Wednesday, May 20, 2009

May 13 review: Killer tornadoes strike northern MO

A powerful supercell produced a few damaging tornadoes across portions of far northcentral/northeastern MO on May 13. The first of these, rated EF-1, killed a woman in a mobile home near the town of Milan MO. The second, also rated EF-1, passed near the town of Novinger MO but caused no injuries/fatalities. The third tornado produced EF-1 damage in northern sections of Kirksville MO, also killing two more people to the northwest of town. This tornado strengthened to an EF-2 as it moved toward the town of Clay MO. The storm went on to produce two more weak tornadoes north of Edina MO.

A pair of elevated TS clusters were ongoing over portions of the Midwest on Wednesday morning, to the north of the surface warm front. The lead cluster over much of IL was driven by sharp moisture advection/convergence in the exit region of a 60-70 kt westsouthwesterly LLJ, while the trailing, more intense cluster was crossing the northern third of MO in advance of a potent low-amplitude 7H shortwave trough. An MCV, remnants of severe convective storms across the southern High Plains the prior evening, was also crossing the Ozark mountains at this time… and began to ignite surface-based storms after 15Z across far southern MO.

In the wake of these convective systems, a pronounced EML was overspreading the central Plains amidst weak shortwave ridging…well downstream and equatorward of the primary, strong shortwave rotating into the northern Plains. In fact, the central Plains capping thermal ridge underwent amplification during the late morning and early afternoon, as evidenced by 7H temps increasing from 11°C to 13°C at TOP between 12Z and 18Z (see RAOB overlay below, with extreme mid-level lapse rates by 18Z). Also note, despite the presence of the stout cap aloft, diurnal mixing still took a pronounced toll on prefrontal boundary layer moisture...with the richest moist layer becoming decidedly shallow by 18Z. Observed 100 mb ML dewpoints on the 18Z TOP and OUN RAOBs only ranged from 60-64°F… not a particularly positive sign. Still, it appears that persistent south-southwesterly low-level flow ahead of the cold front, coupled with evapotraspirative effects, helped boundary layer moisture quality to recover again through the afternoon (ML dewpoints increasing back to the 64-68°F range) per 00Z RAOB data.

Short-term guidance and conceptual models suggested SRH would be marginal for a significant tornado threat over the majority of the area east of the cold front, given 1) veering of near-surface winds to SSW amidst deep vertical mixing, and 2) the primary LLJ axis shifting away from the area. The obvious exception to this would be in the raincooled, gradually recovering BL airmass across northcentral and northeastern MO… where guidance suggested 0-1 km SRH AOA 250 m2/s2 would be possible given not only weaker BL mixing (more backed near-surface component) but also the maintenance of a somewhat stronger LLJ (~40 kt) nearer the primary low-level cyclone riding the U.S.-Canada border. By mid-afternoon (see crude 21Z sfc analysis below), surface observations showed the warm front rapidly redeveloping northward toward the I-80 corridor in IA/northern IL; while to its south, modifying outflow air was evident via a dense VFR stratocu deck and south-southeast sfc winds amidst low 70s°F over upper 60s°F. Given sfc winds just to the north (e.g. far southeastern IA) were slighly more veered, it is possible that an outflow boundary remained anchored along and just south of the Kirksville-Quincy-Winchester corridor. Regardless, observational trends clearly supported better-than-progged SRH/LCLs for a sig tor threat over this modifying airmass. The primary question was the strength of the low-level destabilization, given that BL heating would remain modest and a formidable 7H thermal ridge was still cresting not far west of the area via the 18Z TOP RAOB.

Shortly after 21Z, a combination of frontal lift/convergence, weak height falls aloft, and rapid cooling of the capping layer from the west all resulted in explosive convective initiation along the cold front from the KS Flint Hills up toward Des Moines IA (see visible satellite pic below). More importantly, a prefrontal confluence band—coincident with a narrow low-level thermal ridge with sfc temperatures warming into the low 80s°F—initiated a broken line of storms just east of the I-35 corridor, from Kansas City to Princeton MO. The fact that this latter band of storms was able to initiate ahead of the primary surface boundary was probable confirmation of what afternoon RUC model soundings had suggested: that the very warm capping thermal ridge had held primarily west and south of the MO river; while a very deep/moist boundary layer, with essentially no capping inversion aloft, had evolved amidst the recovering airmass over far northcentral and northeastern MO.

