“An Overview of Environmental Conditions and Forecast Implications of the 3 May 1999 Tornado Outbreak” RICHARD L. THOMPSON AND ROGER EDWARDS By Andy Koehler and Joshua Mertzlufft

Overview Introduction Synoptic Environment Evolution Convective Initiation and Storm-scale Observations Discussion Summary and Implications

Introduction In the late Afternoon and evening hours of 3 May 1999, a violent tornado outbreak affected portions of Central and Northern Oklahoma and Southern Kansas 69 tornadoes documented from 10 tornadic supercells ▫- Comparable to 26 April 1991, most recent outbreak to affect Oklahoma and Kansas

Introduction

This outbreak had many difficult forecast problems ▫Initial supercell developed to the east of two weakly convergent drylines ▫Thick cirrus overcast reduced heating and boundary mixing ▫Well-defined jet streak 4-10 km AGL from eastern New-Mexico and Western Texas

Introduction Intent of the article is to document the environment and evolution of the 3 May 1999 outbreak from the view of operational forecasters ▫Offers the opportunity to describe the event from synoptic scale down to the storm scale ▫Hopes of furthering severe storm research with the goal of benefiting convective outlooks, watches and warnings

Introduction AK8Dkhttp:// AK8Dk

Synoptic Environment Evolution Moisture and Instability Vertical Shear

Moisture and Instability Mean large-scale trough located over western U.S. ( May 3 12 UTC) Embedded short wave over Arizona Large-scale trough amplified over the Rockies by 00 UTC on May 4 th Embedded short wave trough progressed from Arizona to western OK and KS Deepening sfc low located over central high Plains Low-level south to southeast flow over KS, OK, and TX

Moisture and Instability Sfc dewpoints (middle to upper 60’s) spread northward from central TX at 00 UTC, to central OK by 12 UTC, and southern KS by 20 UTC OK mesonat data showed diffuse drylines along the moist side of two moisture gradients Western boundary- denoted typical dryline Eastern boundary-weaker than typical dryline 2 boundaries were supportive of 2 transition zones in the moisture field

Moisture and Instability The diffuse boundary structures were less important than fact that dewpoints had increased to upper 60’s and sfc temps had warmed to the middle 80’s by 20 UTC in extreme SW OK This contributed to very large CAPE values and weak convective inhibition

Moisture and Instability 12 UTC Norman, OK sounding- revealed moist BL about 1 km deep, beneath elevated mixed layer from mb 18 UTC Norman sounding-some deepening of the BL from UTC, with 2-3 degrees C of warming btw 700 and 500 mb 00 UTC sounding- showed substancial warming from 600 mb to the sfc during previous 12 hours (2-4 degrees C)

Moisture and Instability Sounding from Fort Worth, TX revealed LL warming and moistening similar to Norman Both 18 UTC and 00 UTC soundings at Norman yielded a mean BL based CAPE near 5000 J/kg, with less than 10 J/kg of convective inhibition

Vertical Shear Time- height series of profiler winds from plots in the affected areas showed knot strengthening of the flow in the 4-10 km layer After that came a speed maximum in the 4-10km layer The observed strengthening of the lower and middle tropospheric flow in the afternoon, which indicated enhanced supercell potential

Vertical Shear Modest backing and strengthening lower level winds in Purcell, OK resulted in sufficient deep- layered vertical shear for supercells by early afternoon in the warm sector BRN shear values of around 55 m^2/s^2 near convective initiation 0-3 km SRH values of 120 m^2/s^2 for right mover and -46 for left mover in early afternoon

Vertical Shear BRN increased in the late afternoon from 62 to 123 to 166 m^2/s^2 while 0-3 km SRH increased from 80 m^2/s^2 at 1800 UTC to 338 m^2/s^2 by 2300 UTC Hodographs assumed pronounced clockwise curvature which strongly favored right-moving supercells In mid to late afternoon, SRH values across central OK remained in the range of m^2/s^2

Convective Initiation & SS obs All supercells were preceded by development of 2 short-lived convective towers over NW TX This convection formed under relative gap in the cirrus canopy and btw the sfc drylines Cirrus gap shifted northeastward across SW OK, where additional towering cumulus formed (SW of Lawton around UTC)

