May 04, 2003 Tornado Outbreak

A classic "corridor" tornado outbreak occurred this day. We sampled three separate tornadic storms, ultimately witnessing a long-track strong to occasionally violent tornado that moved from 2 miles SW of Pierce City, MO to Springfield, MO. We captured the most dramatic close-range video footage of a strong multiple vortex tornado I've seen.

This event was a classic "Fawbush-Miller" (1) tornado outbreak. As with most such events, a significant upper-tropospheric mobile trough was involved in generating the very strong vertical wind shear and instability over a large portion of the outbreak region. The 500 mb chart valid 12Z May 04 (Fig. 1) illustrates the strength of this trough, evinced in part by the strength of the wind maximum noted in the base of the trough at this time (80kt at Tucson, AZ and El Paso, TX). Fig. 1 is also a very simple example of "compositing" - presenting information from various levels on a single chart. On this particular map, I've only drawn two items: (a) the orientation and likely location at t+3 hours of the 500 mb jet streak and (b) the morning location and orientation of the moisture axis (closely corresponding to the position of the low-level jet stream), including maximum wind speed and highest dewpoint temperature (TD), noted at 850 mb. Full-blown composite charts are much more complete, attempting to consolidate salient information from many pressure levels. However, Fig. 1 is sufficient to show that there is tremendous "cross-over" between mid and lower level wind maxima over the plains, implying the presence of enormous vertical shear (especially given the magnitude of the winds at 850 and 500 mb) and potentially large CAPE (implied by the superposition of -11C to -13C 500 mb temperatures over TD=16C at 850 mb). Moreoever, it is clear that the 500 mb trough is not going to sit over the southwestern U.S. but rather is clearly mobile and will move northeastward into the plains during the day. A short-term forecast of the 500 mb (Fig. 2) and 850 mb (Fig. 3) pressure level conditions suggests that indeed the mid-level trough and associated intense mid-level winds will superpose strong southwesterly lower tropospheric flow, resulting in large vertical shear over a large region. Sounding analysis (not shown) clearly demonstrated that the thermodynamic stratification was classically favorable for tornadic supercells; the morning sounding composite analysis (described in the April 30 synopsis) revealed that the "free atmosphere" (i.e., that portion above the boundary layer) was nearly homogenous (i.e., characterized by a well-defined EML - elevated mixed layer) and that the soundings taken in the "warm sector" possessed a breakable cap atop deep, rich low-level moisture. Daytime heating with little further moistening yielded CAPEs of 3000-4000 J/KG, more than sufficient for severe thunderstorms. The presence of a moderately strong cap (i.e., non-negligible CIN) and deep-layer mean wind vectors oriented across the dryline suggested that discrete supercell storms would rule the day. Recent efforts to develop parameters consolidating the CAPE, shear and other data into single measures of tornado potential have produced some success. The so-called "significant tornado parameter" (STOR; Thompson et al. 2002) can be a useful tool for measuring the likelihood of strong tornadoes given the initiation of discrete thunderstorms. The STOR parameter is designed so that a value of 1 or greater denotes "threshold" strong tornado potential based on the findings of earlier studies establishing empirically-derived approximate threshold values of the component parameters that contribute to STOR (of course, conditions may occasionally be favorable for the production of strong tornadoes even if the STOR parameter fails to highlight the threat); by mid-afternoon May 4, the STOR parameter was "off the charts" (Fig. 4), further underscoring the potential gleaned from analysis of the individual components (CAPE, shear, etc..). One aspect of the day's setup that caught my attention was the restricted width of the unstable warm sector (Fig. 5, note the relatively narrow width of real estate between analyzed warm front and dryline), attributable in part to the cold air mass sitting over the Midwest. This factor, plus the obvious presence of "outbreak conditions" over all of the prospective target area, influenced our decision to play storms that we felt would surely form down the dryline over southeastern Kansas/southwestern Missouri where the warm sector was wider as opposed to storms forming near the warm front/dryline intersection (the triple point; the region that produced the devastating Kansas City tornadoes earlier in the afternoon) near the northern end of the instability axis. The idea is that storms would first develop near the triple point and then "back build" down the dryline as the cap broke progressively further south; moreover, the prevalence of conditions supremely favorable for significant tornadoes (see preceding discussion) suggested that every single isolated supercell would have an excellent chance of producing tornadoes. Under this working notion, any storm that you targeted east of the dryline within the larger target area would produce a tornado so that even if you screwed up intercept on one storm, you'd get a second chance. Such a "luxury" is very rare, indeed, and on this day our idea worked.

