P6.14                        CINCINNATI, OHIO TORNADO OUTBREAK OF 9 APRIL 1999 - A CASE STUDY

1.  INTRODUCTION

        On the morning of 9 April 1999 (hereafter, April 9th), just after 0900 UTC (5 am EDT) the northeast suburbs of Cincinnati, Ohio (specifically, the communities of Blue Ash and Montgomery) was the location of a deadly F4 tornado, where four people perished [the second deadliest tornado day the Cincinnati metro area (CVG in Fig. 1) has ever seen].  This F4 tornado was one of five tornados that impacted southeast Indiana and southwest Ohio during the pre-dawn hours of April 9th.  Two additional tornados were observed in west-central Ohio around this same time.  Plotted in Fig. 1 are all seven tornados with locations of their associated lowest level storm-relative velocity circulations taken from the National Weather Service (NWS) Wilmington, Ohio, (KILN) Weather Surveillance Radar-1988 Doppler (WSR-88D).
        The synoptic and thermodynamic setting quickly evolved to support severe weather development during the early morning hours of April 9th.  By 0600 UTC, the Ohio Valley was under the influence of both a dual upper level jet configuration, as well as a strong low level jet.  This combination helped provide strong and deep upward vertical motion, as well as critical shear to the region.  Simultaneously, moist unstable air was being advected into the area in advance of an approaching frontal system. Mid level drying was also occurring which would play a significant role in the developing severe weather.
        Five of the seven tornados this day were focused along a pre-existing, low level storm-relative shear axis (not shown) that stretched from southeast Indiana through Cincinnati, and on into southwest Ohio.  This shear axis remained nearly stationary ahead of the tornado producing thunderstorms.  Of these five tornados, the F4 was the only tornado in which the parent thunderstorm exhibited supercell characteristics (Moller et al. 1994).   In fact, it displayed both classic and High Precipitation (HP) characteristics. Moller et al. (1994) defined a classic supercell as a single steady-state cell, which deviates to the right of the mean wind (Northern Hemisphere) and contains both a mesocylone and a bounded weak echo region (BWER).  Its storm-top is also displaced above a low-level pendant (Lemon 1980).  An HP supercell contains these same characteristics with the addition that its mesocyclone contains substantial precipitation (Moller et al. 1994).   At the time of the tornado touchdown in the Blue Ash/Montgomery area, this classic supercell soon transitioned into an HP with the tornado becoming completely rain wrapped.  The other four tornados associated with this same shear axis were not as straight forward with regard to any dominating radar signatures.
        During this outbreak, two additional rotating thunderstorms were observed.  One affected west-central Ohio, while the other moved across northern Kentucky.  As denoted in Figure 1, the west-central Ohio storm produced two additional tornados, while the northern Kentucky storm produced only sporadic (non-tornadic) wind damage.  This paper will attempt to provide an understanding of the pre-convective environment surrounding those storms which affected southeast Indiana and southwest Ohio (Section 2), including both a synoptic and thermodynamic assessment.  Storm scale radar analysis will be provided in Section 3.
 

Figure 1.  Tornado tracks (T1 - T5) and F-scale classification during April 9th Cincinnati area tornado outbreak.  Location for lowest level circulation denoted by large dots.
 

2.  PRE-CONVECTIVE ENVIRONMENT

2.1  Synoptic Assessment

        According to the 0000 UTC Eta model, the upper levels (250 mb) were noted as having a dual jet structure in place by 0900 UTC on April 9th (Fig. 2).  The northern jet was oriented northwest-southeast over the eastern Great Lakes, while the southern jet was oriented west-east through the mid-Mississippi Valley.  Maximum winds associated with both jets were 120 kt.  The orientation of these jets would act to enhance the upward vertical motion over the Ohio Valley.
        At mid-levels, a significant dry intrusion (as noted by decreased Theta-E values) would move into the Ohio Valley region between 0600 and 1200 UTC.  This dry air would serve as a necessary ingredient to enhance a storm's downdraft potential.  At 850 mb, strong convergence, driven by a 60 kt southwesterly jet, would be in the same region.  The orientation of the combined upper- (250 mb) and lower-level (850 mb) jets infers the potential for strong and deep upward vertical motion.  According to Uccellini and Kocin (1987), both the right rear entrance region and left front exit region of an upper level jet are favored locations for the development of a transverse ageostrophic secondary circulation.  The role of this circulation is to convert available potential energy into kinetic energy, thus enhancing upward vertical motion.  A cross-section showing the resultant vertical/tangential motion of the ageostrophic wind, which was oriented north-south between central Michigan and eastern Tennessee, demonstrates the effect this upper level dual-jet configuration had on enhancing upward vertical motion in the vicinity of southwest Ohio (not shown).
        At the surface, a strong low pressure system would move east out of the mid-Mississippi Valley.  This low would track along a quasi east-west oriented frontal boundary that was situated about 90 nm north of Cincinnati (Fig. 2).  Surface-based convergence was strongest along this east-west oriented front.
 

