On one day (26 May 1991) a north-south dryline was located over the east Texas Panhandle. A cluster of tornadic thunderstorms formed during late afternoon near the Texas-Oklahoma border in the northeast panhandle and moved into northwest Oklahoma. Initiation occurred at the intersection of the dryline and a convergence line (observed by aircraft and characterized by a convective cloud line) that extended west-southwestward from the intersection point. The convergence line very likely formed in response to heterogeneities in the underlying surface characteristics in the along line direction. Specifically, the surface was covered with sparse vegetation over about 100 km in the north-south direction (Canadian River Valley) west of the dryline with more vegetated areas both to the north and south. Over this dryer area the atmosphere heated more efficiently (as verified by satellite observations), and vertical mixing was likewise more efficient. This resulted in preferential downward transport of westerly momentum over the same area (as verified by the presence of a mesoscale dryline bulge) in comparison to surrounding areas, and a line of convergence along the southern edge of the sparsely vegetated area.
In the second case (16 May 1991) the dryline extended southward from a surface low that was associated with a translating upper level trough. During midday the dryline moved rapidly across western Oklahoma and a synoptic scale bulge developed. In early afternoon a discontinuous jump in position took place wherein a second dryline developed farther east while the original one to the west lost definition. To the north of the bulge an elongated swath of cooler air oriented approximately normal to the dryline was present and resulted from the passage of thunderstorms over the swath during the previous night. Just to the south of the cool swath and just west of the dryline preferential warming was observed through the depth of the boundary layer by aircraft and by satellite (see Figure 1). Lowered pressure just west of the dryline bulge likely backed the winds ahead of the dryline and produced enhanced convergence along a short dryline sector (consistent with pronounced clear air radar return along this sector). As a result this was an along-line location where thunderstorms developed. Convective cloud development was depressed (Fig. 1) along the section of the dryline that passed over the cool swath, owing to increased latent heat flux and reduced sensible heat flux over the wetted soil, and consequently a shallower and more moist boundary layer. On the north side of the cool swath along the northern border of Oklahoma another cluster of thunderstorms developed at the intersection of the dryline and a short cloud line that extended into the dry air. Low-level convergence was enhanced at this intersection as verified by aircraft measurements. The short cloud line existed for 2.5 hours prior to the development of deep convection, and appears to have originated over the region that received the heaviest rainfall on the previous night. The mechanism for development of this short line is unknown, but likely related to enhanced evaporation of water from the surface.
On a third day (15 May 1991) another quiescent north-south dryline was again located in the eastern Texas Panhandle. The MM5 model was run in triple-nested (18, 6, and 2 km spacing) format from initial condition specified on the large scale at 12 UTC that morning in an attempt to simulate the dryline location and convective initiation along it. The simulation produced a late afternoon dryline that was about 30 km farther west than observed, and deep convective initiation occurred about 90 minutes earlier in the simulation than was observed by radar. The character of modeled and observed convection (scattered line) was similar. Prior to convective initiation the simulation included pronounced boundary layer horizontal convective rolls in the dry air west of the dryline. Deep convection first occurred just east of the dryline near the intersections of particularly vigorous rolls and the dryline. The downdraft portion of the roll circulations to the immediate west of the surface dryline appears to have preferentially transferred westerly momentum from aloft to the surface, as indicated by mesoscale bulges in the dryline near these roll intersections. Enhanced low-level convergence and upward motion ahead of these bulges appear to have provided a favorable environment for convective initiation. Unfortunately, the initiation region of the observed convection was not within range of any radar that could sample clear air return.
The common thread among all these cases is that processes in the dry air (in a variety of forms) appear to have increased low-level convergence in local areas along the dryline leading to convective initiation. The paper will present highlights of the analyses performed in each case.