Patrick J. Spoden*, Christopher N. Jones, James Keysor, and Mary Lamm

National Weather Service, Paducah, Kentucky

*Corresponding author address: Patrick J. Spoden, National Weather Service, 8250 US Highway 60, W. Paducah, KY 42086 e-mail w-pah.webmaster@noaa.gov



Radar operators have long known that bowing line segments are likely to produce damaging winds due to numerous studies on these types of systems (Fujita 1978; Przybylinski and Gery 1983; and others). Johns and Hirt (1983) defined a larger scale convective system utilizing Maddox’s 1980 definition of an extratropical mesoscale system as a derecho. In addition, Johns and Hirt defined a progressive derecho as, "A short curved squall line oriented nearly perpendicular to the mean wind direction with a bulge in the general direction of the mean flow." Przybylinski and DeCaire (1985) also assisted radar operators by defining several radar signatures associated with derechos. Additional studies (e.g., Przybylinski et al. 1996; and others) state that derechos are often composed of several mesoscale reflectivity characteristics such as rear-inflow notches (RINs) and "S" shapes. These characteristics have helped operators focus on areas where enhanced wind damage or tornadoes are likely. Weak tornadoes have been documented to form along the leading edge of the northern portion of the bow by Przybylinski and Schmocker (1993), Burgess and Smull (1990), Przybylinski et al. (1996), and Funk et al. (1996). Stronger (F2-F4 intensity) tornadoes have been documented by Wakimoto (1993) and Przybylinski (1988) in association with these types of mesoscale convective systems (MCS).

With the advent of the WSR-88D network, operators can not only survey the mesoscale reflectivity characteristics, they can also view the overall flow structure of these MCSs. Smull and Houze (1987) documented the MCSs flow structure of a convective bowing squall line and revealed the presence of two mesoscale storm-relative airflows. These airflows were identified as: 1) the descending mesoscale rear inflow jet (RIJ), and 2) ascending front-to-rear flow. Smull and Houze also noted that the RIJ influenced the shape, intensity, and propagation of the leading convective squall line. Numerical simulations by Weisman et al. (1990) and Weisman (1993) were able to re-create the typical flow structure associated with bow echoes.

This study will examine the Paducah, Kentucky (KPAH) WSR-88D data (archive II and archive IV) which revealed the mesoscale flow structure, bookend vortices, and circulations along the leading edge of an asymmetric derecho that moved rapidly across the Lower Ohio Valley on 5 May 1996. The derecho produced widespread wind damage with wind gusts to 41 m s-1. This derecho formed in an environment similar to that described as "warm season" by Johns (1993) and traveled along a quasi-stationary boundary nearly parallel to the mid-tropospheric flow. The mesoscale flow structure was found to be similar to that by Smull and Houze (1987) with an elevated RIJ. In this case, with an asymmetric derecho, the RIJ was deflected toward the northern portion of the line. One of the several interesting features associated with this derecho was the presence of two bookend vortices (coexisting for almost one hour) along the northern edge of the line. Four additional circulations developed along the leading edge of the northern portion of the line. These circulations were long-lived and satisfied mesocyclone criteria, one forming due in part to a convective scale downdraft. There were other short-lived circulations, but most existed for only one or two volume scans. There were no confirmed tornadoes during this event.


A composite map of 1200 UTC (times hereafter UTC) 5 May 1996 (Fig. 1) revealed a zonal flow pattern aloft with a surface quasi-stationary boundary evident from southeast Kansas to central Kentucky.


Fig1m5.bmp (599018 bytes)


Figure 1 – Composite map from 1200 UTC 5 May 1996 – All symbols are standard

The near zonal flow coincides with Johns and Hirt (1987), who found that derecho formation often occurs in a 500 mb flow from a direction of 240 or greater. At 300 mb, jet streaks of 55 m s-1 were noted over the Northern Plains states and Great Lakes and weak diffluence was occurring over the middle Mississippi River Valley (not shown).