Figure 2.  Composite chart valid at 0900 UTC on April 9th.  Branches of the polar, subtropical, and low-level jets, along with surface low and associated frontal boundaries are shown.  Contours of surface-based CAPE (J/Kg) also shown.
 

2.2 Thermodynamic Assessment

        The environment across the Ohio Valley varied considerably with respect to shear and instability.  According to the 0000 UTC Eta hourly model sounding output for Cincinnati, helicity was expected to be over 400 m²/s² by 0900 UTC while the positive shear was expected to be greater than 30 (x10^-3 /s).  The model output wind profiles indicated the strong shear across west-central Ohio was primarily a result of veering low-level winds in association with the surface frontal boundary.  Further south (near the Ohio river), while directional shear was evident, speed shear appeared be more dominant.  Forecast surface-based Convective Available Potential Energy (CAPE) from the 0000 UTC Eta model indicated values less than 500 J/kg would be realized across the Ohio Valley between 0600 and 1200 UTC.
        Both the position and the magnitude of this forecast surface-based instability axis proved to be substantially unrepresentative.  The surface RUC analysis, for the period 0600 - 1200 UTC, indicated that the axis of instability (as noted in the CAPE field), was actually further east and much stronger than forecast by the 0000 UTC Eta.  At 0600 UTC, the area of maximum instability (3500 J/kg) was located over Arkansas.  As it translated northeastward, it weakened somewhat, but was still impressive (~2500 J/kg) when it reached southwest Ohio at 0900 UTC (see Fig. 2).  When assessing the anticipated shear against the most likely scenario regarding instability (helicity or positive shear values versus CAPE), the potential for strong to violent tornados were indicated (Johns and Doswell 1992; Johns et al. 1993).

3.  RADAR ANALYSIS

 Between 0600 and 0800 UTC on April 9th, a line of thunderstorms was observed by the KILN radar to be approaching the Weather Forecast Office (WFO) Wilmington, Ohio, (ILN) County Warning Area (CWA) from the west.  This convection was oriented quasi north-south along, or just ahead of the (near) surface-based instability axis which was positioned ahead of the approaching cold front.  By 0758 UTC, KILN reflectivity imagery indicated a broken line of convection extending from northeast Indiana, south-southwestward into north central Kentucky (Fig. 3).  Four concentrated areas of convection appeared to stand out along this line; three of which would directly impact ILN's CWA.  Below is a discussion surrounding the most intense area of convection as it impacted southeast Indiana and southwest Ohio.
 

Figure 3.  Wilmington, Ohio, WSR-88D (KILN) composite reflectivity valid at 0758 UTC on April 9th.  Color table depicts the intervals of reflectivity in units of dBZ.  Values < 25 dBZ have been filtered.
 
 
 

3.1    Southeast Indiana/Far Southwest Ohio

        The intense convection, which impacted southeast Indiana just prior to 0800 UTC, was responsible for producing the first tornado (T1) of the morning (between 0743 and 0818 UTC).  This F3 tornado tracked east-northeast and dissipated approximately 35 minutes after it first touched down.  Reflectivity imagery from three local WSR-88Ds [KILN, Louisville (KLVX in Fig. 3), and  Indianapolis (KIND in Fig. 3)] did not portray any of the more classic supercell signatures (hook, BWER/WER) associated with this tornadic thunderstorm.  Distance to this storm from these radars (60-80 nm) appeared to limit optimum sampling, as the lowest elevation scans from all three radars could not view that portion of the storm lying within 7-10 kft of the surface.  Note that storm tops were less than 30 kft AGL.
        Storm-relative velocity data displayed a moderate and rather deep circulation, (i.e., tight rotational velocity > 30 kts through at least 12 kft in depth) for much of the tornado's life time (Fig. 4).  This circulation also appeared to be situated along a pre-existing, quasi east-west oriented shear axis (not shown) which extended into southwest Ohio.  After 0813 UTC this tornadic thunderstorm slowly began to weaken and move to the southeast away from the influence of this shear axis.  Just prior to going sub-severe, this storm produced damaging (non-tornadic) winds in southeast Indiana (Fig. 1).

Figure 4.  Maximum storm-relative rotational velocities associated with tornados across southeast Indiana and southwest Ohio.  Dots (.) represent data points from KILN radar.  Stars (*) represent data points from KLVX and KIND radars.  Bold contour is 20 kt.  T1 through T5 denote tornados as discussed in the text.
 