Unbelievable day in the southeastern plains today! We began the day in OKC, determined to consider playing the tail end of this system. My concern (in addition to storm motion) was that the best show might occur in a narrow and early window over SC-E Nebraska (16-19Z) and then it would be over. Upon waking and looking at the data, it became apparent that the setup was indeed outbreak-like and that therefore one could likely pick about any storm and expect to see large tornadoes. Analysis soon suggested that the cap would break explosively in the HCRs (horizontal convective rolls - bands of congested cumulus from which the most important storms of the day often develop; these bands are generally oriented along the low-level shear vector) embedded within the highly unstable, strongly sheared warm sector. Stopping in El Dorado for fuel we watched as our surface winds slowly veered and our dewpoint spiked lower. Surface analysis revealed an intense low-level dry punch surging inexorably eastward towards SE KS and NE OK. I couldn’t believe it when I downloaded the SPC STOR and saw the 8-11 values blanketing the entire risk area from SE NE to N TX.

We decided on the spot to target SE KS, first intercepting storms on the northern portion of the E KS theta-e tongue and then dropping southward on different storms. We drove east on Hwy 54 to Iola and then north on 169 to Welda, KS where we sat for close to an hour watching convective towers get shredded to bits. Radar downloads revealed an intensifying cell moving NE towards the area south of CNU; rather than jumping on that as soon as we saw the first reflectivity core in NE Osage County, Oklahoma we sat watching convective towers closer to us. However, as the cell slowly matured and moved into Chautauqa Co. (see the radar downloaded via cell phone at the time, shown in Fig. 6 below), we could no longer sit still as we became convinced that this storm would become a prolific tornado producer. Driving south and then navigating around the NE side of CNU, we dropped SE in a desperate attempt to get ahead of this incredible storm (radar shot at the time, Fig. 7). As we passed the junction of Hwy 57/59 driving south on Hwy 59, we drooled at the spectacle of the rock hard convective bomb to our ESE. Following the western edge of the updraft down to cloud base, we immediately spotted a classic stovepipe tornado spewing debris about 7-8 miles to our E, somewhere near or just east of the Neosho wildlife area. Cursing at our failure to jump right on this storm as quickly as we noticed it on radar, we drove east on 160 out of Parsons, quickly taking up the rear of a very long and very sloooow moving caravan of storm chasers. We busted north and east on Hwy 126 intending to eventually intercept the classic tornadic supercell east of Girard, KS at or near the MO border. I gasped at the radar depiction of the storm’s hook overtaking Girard and wondered what kind of devastation was in store for towns east of this storm. We were making great time and might have actually outflanked the storm as planned. However, fresh radar downloads revealed a very good looking tail end storm in Nowata Co. OK and we became convinced that this storm would eventually go on to produce significant tornadoes provided the dreaded 80-mph left mover to its SSE didn’t interfere too badly (this bit of bad luck killed what might have been a strong tornadic storm the day before near Childress, TX). On the way south to intercept the OK storm, we quickly detoured to check out the updraft base of another supercell moving NE through Labette Co., KS and quickly spotted a truncated cone tornado emanating from cloud base to our distant west. The storm started to look a bit more organized shortly afterwards and we pressed south, worrying that the Labette Co. storm would be interfered with by the tail end charlie and would too seriously detract us from intercepting the southern storm. Driving through scattered golfball-sized hail in the FF core, we observed the updraft (completely separated from FF core) and immediately noticed a fat, nubby funnel cloud just NE of Fairland, OK (we later learned that this was a weak tornado but couldn't see the ground-based circulation for all of the trees and hills). Soon thereafter, the storm became rather disorganized as it merged with the left mover. Nonetheless, we persisted chasing this storm noting that inflow was still quite strong and moist southeast of the RF gustfront. The storm began to intensify and take on scary HP/classic hybrid characteristics east of Neosho (which was devastated by a tornado on 4/24/75). Racing east on Hwy 86, we attempted to get to Hwy 60 north of Newtonia but the storm’s horrendous looking rear-flank core soon overtook the road before we got there and, fearing very large hail and powerful winds, we turned around and continued E and SE on Hwy 86, eventually turning north on Hwy 37 at the town of Purdy. All this time, we were experiencing the storm’s ambient inflow which was 76/66 on 1725 winds. A new radar download revealed that this storm had retained classic supercell character with a vengeance and we wasted no time at all getting north to Monett, MO. The clefted-out barrel-shaped meso of an incredible beefy tornadic storm revealed itself to us soon after turning north out of Purdy and, 1 mile south of Monett, we observed a very large strong/violent tornado just northeast of town. We turned east on 60 and for the next 35-40 minutes experienced one of the most thrilling and fascinating tornadic experiences I’ve had. We watched in amazement as this fat, inky black-blue tornado rampaged across the countryside between Monett and Verona, spewing huge plumes of dirt/debris skyward. The tornado morphed into a fabulously fluid and dynamic multiple vortex tornado with evolutions that are best viewed on video rather than described. We got closer and closer to this high-contrast, long-lived tornado, watching as it lofted debris into cloud base and fearing for all in its path. I relayed reports to people with access to NWS numbers the best I could. I fear the tornado hit the north side of Aurora, MO. At times we got as close as 2/3 of a mile from the tornado as we drove northeast on Hwy 60 and then east on farm road 2200 south of Marionville and then Hwy 173 east of town. We got farther south of the tornado than we would have liked by this time (after being so close for so very long) but we got view of it again beneath the center of the incredible ground-dragging, clefted mesocyclone SW of Springfield: now it was an intense looking high-contrast stovepipe tornado with a fan-shaped debris cloud near Battlefield, MO. We watched as the tornado endured a classic rope-out on the west side of SGF, sparing the fortunate citizens of that city a bad tornado disaster.