Convective Initiation & SS obs This convection rapidly evolved into a storm split (the first right-moving supercell, storm A) Storm B developed explosively within a small cluster of updrafts west of Altus around UTC, just west of eastern dryline Storms A & B-most prolific tornado producers of the outbreak Combined total of 35 tornadoes (including an F5 that moved across southern OK City and an F4 that hit Abell and Mulhall)

Supercell Character & Structure Central OK thunderstorms-displayed visual characteristics of classic supercells Storms A & B- tornadic periodically for 4-6 hrs Visual obs, along with radar reflectivity & velocity signatures indicated maintenance of classic precip distributions about the updraft and forward-flank regions once in tornadic phases

Supercell Character & Structure Storm-relative winds were supportive of CL supercells (21 m/s in the 9-11 km layer) During first 1-2 hrs of storms A,B,E, & G, storms exhibited visually skeletal cloud structures (which are often associated with LP supercells) However, radar imagery actually revealed CL precip patterns

Supercell Character & Structure Certain characteristics almost constantly observed in storms A & B ▫Relatively low cloud bases in the regions of strongest apparent LL rotation ▫Clear slots wrapping cyclonically around wall clouds ▫Above conditions are characteristic of many tornadic supercells

Supercell Character & Structure Cloud base heights appeared to lower as each storm passed northeastward over progressively smaller sfc dewpoint depressions Similar trends were observed in the lifted condensation levels (LCL’s)

Supercell Character & Structure Largest and most damaging tornadoes occurred after the LCL’s lowered from roughly UTC where storm A had developed in SW OK, to less than 500 m in the OK City area by 00 UTC Region of lower LCL’s coincided with region of largest SRH This conforms to the trend toward increasing tornado threat as SRH increases Structures of other supercells- generally consistent with the observations of storms A & B

Discussion Buoyancy and vertical shear profiles Convective initiation and low-level boundaries Variations from classical supercell structures

Buoyancy and vertical shear profiles Some aspects of the tropospheric evolution followed the typical sequence for Thunderstorm outbreak ▫Steep mid-layer lapse rates overspread the area ▫Boundary later moisture increased ▫Daytime heating further increased instability However anticipation of the severe tornado outbreak was hindered by poor operational model forecasts of wind speeds in mid and upper troposphere

Buoyancy and vertical shear profiles Large scale pattern recognition didn’t suggest the outbreak would ensue Perceived problems were derived from models that showed modest vertical wind shear and surface analysis that didn’t provide clear focus for convective initiation This being said, regional profiler data and soundings were consistent with other historical outbreaks

Buoyancy and vertical shear profiles

Each event is uncontaminated by convective outflow Each show a deep later of strong positive buoyancy Hodographs were very similar in lowest 3km

Convective initiation & LL boundaries Sfc analyses revealed a dryline on the meso- alpha to synoptic scales, which appeared to be a diffuse double dryline structure in a finer scale analysis Deep sfc low existed well to the NW of OK

Convective initiation & LL boundaries Sfc convergence was ill-defined along each dryline Moisture gradient was not particularly large Ziegler and Hane (1993) suggested that BL conv is critical for maintaining a pronounced moisture gradient along a dryline, as well as for thunderstorm initiation along it A large plume of cirrus overspread much of TX panhandle and western OK by early afternoon Raised questions about cont’d heating and mixing in BL through mid-afternoon

Convective initiation & LL boundaries The widespread high clouds & lack of conv in the dryline regions introduced considerable uncertainty regarding timing & location of convective initiation However, the gap in the cirrus appeared to be crucial in allowing cont’d sfc heating & mixing to maintain weak enough capping for initiation of storms A & B

Convective initiation & LL boundaries Upper-tropospheric divergence and possible weak inertial instability may have contibuted to a favorable environment for strong storm top divergence and sustained updrafts Smaller scale- storm A appeared to have formed near updraft portion of a large horizontal convective roll (HCR) in the BL This HCR updraft was a potential mechanism for convective initiation

Convective initiation & LL boundaries The presence and depth of HCR’s are difficult to anticipate and detect in an operational setting Forecasts for convective initiation will necessarily have large uncertainty when HCR’s or other subtle BL processes dominate

Convective initiation & LL boundaries VIS satellite imagery- revealed a series of billow clouds oriented meridionally over northern TX and south-central OK, to the east of where storms A & B developed These clouds associated with a relatively shallow, capped BL with backed sfc winds The storms approached this area and encountered a confluent zone & associated backed sfc winds, which yielded 0-3 km SRH values of m^2/s^2 The storms produced more numerous and progressively stronger tornadoes as they moved into this area of increased SRH