Over Oklahoma, strong warm air advection was occurring at 850 mb ahead of a low centered in western Oklahoma. Thermal ridging (16 to 25C) extended into southeast Kansas/southwest Missouri near the derecho genesis region. Previous research (Johns et al. 1990), documented the tendency of derechos to move just to the north of the thermal ridge. A 15 m s-1 low-level jet (LLJ) was evident from Texas into Oklahoma, with moisture-pooling present near the nose of the LLJ (15C isodrosotherm).

The pre-convective environment, ahead of the surface low pressure, was moderately unstable. The sounding from Springfield, Missouri (KSGF) at 1200 revealed a Convective Available Potential Energy (CAPE) of 2104 J kg-1 and a lifted index (LI) of -6C, using a mean mixing layer parcel. Convective inhibition (CIN) for the sounding was -37 J kg-1. Using the most unstable parcel, the KSGF sounding for 1200 had a CAPE of 2530 J kg-1. The KSGF 1200 sounding was modified for the 1900 observed surface temperature and dew point at KPAH, and yielded a CAPE of 3205 J kg-1 and a LI of -7C.

The KSGF 1200 sounding indicated weak environmental shear (6 m s-1) in the surface to 2.5 km layer. However, the shear in the surface to 5 km layer was 17 m s-1. These magnitudes are weaker compared to Weisman’s (1993) simulation for long-lived bow echoes. The modified sounding depicted a higher CAPE value than the 2200 J kg-1 he determined was necessary to generate a long-lived bowing convective system. This is similar to recent case study findings of Burgess and Smull (1990), Przybylinski and Schmocker (1993), and Przybylinski et al. (1996).


At 1700, the KPAH WSR-88D revealed an asymmetric bowing reflectivity structure with an apex along or just south of the quasi-stationary boundary. The derecho was already in the stage 4 phase of an idealized bow echo described by Weisman (1993). Fig. 2 shows the progression of the leading edge of the derecho as it moved through the 230 km range of the KPAH radar. Reflectivity data from 1802 (not shown) showed the squall line extended from 75 km south of St. Louis to near the Missouri-Arkansas border. The overall length of the solid convective line was 144 km. A stratiform rain region trailed the northern part of the line with embedded areas of 30 dBZ, reflecting the melting level. Near the northern end of the line, a comma-shaped echo pattern was identified suggesting the probable existence of a bookend vortex. Stationary convection, 120 km east of the KPAH radar site, denoted the location of the quasi-stationary boundary which the apex of the derecho would follow as it moved through west Kentucky.

Fig2m5.bmp (90210 bytes)

Fig. 2. Map of derecho position every sixth volume scan ( 35 minutes). Times in UTC. Dashed lines indicate broken convective line segments. Origins of the bookend vortices (circulations #1 and #3) are denoted.

Through 1900 the overall reflectivity characteristics along the northern part of the line exhibited a solid line of convection. In contrast, individual cells were clearly observed along the southern part of the line. By 1906 the apex of the derecho was approximately 35 km west of KPAH (Fig. 3a). The reflectivity field showed two large comma-shaped echoes along the northern flank of the line suggesting the presence of two bookend vortices. The northernmost vortex was in the dissipation stage and was indicated by the lower reflectivity hole just northwest of the second, more dominant vortex. The second vortex persisted until 2003 (see section 4).