First view of tornado as a large strong tornado north of Monnett. It had just moved out of Pierce City at this time.	Tornado begins a lengthy multiple vortex stage.	View towards the north of tornado and classic larger-scale storm structure; stopped on Hwy 60 at Aurora, MO.	Close up shot of debris cloud beneath truncated cone funnel on outskirts of Aurora; note large debris chunks in air.

Next few shots are along Hwy 60 between Aurora and Marionville.

Tornado morphs to more solid cone shape at Marionville.	Just east of town, another classic multiple vortex evolution is taking place beneath a meaty cylindrical cloud base.

Center stovepipe funnel surrounded by debris and dirt.	View to the north along Jasmine Road south of Clever; tornado is once again a fat stovepipe headed for Springfield, MO (gasp!).	Luckily for residents of Springfield, tornado ropes out on the southwestern edge of town; shot to the NW from Hwy 14 and Hwy 160.

(1) Fawbush and Miller were two USAF officers who got extremely lucky with a tornado forecast that ushered in the modern era of severe thunderstorm forecasting. Read an interesting article about that fateful first experiment at tornado forecasting here.


Fig. 1: Bare-bones composite chart.	Fig. 2: 12-hr RUC forecast 500 mb chart (geopotential height and wind)	Fig. 3: 12-hr RUC forecast 850 chart (same data as in Fig. 2)	Fig. 4: STOR parameter at 18Z

Fig. 5: 19Z surface analysis	Fig 6: Base reflectivity 2038Z (Tulsa radar)	Fig. 7: Same as in Fig. 6, except at 2104Z; note the pronounced hook/inflow notch on the storm's southwestern flank, along with elongated forward flank anvil. This storm produced the Girard, KS tornado shortly after this time. Our position at the time indicated by white dot.	Our route (blue line), tornado beginning (upside down triangles) and path (black hatched line). Note the small detour to the north at Newtonia where we tried to cut in front of the storm; we didn't make it because the intense hail core on the south edge of the developing hook cut us off.