        The role of this shear axis in assisting in the process of toronadogenesis is unclear.  For between 0835 and 0855 UTC, two additional tornados were observed; both occurring in close proximity to this shear axis.  As mentioned earlier in this section, it "appeared" that the lowest level rotational feature, which was once associated with the first tornado (T1), became disassociated from its initial parent cell.  Now in a weaker state, this rotational feature nonetheless seemed to continue its east-northeast movement along this axis.  By 0837 UTC, reflectivity imagery (not shown) indicated strong low-level return (50 dBZ) in southeast Indiana with negligible returns noted above ~12 kft AGL.  Coincident to this, the lowest level rotation was noted to have strengthened somewhat (20-25 kts), while rotation aloft was rather weak.  It was around this time that the second tornado (T2) occurred.  Whether it was solely low-level outflow interaction between separate convective cells from opposite ends of the county,  the juxtaposition of this "traveling" rotational feature with this low-level convective element, or both that aided in this process of tornadogenesis is unclear.
        The third tornado (T3) occurred in the town of Addyston, Ohio, around 0847 UTC.  The cause(s) leading to its development seemed to be the same as that for the second tornado (T2) which occurred just minutes earlier.  Both were weak (see Fig. 1).

3.2  Southwest Ohio

        It was over southwest Ohio, specifically over the northeast suburbs of Cincinnati, where the strong dynamic and thermodynamic forcing would be realized in the development of a violent F4 tornado (T4.)  The inception of this storm could be traced back to the approximate time when the Addyston tornado (T3) occurred.  It was here, just west-northwest of Cincinnati, that rotation in the vicinity of the shear axis aloft began to strengthen.  Reflectivity aloft was also observed to increase in intensity.  Between 0857 and 0902 UTC, a classic supercell had developed over the northern suburbs of Cincinnati.  Within the next 10 minutes, this storm would take on the structure of an HP supercell.  Shortly after this (~0912 UTC), tornado touchdown was observed.  However, soon after touch down, this supercell began to weaken.  But before dissipating, the supercell re-intensified for a short period to produce an F2 tornado (T5).  Touchdown for this tornado (T5) was about 15-20 nm northeast of Cincinnati at 0932 UTC.  This too occurred in proximity of the pre-existing shear axis.  All told, the total life span of this supercell was about 45 minutes.
        The KILN radar provided excellent remote sensing data for this event.  Rotational velocities increased to 20-25 kts to a depth of 15-18 kft AGL about 25 minutes before the tornado (T4) touched down (Fig. 4).  Further strengthening occurred 10-15 minutes later with 30-40 kts of rotation noted to a depth of 13-17 kft AGL.  The strongest rotation detected prior to touchdown (40-45 kts) was observed 5-8 kft AGL.  At the time of the tornado the WSR-88D Tornado Detection Algorithm (Mitchell et al. 1998) displayed a Tornadic Vortex Signature with Low Level Delta Velocity and Maximum Delta Velocity of 77 kts at 3.2 kft AGL.  For this storm, tornadogenesis, based on rotational velocities, would be characterized as descending or "archetypical" whereby a strong rotational couplet begins in the mid levels before descending to the ground as a tornado (Trapp et al. 1999).  Figure 5 shows SRM imagery for the 0.5 and 2.4  tilts at 0912 UTC, the time the tornado (T4) first touched down.
        From a reflectivity standpoint, unlike the tornadic storms which occurred earlier across Indiana and Ohio (T1 - T3), this supercell evidenced many, if not all, pertinent storm structure related signatures.  Note, the BWER at 0912 UTC in Fig. 5.  A pronounced rear flank downdraft (RFD) was noted almost from the supercells' inception, apparently driven by mid-level drying which helped accelerate the downdraft through evaporative cooling (inferred by tight moisture gradient aloft on southwest side of storm in Fig. 5).  The strengthening of this RFD would also act to enhance the storm-relative inflow.  Both of these processes most likely played a crucial role in helping to accelerate the transition of this storm into that of a dangerous supercell, as it moved across the northern suburbs of Cincinnati.  In addition, other parameters such as cell based VIL, Echo Top, and height of the maximum dBZ were also assessed.  Although a tendency for falling values around the time of touchdown was observed (not shown), no clear predictor among these parameters could be established.
 