A cross-section through the apex of the squall line at 1906 revealed a multicellular nature to the convection (Fig. 3b). At least two higher reflectivity cores could be seen within the squall line. A large WER persisted along the leading edge of upright convection. The WER extended to a depth of 2.7 km. Areas of higher reflectivity (i.e., the melting level) extended rearward, centered around the 3 km level. A manually derived storm-relative cross-section (26 m s-1 speed subtracted) revealed two mesoscale airflows, front-to-rear flow and an opposing RIJ. A streamline analysis of the airflow within the derecho indicated the front-to-rear flow was lifted to the 6 km level ahead of the convection, and the flow remained above that level to the rear of the system (Fig. 3c). Storm relative velocities (SRM) in the front-to-rear flow were 0 to 7 m s-1 with local velocity maximums of 7 to 12 m s-1. The RIJ had SRM velocities of 7 m s-1 with local maximums of 18 m s-1 at a height of 0.6 km. There were two main radial convergent areas embedded within the front-to-rear flow, both located within the squall line. The first, along the leading edge of the WER, extended from 4.5 to 7.6 km. The second area was located on top of the WER, about 5.5 km upshear from the first, and extended from 8.4 to 10.7 km. These areas of radial convergence are similar to the Mid-Altitude Radial Convergence (MARC) velocity signature described by Schmocker et al. 1996.


Fig4m5.bmp (138654 bytes)

Fig5m5.bmp (138654 bytes)


Fig6m5.bmp (138654 bytes)

Fig. 3a, b, and c. WSR-88D reflectivity plan view at 0.5 elevation for 1906 UTC (a), vertical reflectivity cross-section constructed from 1906 UTC volume scan (b), and same as (b) except for storm relative velocities (c). Shaded area in (c) depicts reflectivities of 50 dBZ or greater.



A rotating comma-head reflectivity structure or mesolow (Smith 1990) is a common feature with derechos. Smith found that an east-west band of convection (warm advection wing) often develops north of the mesolow. This case was similar, in that an east-west band of convection developed around 1900. Modeling studies by Weisman (1990) showed a cyclonic flow pattern within the northern comma-head and defined this feature as a bookend vortex. Unlike previous studies, two long-lived cyclonic bookend vortices were identified with this asymmetric derecho. Fig. 4 shows the tracks of all six circulations which developed within a 230 km range of the KPAH WSR-88D. Circulations #1 and #3 were bookend vortices. In the following comparisons, magnitudes of rotational velocities (Vr) and core diameters were calculated by averaging throughout the column.

The first bookend vortex (Fig. 4, circulation #1) was detected by the KPAH WSR-88D at 1756. The initial diameter of 10.4 km fluctuated during its lifetime from as small as 6.1 km to as large as 10.9 km. Maximum Vr averaged 13 m s-1 throughout the lifetime of this bookend vortex. One report of severe weather was associated with this circulation, a wind gust of 24 m s-1. The gust occurred south of the vortex track, about 20 km north of Cape Girardeau, Missouri (KCGI). A close examination of the northern portion of the line beginning at 1744 (Fig. 4, circulation #1) showed a small bowing line segment near the center of the derecho with an apparent semi-occlusion (or outflow boundary) located just north of this segment. This feature was seen only in archive II data. By 1756 the convection along the warm frontal portion of the occlusion became the dominant line segment and the typical comma-head reflectivity pattern could be clearly identified through 1854. By 1906, the northern comma-head structure had dissipated.


Fig8m5.bmp (66526 bytes)


Fig. 4. Map of circulations #1 - #6 with starting and ending times in UTC. KCGI and KPAH represent Cape Girardeau, MO and Paducah, KY, respectively.

The second bookend vortex (Fig.4, circulation #3) developed around 1819, approximately 37 km due south of the first bookend vortex. These two bookend vortices existed in parallel from 1819 through 1906. Circulation #3 continued until 2003. At 1819 the initial diameter of this vortex (Fig. 5a) was 5.2 km and rapidly increased to larger than 9.0 km by 1831. The diameter never decreased below 10 km after 1837 and obtained a diameter of 17.8 km at 1906. Rotational velocities (Fig. 5b) were much stronger with the second vortex reaching a peak of > 23 m s-1 at 1912, when the diameter was 17.4 km. Damaging wind reports from 15 km west of KCGI to 60 km northeast of KCGI appeared to be associated with the enhanced rear-inflow south of circulation #3. This swath of high winds is reflected by the 34 m s-1 wind gust recorded at KCGI at 1845. The reflectivity structure seen in Fig. 2 also shows the development of a second comma-shaped echo at 1819. Similar to the first bookend vortex (circulation #1), a semi-occlusion (T-bone reflectivity structure) developed at 1819 north of a small bowing line segment. The warm frontal portion of the occlusion became the main convective line by 1831 with the comma-head structure again visible. As with the development of circulation #1, the semi-occlusion was only seen in archive II data. The comma-head moved northeast along the derecho and became elongated (northwest-southeast) by 1928. The elongated nature of the comma-head could be seen until 1951. Both the circulation and the associated comma-head structure were no longer visible after the 2003 volume scan.

