Theodore W. Funk and Van L. DeWald


Yeong-Jer Lin

National Weather Service Forecast Office
6201 Theiler Lane
Louisville, Kentucky 40229


Dept. Of Earth and Atmospheric Sciences
Saint Louis University
St. Louis, Missouri 63103


During the early morning of 14 May 1995, a non-severe cluster of convection evolved in less than one hour into an intense bow echo that then raced eastward cutting a swath of major wind damage across north-central Kentucky. Widespread straight-line winds over 60 mph with maximum reported gusts around 100 mph, hail, and at least two tornadoes of F0-F2 intensity downed many trees and power lines, and damaged or destroyed numerous buildings and structures throughout several counties just south of Louisville, Kentucky. Over 10 million dollars of damage was reported in one county alone.

Significant mid-altitude radial convergence (MARC; Przybylinski et al. 1995; Schmocker et al. 1996) resulted in downburst activity that contributed to rapid convective intensification, bow echo development (Przybylinski 1995; Weisman 1993), and the onset of damaging surface winds along the leading convective line. The apex of the bow passed within 15-35 km of the Louisville-Ft. Knox (KLVX) WSR- 88D radar site (RDA). As part of a COMET Cooperative Research Project, this study uses KLVX radar data to examine the rapid formation and evolution of the intense bow structure, including significant MARC and subsequent development of cyclonic circulations near the bow apex that resulted in tornadoes embedded within the damage swath.


The 0000 UTC 14 May synoptic environment was characterized by strong forcing, similar to the "dynamic" bow echo pattern described by Johns (1993). Thunderstorms initiated ahead of a migrating low pressure system and cold front, as a 15-20 m/s low-level jet advected high equivalent potential temperature (theta-e) air into a position ahead of the front (Fig. 1). In addition, upper-level divergence existed to the right of a 35-40 m/s jet core at 250 mb. Meanwhile, the VAD Wind Profile from the KLVX WSR-88D at 0548 UTC indicated 20-25 m/s total shear in the 0-2 km layer, with near constant winds above (Fig. 1), a profile typically associated with bow echo occurrence (Johns 1993).

Composite Synoptic Chart at 0000 UTC 14 May 1995 Fig. 1: Composite synoptic chart at 0000 UTC 14 May. Thicker (thinner) arrow is the 250 (850) mb jet, while dashed lines are 850 mb theta-e (deg K). Also shown are surface fronts, 500 mb heights (in dm; thin solid lines), MCS location at 0600 UTC (shaded area), and the 0548 UTC vertical wind profile (VWP; in kts and 1000s of ft) from the KLVX (Louisville) WSR-88D.

A moderate-to-very unstable environment existed across the Ohio Valley at 0000 UTC 14 May. Sounding data from Nashville, Tennessee (not shown) indicated a lifted index of -6 to -8 and a surface-based CAPE of nearly 4000 J/kg. Low-level moisture was plentiful, with surface dewpoints in Kentucky in the lower to middle 70s and precipitable water values around 1.5 inches. The Nashville sounding also exhibited a significant dry intrusion above 750 mb (resulting in a K Index of 19), which could act to strengthen convective downdrafts through evaporative cooling of entrained environmental air. Moreover, the sounding revealed a 0-3 km storm-relative helicity of nearly 330 m2/s2, a value associated with possible mesocyclone and tornado development (Davies-Jones et al. 1990).


KLVX WSR-88D base reflectivity data (0.5 elevation) at 0538 UTC 14 May (Fig. 2) showed a cluster of convection in existence along the Ohio River 75-110 km west-southwest of the RDA. Within the cluster, two areas of highest cloud tops and reflectivity values (denoted by "A" and "B") were located near the center of the MCS, with lower values near the leading edge ("C"). At this time, heavy rain was occurring, although no surface wind damage was reported. However, by 0603 UTC (25 minutes later), convective area "A" began to pivot northeastward as the reflectivity gradient ("C") was intensifying along the leading edge of dissipating convective cluster "B" (Fig. 2). Increased convergence resulted in a continued rapid intensification of convection along the leading line by 0618 UTC (15 minutes later), which produced a mature, severe bow echo with an intense leading reflectivity gradient at "C." Very damaging downburst surface winds were in progress along the bow apex, similar to findings by Przybylinski (1995) among others. The bow echo raced eastward at speeds of 55 kts during its mature stage producing major wind damage. A tight leading reflectivity gradient still was evident at 0633 UTC (Fig. 2), while a weak echo channel (WEC) behind the leading line denoted the location of strong low-level rear-to-front flow (not shown). For example, base radial velocity values within the rear inflow jet peaked at over 46 m/s (90 kts) at 1300 ft elevation immediately behind the mature bow apex. By 0658 UTC, a WEC again was evident with wind damage persisting along the apex. The bow and its intense leading reflectivity gradient began to weaken by 0718 UTC (Fig. 2) and thereafter, with a corresponding gradual decrease in the severity of surface wind damage.