Given westnorthwest-oriented deep layer shear vectors atop the cold front and prefrontal trough, discrete cellular mode actually dominated for the first few hours after initiation. Additionally, stronger flow fields aided the northern storms in moving eastward off the boundaries, while markedly rightward-deviant motions amidst very strong CAPE helped southern storms move off the cold front across portions of KS and eventually northern OK. With a consolidating/strengthening supercell emerging across far northcentral MO and crossing the effective outflow boundary near Milan MO… the stage was set for a cyclic tornadic supercell.

A representative thermodynamic environment for this tornadic supercell was attained by modifying the 00Z DVN RAOB alongside the 00Z TOP RAOB (Kirksville sits about halfway between each). The surface warm front lifted through DVN (~120 miles NE of Kirksville) just prior to the 00Z RAOB launch, which means its sounding--characterized by 2100 J/kg MLCAPE and no MLCINH, despite a saturated boundary layer--was strongly representative of the modifying outflow pool. Meanwhile, just behind the low-level cold front, not only was TOP’s capping inversion aloft (having cooled very markedly the past few hours) similar in magnitude to DVN’s… but the entire thermal profile above the frontal inversion was nearly identical to DVN’s as well. It’s possible that mid-level thermal ridging near Kirksville wasn’t quite as suppressed as at TOP (and yet hadn’t advanced as far NE as DVN)… but this is assumed to be negligible. Below is the modified DVN sounding to represent the Kirksville event. Note moderate-bordering-on-strong instability (MLCAPE = 2455 J/kg); a very deep/moist boundary layer, with remnant EML above 650mb; and strong boundary layer RH (MLLCL < 800 m). Although MLCINH is weak to nil... note that low-level instability is not particularly strong (0-3 km MLCAPE = 41 J/kg), given the nearly saturated, weak-lapse-rate environment below 3 km.

Creating a single, strongly representative sr-hodograph for this event is more difficult given the gradual veering of low-level winds observed in the modified outflow air between 21-00Z. Regardless, the DVN RAOB and Winchester IL profiler were ideally situated to sample the vertical shear within this airmass, and both indicated a very strongly sheared low-level environment for the duration of this event--with very large 0-1 km SRH. Hodographs #1 and #2 below are those observed on the Winchester profiler (N of the effective outflow boundary and ~100 miles ESE of Kirksville) at 21Z and 00Z. Note that veering 0-1 km flow by 00Z is offset by acceleration of the synoptic LLJ to near 50 kt... actually resulting in an increase in SRH, given low-level winds were still strongly backed in a storm-relative sense. The third hodograph, from the raw 00Z DVN RAOB (120 miles NE of Kirksville)--likely launched around 23Z--is a good "medium" between the two... and also produced the largest low-level shear of the three, with 0-1 km SRH at a whopping 460 m2/s2. [Short-term model guidance vastly underestimated the size and pronounced "sickle shape" of these hodographs within the modifying outflow air... with NAM and NAM-KF models forecasting more veered near-sfc winds and a weaker LLJ.] Deep layer shear also appeared favorably strong via these hodographs, with around 50 kt of 0-6 km bulk shear. One final thing to note is that mid- and upper-level storm-relative winds on the three hodographs are pretty modest (20-35 kts)... and when coupled with anomalously rich PWATs (> 1.5"), this probably contributed to HP supercell processes that acted to temper the strength of these tornadoes. [Given that a very rare combination of ~2500 MLCAPE and ~400 0-1 km SRH existed amidst ideal low-level thermodynamics (albeit modest 0-3 km CAPE)... it would seem that the tornadoes from this storm could have been far more violent than they were.]

Finally, below are representative profiler hodographs for the tornadic supercells affecting the Burlington KS and Lamont OK areas, respectively. The latter tornado was a surprisingly long-lived and nearly stationary cone. The thermodynamic environment in these areas was characterized by higher LCL/LFC, but very strong instability (MLCAPE 3500-4000 J/kg) and steep low-level lapse rates/no CINH. This may have facilitated strong near-ground stretching to help in the production of tornadoes. Also, low-level hodographs remained a bit more kinked than forecast... and given the strongly deviant storm motions (almost too strongly deviant in the Lamont case, given the increasingly parallel nature of the storm motion with the near-sfc shear vector)... 0-1 km SRH was actually pretty favorable, in the 140-170 m2/s2 range. Note that deep layer shear is solidly on the low end for supercell storms though (~30 kts of 0-6 km bulk shear)... with the large instability likely playing a role in allowing for sustained supercells.