12
ORIGINAL PAPER Lightning characteristics of derecho producing mesoscale convective systems Mace L. Bentley 1 John R. Franks 1 Katelyn R. Suranovic 2 Brent Barbachem 3 Declan Cannon 4 Stonie R. Cooper 5 Received: 17 July 2015 / Accepted: 2 November 2015 Ó Springer-Verlag Wien 2015 Abstract Derechos, or widespread, convectively induced wind storms, are a common warm season phenomenon in the Central and Eastern United States. These damaging and severe weather events are known to sweep quickly across large spatial regions of more than 400 km and produce wind speeds exceeding 121 km h -1 . Although extensive research concerning derechos and their parent mesoscale convective systems already exists, there have been few investigations of the spatial and temporal distribution of associated cloud-to-ground lightning with these events. This study analyzes twenty warm season (May through August) derecho events between 2003 and 2013 in an effort to discern their lightning characteristics. Data used in the study included cloud-to-ground flash data derived from the National Lightning Detection Network, WSR-88D imagery from the University Corporation for Atmospheric Research, and damaging wind report data obtained from the Storm Prediction Center. A spatial and temporal anal- ysis was conducted by incorporating these data into a geographic information system to determine the distribu- tion and lightning characteristics of the environments of derecho producing mesoscale convective systems. Primary foci of this research include: (1) finding the approximate size of the lightning activity region for individual and combined event(s); (2) determining the intensity of each event by examining the density and polarity of lightning flashes; (3) locating areas of highest lightning flash density; and (4) to provide a lightning spatial analysis that outlines the temporal and spatial distribution of flash activity for particularly strong derecho producing thunderstorm episodes. 1 Background The National Lightning Detection Network (NLDN) began recording occurrences of cloud-to-ground lightning strikes in 1989, and the scientific community has since expanded its understanding of the frequency and behavior of cloud- to-ground (CG) lightning associated with a plethora of severe and non-severe convective phenomena. However, we still know relatively little about the characteristics of CG lightning emanating from derecho producing mesos- cale convective systems (hereafter, DMCSs; Holle 2014). A DMCS is a convectively induced windstorm emanating from rapidly moving thunderstorms variously known as bow echoes, squall lines, or quasi-linear convective sys- tems (Przybylinski 1995; Johns and Hirt 1987; Fujita and Wakimoto 1981). These fast-moving, powerful storms have the capability to emit thousands of lightning strikes over a vast region in a relatively short amount of time. DMCSs are most common east of the Rocky Mountains and during the warm season months of May through Responsible Editor: M. Kaplan. & Mace L. Bentley [email protected] 1 Department of Integrated Science and Technology, James Madison University, Harrisonburg, VA, USA 2 Department of Geography, The George Washington University, Washington, DC, USA 3 Department of Modeling Simulation and Visualization Engineering, Old Dominion University, Norfolk, VA, USA 4 Aviation Weather Center, National Weather Service, Kansas City, MO, USA 5 Nebraska Mesonet, School of Natural Resources, University of Nebraska-Lincoln, Lincoln, NE, USA 123 Meteorol Atmos Phys DOI 10.1007/s00703-015-0417-x

Lightning characteristics of derecho producing mesoscale convective systems

  • Upload
    jmu

  • View
    0

  • Download
    0

Embed Size (px)

Citation preview

ORIGINAL PAPER

Lightning characteristics of derecho producing mesoscaleconvective systems

Mace L. Bentley1 • John R. Franks1 • Katelyn R. Suranovic2 • Brent Barbachem3•

Declan Cannon4 • Stonie R. Cooper5

Received: 17 July 2015 / Accepted: 2 November 2015

� Springer-Verlag Wien 2015

Abstract Derechos, or widespread, convectively induced

wind storms, are a common warm season phenomenon in

the Central and Eastern United States. These damaging and

severe weather events are known to sweep quickly across

large spatial regions of more than 400 km and produce

wind speeds exceeding 121 km h-1. Although extensive

research concerning derechos and their parent mesoscale

convective systems already exists, there have been few

investigations of the spatial and temporal distribution of

associated cloud-to-ground lightning with these events.