WSR-88D Reflectivity Image at 0538 UTC 14 May 1995 WSR-88D Reflectivity Image at 0603 UTC 14 May 1995
WSR-88D Reflectivity Image at 0618 UTC 14 May 1995 WSR-88D Reflectivity Image at 0633 UTC 14 May 1995
WSR-88D Reflectivity Image at 0658 UTC 14 May 1995 WSR-88D Reflectivity Image at 0718 UTC 14 May 1995
Fig. 2: Low-level base reflectivity data (0.5 deg elevation) from the KLVX WSR-88D showing evolution of an MCS into an intense bow echo. Contours are at 10 dBZ intervals starting at 20 dBZ; values over 40 (50) dBZ are shaded light (dark). "A" and "B" denote two convective areas within the MCS, while "C" marks the leading edge of the MCS. "C1", "C2", and "C3" show the location of 3 cyclonic circulations. "WEC" denoted a weak echo channel behind the leading line.

During the period of bow echo intensification and maturity, three cyclonic circulations (C1, C2, and C3 in Figs. 2 and 3) were noted along the bow. Circulations 1 and 2 produced F1-F2 tornadoes during their mature stages while Circulation 3 was non-tornadic.

Identified Cyclonic Circulation Tracks on 14 May 1995 Fig. 3: Identified cyclonic circulation ("Circ") tracks. Small black dots along tracks are 5-min interval circulation positions. Small tick marks along tracks denote the occurrence of tornadoes ("T"). "W" represents locations of wind damage, while "mw" is where maximum wind damage occurred. The radar site (RDA) is shown by the circled black dot in northern Hardin county. 



Various researchers have studied convergence zones within convective storms to determine their effect on severe weather phenomena. For example, Lemon and Parker (1996) and Lemon and Burgess (1992) documented the existence of a "deep convergence zone" within supercell storms which extended to an average vertical depth of 10 km and was associated with mesocyclones, tornadoes, and damaging winds. Eilts et al. (1996) researched over 85 microburst-producing pulse thunderstorms and determined that deep convergence in the storms' middle-levels was one of the most effective radar precursors to surface wind damage. Przybylinski et al. (1995) studied the mid-altitude radial convergence (MARC) signature for its utility in predicting the onset of damaging surface winds associated with squall lines and incipient bow echoes. MARC is defined as the "delta V" or difference between the maximum inbound and outbound velocity values within 6 km along a radial. Their results suggested that MARC velocity values of 25 m/s (50 kts) or more at an altitude of 3 to 7 km preceded the onset of damaging surface winds by up to 20 minutes. Schmocker et al. (1996) substantiated these results for several MCSs over the middle Mississippi Valley and added that the average vertical depth of MARC was 6.2 km with the level of maximum MARC between 5 and 6 km.

In the 14 May case, a well-defined MARC signature with excellent temporal continuity (Fig. 4) was observed in storm-relative velocity (SRM) data prior to bow echo formation and the onset of damaging surface winds. A significant convergence signal was noted at 0523 UTC, or approximately 30-35 minutes prior to bow echo formation and the incipient wind damage report in Breckinridge county (Figs. 3 and 4). Maximum observed MARC values of nearly 38 m/s (75 kts) occurred at 0538 and 0543 UTC at an altitude of 4.5-5.5 km (Fig. 4), or 10-20 minutes prior to the onset of reported damage. Maximum wind damage from 0613-0633 UTC (wind gusts around 100 mph in Hardin county; "mw" in Fig. 3) from the mature bow echo occurred about 30-40 minutes after the period of maximum MARC (0538-0543 UTC; Fig. 4). SRM data at 0538 UTC (Fig. 5) shows that the greatest MARC at 5.2 km altitude was coincident with the area of highest reflectivity values within convective cluster "B" (Figs. 2 and 5). During the period of strongest MARC (0533-0548 UTC), convergence was noted throughout a vertical depth of 6-8 km (Fig. 4). In addition, the distance between maximum inbound and outbound radial velocities (i.e., MARC width) consistently ranged from only 1-3 km, well less than the 6 km maximum Schmocker et al. (1996) cited as significant in order to promote surface wind damage. MARC values still were noteworthy but gradually decreased after 0548 UTC (Fig. 4) as subsequent downburst winds contributed to development of the bow echo. It should be noted that base velocity data (not shown) also revealed the presence of low-to-mid-level rear-to-front flow near cluster "B" in the MCS which also may have contributed to bow echo formation. However, this airstream was present within the MCS well prior to bow formation, i.e., when highest cloud tops were in the center of the system. Thus, MARC, with its consistently strong signal, certainly appeared to play a vital role in promoting actual bow echo formation and severe wind damage. By 0603 UTC, observed MARC values continued to decrease slowly as the original convection (cluster "B" in Fig. 2) weakened and the leading bow echo strengthened along its leading edge (area "C" in Fig. 2).