Fig9m5.bmp (842302 bytes)

Fig10m.bmp (799262 bytes)

Fig. 5a and b. Time-height series plot of circulation #3 core diameter in km (a), and rotational velocities in m s-1 (b).


Table 1 summarizes the life history, intensity of shear, and core diameter of the other vortices associated with this derecho. Circulation #2 appeared to exhibit the weakest Vr of all circulations documented in this event. Circulation #5 which developed along the northern portion of the apex of the squall line revealed the strongest Vr and largest average diameter of this group of vortices. Several unconfirmed tornado reports and the existence of enhanced damage were docmented along the path of this circulation.

Another interesting evolution observed in this case was the relationship of circulation #6 and a small local velocity maxima. The local inbound velocity maxima, nearly parallel to the radial, was associated with high reflectivity cores embedded within the convective line. There were no observed rear-inflow velocity maximums in the SRM data. Thus, the local maxima detected within the highly reflective core region was a reflection of a convectively induced downdraft rather than a component of the RIJ. At 1843, the local inbound velocity maxima of 23 m s-1 impinged upon the leading convective line edge of the derecho resulting in the rapid formation of circulation #6. This vortex moved closest to the KPAH WSR-88D, and was tracked throughout much of its lifetime. Numerous storm reports of enhanced damage clearly marked the path of this circulation.



Circ #2


Circ. #4

Circ. #5

Circ. #6

Development (UTC)





Dissipated (UTC)





Max. Diamter (km)





Mn.. Diameter





Ave. Diameter





Max. Vr

( m s-1)





Table 1. Characteristics of circulations #2, #4, #5, and #6.


The asymmetric derecho which progressed through the Lower Ohio Valley on 5 May 1996 possessed several similarities with previous studies of this type of MCS. This study documented the presence of an elevated RIJ and front-to-rear flow as the two main airflows within the derecho. Bookend vortices, located at the northern end of the complex, assisted in focusing the RIJ and aided in the development of at least one circulation located along the leading edge of a convective line. Six different circulations were documented along the northern half of the derecho. Circulations #1 and #3 were already, or became, bookend vortices and were the largest in diameter throughout most of their life. The initial average diameter of circulation #1 was 10.4 km and maximized at 10.9 km, while circulation #3 had an initial diameter of 5.2 km and grew to a maximum size of 17.8 km. The other circulations had initial diameters of less than 4.0 km and never grew larger than 10.0 km. This suggests that circulations #1 and #3 were already bookend vortices, starting out large and remaining large (in diameter) throughout their lifetime. The operational forecaster can perhaps benefit from these findings by first identifying the larger circulations, and then look for an enhanced mesoscale RIJ south of the vortex. Such enhanced rear-inflow can result in increased bowing of the line and a higher intensity of damaging winds.


The authors would like to thank Ron Przybylinski and Gary Schmocker of NWSFO LSX, and Dr. Frank Lin of St. Louis University for their guidance and thorough review of the manuscript. Additional thanks are extended to Beverly Poole, MIC and the rest of the NWSO PAH staff for their unwavering support in this research. This paper was funded from a subaward (S97-86991) under a cooperative agreement between NOAA and the University Corporation for Atmospheric Research (UCAR). The views expressed herein are those of the authors and do not necessarily reflect the views of NOAA, its subagencies, or UCAR.


Available upon request

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