Convective initiation & LL boundaries Longevity of the storms after 00 UTC was in question due to a stronger cap well east of the confluence boundary Storm A ceased producing tornadoes by 0130 UTC and dissipated shortly after 02 UTC Storm D reached peak intensity only 30 miles south of where storm A dissipated These obs underscore the difficulties faced by forecasters in anticipating storm longevity & tornado production in an evolving environment with subtle but important variations

Variations from classical supercell structures At least two possible cases of structure storm interference were observed during the tornado outbreak ▫One in Wichita where the supercell weakened substantially after being overtaken by a large area of thunderstorms ▫Supercell in Parcell, OK was structurally altered by convection from the southwest. This storm was suggested to evolve into a violent tornado until the aforementioned disruption

Variations from classical supercell structures Not all wind profile parameters were strongly supportive of classic supercells ▫One model showed a weakness in storm-relative winds in the middle troposphere, which was seen as a limiting factor ▫It can lead to excessive cold outflow in the rear- flank which can undercut the mesocyclone before the formation of significant tornadoes

Summary & Implications The 3 May 1999 event in OK showed some general large-scale characteristics of historical severe thunderstorm outbreaks in the southern and central Plains. Many important points to be learned from this outbreak This event illustrates that outbreaks of strong & violent tornadoes are not necessarily associated with what many operational forecasters would consider to be the most evident large-scale patterns in numerical model outputs

Summary & Implications The potential for a severe weather episode was anticipated by SPC forecasters a day in advance, but major tornado outbreak was not forecast initially Much difficulty in anticipating the magnitude of the tornado threat hours in advance was the result of poor numerical model forecasts of mid- and upper- tropospheric flow, and associated deep-layered vertical shear However, obs data suggested a substancially greater and increasing threat of supercells with significant tornadoes, based on parameters derived from sfc analyses and profiler data from mid-morning to early evening on 3 May 1999

Summary & Implications When forecasting a threat for tornadoes, the mode of convective initiation and the # & spacing of supercells that form are critical to the # of tornadoes expected Storm spacing and motions were such that the supercells remained in an environment of favorable vertical shear and instability for several hours without numerous storm collisions This allowed supercells to produce large number of tornadoes

Summary & Implications The predominance of a supercell convective mode and lack of a squall line may have been attributable to the lack of strong LL conv near the drylines Lack of convergence suggested that supercells may not even develop Presented major forecasting challenges that day

Summary & Implications Subtle clues/precursors to storm initiation important Were present in high-resolution visible satellite imagery and WSR-88D data Important features: complex dryline, confluence zone, and HCR structures Even in the presence of mesoscale sfc obs, forecasters also may need to rely on careful interpretation of available satellite and radar imagery to recognize small areas where convective initiation is possible

Summary & Implications Majority of significant tornadoes during VORTEX- 95 occurred in association with sfc boundaries LL baroclinic boundaries were either ill-defined or not present during the May 3 rd outbreak This illustrates that the apparent weakness or absence of such boundaries in the warm sector does not preclude significant tornadoes, given the initiation of supercells in an otherwise highly favorable environment (characterized by large CAPE, strong deep-layered vertical shear, and strong LL shear)

Summary & Implications Observations suggested that subtle boundaries were important to initiation and evolution of several tornadic supercells in the May 3 rd outbreak It remains crucial for forecasters to monitor high-resolution data to anticipate local areas of enhanced or diminished tornado threat

Summary & Implications May 3 rd tornado outbreak exposed some problems in the application of recent supercell and tornado forecast techniques Some forecasters at the SPC anticipated a possibility of HP supercells because of: 1.) potential for storms to be seeded by adjacent storms, and 2.) inaccurate model forecasts of weak mid- and upper-tropospheric flow and associated weak storm-relative winds

Summary & Implications Operational forecasts of supercell types and tornado potential will continue to be difficult for the foreseeable future Especially when the processes responsible for creating “favorable” supercell tornado environments occur on the mesoscale, or even storm-scale The more general combinations of CAPE and vertical shear clearly supported tornadic supercells by the late afternoon and evening of 3 May 1999 This suggests that the background environment, as opposed to just storm-scale variations, can be a dominant controlling factor in regional tornado outbreaks

Summary Introduction Synoptic Environment Evolution Convective Initiation and storm-scale observations Discussion Summary and Implications

Questions?