization of the moist air mass.