Magnitudes of Mid-Altitude Radial Convergence (MARC) from 0523 to 0603 UTC 14 May 1995 Fig. 4: Magnitudes of MARC (in kts) within convective area "B" in Fig. 2. Time and depth (altitude) are shown. The "W" and adjacent line reveal the period of damaging surface winds.
WSR-88D Mid-Level Reflectivity Image at 0538 UTC 14 May 1995 WSR-88D Mid-Level Storm-Relative Velocity Image at 0538 UTC 14 May Fig. 5: Base reflectivity (far left) and storm-relative velocity (SRM) data (near left) at 0538 UTC (5.2 km altitude). Reflectivity contouring and shading are the same as in Fig. 2. Convective area "B" is shown. In SRM data, solid (dashed) lines are velocities (in kts) directed away (toward) the radar site (RDA) located to the east-northeast of the MARC signature.

Vertically Integrated Liquid (VIL) values (not shown) from cluster "B" (Fig. 2) were highest (65 kg/m2) at 0523 UTC, i.e., as MARC was beginning to increase (Fig. 4). However, as MARC increased and peaked by 0538 and 0543 UTC, VIL values lowered. After 0543 UTC, both MARC and VIL decreased, as cluster "B" weakened and the bow echo formed. VIL values then increased downwind as convection matured within the bow echo. Thus, weakening VIL values were deceptive in this case, and not indicative that subsequent surface wind damage was imminent. Nevertheless, such VIL trends in conjunction with the presence of MARC should alarm meteorologists that collapsing convection may result in damaging downburst winds given favorable environmental conditions.


Cyclonic circulation 1 (C1; Fig. 3) first was detected as a weak vortex below 3 km altitude at 0603 UTC (Fig. 6). The vortex quickly deepened and strengthened thereafter, and met mesocyclone rotational velocity (Vr) criteria. Maximum Vr values reached over 21 m/s (40 kts) in the lowest 3 km of the atmosphere. C1 was strongest at 0613 and 0618 UTC in the lowest radar elevation angle, i.e., at 0.5 km altitude 15-20 km from the RDA. It possessed a tight 1 km diameter. During its mature stage, the vortex produced an F1 tornado that was located near the intersection of an east-west area of convection (cluster "A") with the northern portion of the bow echo (Fig. 2). The development of C1 appears similar to that documented by Wakimoto and Wilson (1989), i.e., vortex genesis occurred along a low- level convergence boundary between the east-west line and the bow echo, then subsequent vertical stretching within the convective updraft resulted in vortex strengthening and deepening. In addition, substantial observed storm-relative low-level flow into C1 contributed to its rapid development. C1 gradually dissipated after 0618 UTC. Note that little or no SRM data or Vr values were available at 0623 and 0628 UTC (Fig. 6). Conversely, data was available at 0633 UTC although only weak shear remained below 1 km depth.

Magnitudes of Rotational Velocity for Circulation 1 Fig. 6: Magnitudes of rotational velocity (Vr in kts) for Circulation 1 (C1). The "T" and adjacent line show the period of a tornado associated with the circulation.

As C1 reached its peak, Circulation 2 (C2; Fig. 3) initially was detected at 0613 UTC in the low-to-mid-levels along the leading bow apex (Fig. 7). C2 then matured into a deep-layered circulation/mesocyclone, with rapid strengthening at and below 1 km altitude where highest Vr values reached 21-23 m/s (40-45 kts) from 0633 to 0643 UTC. The boundary layer vortex intensification correlated strongly with tornadogenesis at about 0625 UTC. A storm survey revealed that the F1-F2 tornado remained on the ground through about 0646 UTC, when vortex dissipation commenced. From 0633-0643 UTC, C2's diameter was about 0.5 km at 0.5 km altitude 15-20 km from the RDA. Thus, it appears the radar may have been sensing at least a portion of the actual tornadic circulation. The tornado dissipated around 0646 UTC as low-level rotation weakened rapidly (Fig. 7). During its mature stage, C2 was located just north of the bow apex within the rain shield and on the northern fringe of the maximum straight-line wind damage associated with the strong rear inflow jet.

Magnitudes of Rotational Velocity for Circulation 2 Fig. 7: Same as Fig. 6, except magnitudes of rotational velocity (Vr) for Circulation 2 (C2).