Figure 5.  Wilmington, Ohio, WSR-88D storm-relative mean radial velocity map (top panels) and base reflectivity (bottom panels) at 0912 UTC on April 9th.  Elevation angles are 0.5 and 2.4  for the left and right panels, respectively.  Color table between the top panels depict the intervals of velocity in knots.  Storm motion subtracted from base radial velocities was 244 /51 kt.  Color table between the bottom panels similar to that in Fig. 3,
except values < 20 dBZ filtered.  Images centered ~25 nm southwest
of radar.  Circle identifies lower-level rotation.

4.  SUMMARY

        During the pre-dawn hours of April 9th a line of thunderstorms began to intensify as it moved into eastern Indiana.  Three concentrated areas of convection along this line would directly impact the WFO Wilmington, Ohio, CWA.  Strong dynamics, supported by both a dual upper-level jet configuration and a coupled low-level jet, would become simultaneously linked with an eastward moving moist unstable pre-frontal airmass over the Ohio Valley.  The resultant strong and deep upward vertical motion, along with critical shear values, would allow this convection to achieve severe limits, and ultimately culminate in the generation of seven tornados.  One tornado, which touched down in the northeast suburbs of Cincinnati, attained an F4 classification.  Four deaths were attributed to this storm.  For this paper, discussion was limited only to those storms impacting southeast Indiana and southwest Ohio, where five of the seven tornados were observed.
        The intense convection, which impacted southeast Indiana just prior to 0800 UTC, was responsible for producing the first tornado (F3) of the morning.  Although this storm displayed a moderate and rather deep circulation over time, no classic supercell signatures (hook, BWER/WER) were apparent from analysis of available reflectivity imagery.  It appeared that distance to this storm from adjacent WSR-88Ds limited optimum sampling in the lower levels.  Between 0835 and 0855 UTC, two additional tornados were observed (one in southeast Indiana, the other in extreme southwest Ohio).  Both occurred in close proximity to a quasi east-west oriented shear axis (storm-relative) that was also evident during the earlier F3 event. The cause for development of these two tornados is uncertain.  Possibilities include pure spin up along interacting outflow boundaries, or cell development in the vicinity of this shear axis, or both.
        It was, however, over southwest Ohio (in the vicinity of Cincinnati) where the strong dynamic and thermodynamic forcing would be realized in the development of a violent F4 tornado which occurred around 0912 UTC.  According to KILN WSR-88D data, this tornadic thunderstorm quickly took on the characteristics of a classic supercell as it moved across the northern suburbs of Cincinnati.  In the short time it then took to produce this killer tornado, the storm had transformed into an HP supercell.  Before dissipating, this supercell produced yet a second tornado (F2) approximately 15-20 nm northeast of Cincinnati.  As was the case with the earlier noted tornados, both of these latest tornados likewise occurred in close proximity to the observed shear axis.

5. ACKNOWLEDGMENTS

        The authors would like to thank Tom Salem of Eastern Region SSD for his thorough review of this paper, and Brian Coniglio, ILN forecaster, for his help in generating the graphics.

6.  REFERENCES

Johns, R. H. and C. A. Doswell, III, 1992: Severe Local Storms Forecasting.  Wea. Forecasting, 7,  588-612.
Johns, R. H., J. M. Davies, and P. W. Leftwich, 1993: Some Wind and Instability Parameters Associated with Strong and Violent Tornadoes. Part
            II.  Variation in the Combination of Wind  and Instability Parameters.  The Tornado:  Its Structure, Dynamics, Prediction, and Hazards,
            Geophys.  Monogr., No. 79, Amer. Geophys. Union, 573-590.
Lemon, L. R., 1980: Severe thunderstorm radar identification techniques and warning criteria: A preliminary report.  NOAA Tech. Memo. NWS NSSFC_1,
            60 pp. [NTIS PB273049]
Mitchell, D. E., S. V. Vasiloff, G. J. Stumpf, A. Witt, M. D. Eilts, J. T. Johnson, and K.W. Thomas, 1998: The National Severe Storms Laboratory Tornado
            Detection Algorithm.  Wea. Forecasting. 13, 352-366.
Moller, A. R., C. A. Doswell III, M. P. Foster, and G. R. Woodall, 1994: The Operational Recognition of Supercell Thunderstorm Environments and Storm
            Structures. Wea. Forecasting, 9, 327-347.
Trapp, R. J., E. D. Mitchell, G. A. Tipton, D. W. Effertz, A. I. Watson, D. L. Andra Jr., and M. A. Magsig, 1999: Descending and Nondescending Tornadic
            Vortex Signatures Detected by WSR-88Ds.  Wea. Forecasting, 14, 625-639.
Uccellini, L. W. and P. J. Kocin, 1987: The Interaction of Jet Streak Circulations during Heavy Snow Events along the East Coast of the United States. Wea.
            Forecasting, 2, 289-308.