Finally, a third cyclonic circulation (C3; Fig. 3) was noted along the bow apex as C2 began weakening. C3 was non-tornadic but associated with wind damage in Nelson county (Fig. 3). Each of the 3 observed circulations in this case developed and matured just south (on the right flank) of its predecessor, similar to findings by Funk et al. (1996) and DeWald et al. (1998) for organized bowing segments within squall lines across Kentucky.


The 14 May 1995 intense bow echo over north-central Kentucky developed rapidly from a non-severe MCS in response to strong mid-altitude radial convergence (MARC) and some rear inflow. MARC correlated highly with subsequent bow echo development and the onset of damaging winds. The onset of significant MARC preceded the first surface damage report by about 30-35 minutes, while maximum MARC values were detected 10-20 minutes before the initial damage. MARC showed a greater correlation to severe wind gusts than did VIL since VIL is more useful in hail situations. After bow echo formation, 3 cyclonic circulations were detected along or near the leading bow, 2 of which met mesocyclone Vr criteria and produced F1-F2 tornadoes within or near the axis of severe straight-line wind damage.

This case and previous studies suggest that in conjunction with other radar data, MARC may be a viable velocity signature that meteorologists can use to help assess the potential for subsequent surface wind damage within MCSs. However, MARC is limited when velocities within mid-level convergence zones are oriented normal to the radar beam and thus greatly underestimated or masked. Once an organized bow echo forms, meteorologists should monitor radar data closely for tornado development along or just north of the bow apex given favorable environmental or convectively-induced wind shear.


This paper was funded from a sub award under a cooperative agreement between NOAA and the University Corporation for Atmospheric Research (UCAR).


Davies-Jones, R.P., D. Burgess, and M. Foster, 1990: Test of helicity as a tornado forecast parameter. Preprints, 16th Conf. on Severe Local Storms, Kananaskis Park, Alberta, Canada, Amer. Meteor. Soc., 588-592.

DeWald, V.L., T.W. Funk, J.D. Kirkpatrick, and Y.-J. Lin, 1998: The 18 May 1995 squall line over south-central Kentucky: An examination of complex storm reflectivity trends and multiple mesocyclone development. Preprints, 16th Conf. on Weather Analysis and Forecasting, Phoenix, Amer. Meteor. Soc., 148-151.

Funk, T.W., K.E. Darmofal, J.D. Kirkpatrick, M.T. Shields, R.W. Przybylinski, Y.-J. Lin, G.K. Schmocker, and T.J. Shea, 1996: Storm reflectivity and mesocyclone evolution associated with the 15 April 1994 derecho, Part II: Storm structure and evolution over Kentucky and southern Indiana. Preprints, 18th Conf. on Severe Local Storms, San Francisco, Amer. Meteor. Soc., 516-520.

Johns, R.H., 1993: Meteorological conditions associated with bow echo development in convective storms. Wea. Forecasting, 8, 294-299.

Lemon, L.R. and D.W. Burgess, 1992: Supercell associated deep convergence zone revealed by a WSR-88D. Preprints, 26th Conf. on Radar Meteor., Norman, Amer. Meteor. Soc., 206-208.

____, and S. Parker, 1996: The Lahoma storm deep convergence zone: Its characteristics and role in storm dynamics and severity. Preprints, 18th Conf. on Severe Local Storms, San Francisco, Amer. Meteor. Soc., 70-75.

Przybylinski, R.W. 1995: The bow echo: Observations, numerical simulation, and severe weather detection methods. Wea. Forecasting, 10, 203-218.

____, Y.-J, Lin, G.K. Schmocker, and T.J. Shea, 1995: The use of real-time WSR-88D, profiler, and conventional data sets in forecasting a northeastward moving derecho over eastern Missouri and central Illinois. Preprints, 14th Conf. on Wea. Analysis and Forecasting, Dallas, Amer. Meteor. Soc., 335-342.

Schmocker, G.K., R.W. Przybylinski, and Y.-J. Lin, 1996: Forecasting the initial onset of damaging downburst winds associated with a mesoscale convective system (MCS) using the mid-altitude radial convergence (MARC) signature. Preprints, 15th Conf. on Wea. Analysis and Forecasting, Norfolk, Amer. Meteor. Soc., 306-311.

Wakimoto, R.M. and J.W. Wilson, 1989: Non-supercell tornadoes. Mon. Wea. Rev., 117, 1113-1140.

Weisman, M.L., 1993: The genesis of severe, long-lived bow echoes. J. Atmos. Sci., 50, 645-670.

Back to Research and Papers

USA.gov is the U.S. government's official web portal to all federal, state and local government web resources and services.