This study analyzes twenty warm season (May through

August) derecho events between 2003 and 2013 in an effort

to discern their lightning characteristics. Data used in the

study included cloud-to-ground flash data derived from the

National Lightning Detection Network, WSR-88D imagery

from the University Corporation for Atmospheric

Research, and damaging wind report data obtained from

the Storm Prediction Center. A spatial and temporal anal-

ysis was conducted by incorporating these data into a

geographic information system to determine the distribu-

tion and lightning characteristics of the environments of

derecho producing mesoscale convective systems. Primary

foci of this research include: (1) finding the approximate

size of the lightning activity region for individual and

combined event(s); (2) determining the intensity of each

event by examining the density and polarity of lightning

flashes; (3) locating areas of highest lightning flash density;

and (4) to provide a lightning spatial analysis that outlines

the temporal and spatial distribution of flash activity for

particularly strong derecho producing thunderstorm

episodes.

1 Background

The National Lightning Detection Network (NLDN) began

recording occurrences of cloud-to-ground lightning strikes

in 1989, and the scientific community has since expanded

its understanding of the frequency and behavior of cloud-

to-ground (CG) lightning associated with a plethora of

severe and non-severe convective phenomena. However,

we still know relatively little about the characteristics of

CG lightning emanating from derecho producing mesos-

cale convective systems (hereafter, DMCSs; Holle 2014).

A DMCS is a convectively induced windstorm emanating

from rapidly moving thunderstorms variously known as

bow echoes, squall lines, or quasi-linear convective sys-

tems (Przybylinski 1995; Johns and Hirt 1987; Fujita and

Wakimoto 1981). These fast-moving, powerful storms

have the capability to emit thousands of lightning strikes

over a vast region in a relatively short amount of time.

DMCSs are most common east of the Rocky Mountains

and during the warm season months of May through

Responsible Editor: M. Kaplan.

& Mace L. Bentley

[email protected]

1 Department of Integrated Science and Technology, James

Madison University, Harrisonburg, VA, USA

2 Department of Geography, The George Washington

University, Washington, DC, USA

3 Department of Modeling Simulation and Visualization

Engineering, Old Dominion University, Norfolk, VA, USA

4 Aviation Weather Center, National Weather Service,

Kansas City, MO, USA

5 Nebraska Mesonet, School of Natural Resources, University

of Nebraska-Lincoln, Lincoln, NE, USA

123

Meteorol Atmos Phys

DOI 10.1007/s00703-015-0417-x

August, with July being the peak month of occurrence

(Johns and Hirt 1987; Bentley and Mote 1998; Bentley and

Sparks 2003; Coniglio and Stensrud 2004).

A defining feature of the DMCS is the bow echo, which

forms either from a large, organized cluster of thunder-

storms or a single, very strong storm (Przybylinski 1995).

Cool, dense air spreads outward along the gust front, the

region where the bow echo forms (and the feature that

gives the storm its signature radar appearance). Several

internal mechanisms including an elevated rear-inflow jet,

mesohigh, and wake-low develop during organization

leading to a self-propagating DMCS (Weisman 1993;

Rotunno et al. 1988).

Derechos present a significant hazard to humans and

especially urban and suburban infrastructure. The 29 and

30 June 2012 Midwest and Mid-Atlantic derecho exem-

plifies a particularly intense DMCS. The bow echo formed

on June 29th over northern Indiana and traveled east across

several states and the District of Columbia before moving

over the Atlantic nearly 11 h later. This derecho was one of

the most destructive and deadly severe thunderstorm

complexes in North American history. A large swath of

damaging winds, torrential rain, and frequent cloud-to-

ground lightning resulted in uprooted trees, downed power

lines, and roofs and walls ripped off of their facades—

causing several hundred million dollars in damages. The

storm claimed 24 lives and millions along the east coast

went without power for several days, and even weeks. This

particular event produced nearly 70,000 CG flashes over its

ten ? hour time span (Holle 2014).

Given the propensity for DMCS episodes to produce

high CG flash rates, evidence suggests that these events

may be a significant contributor to the overall lightning

climatology east of the Rocky Mountains (Holle 2014).

Warm season mesoscale convective systems (MCSs) are

known to be exceptional lightning producers (Rutledge and

MacGorman 1988; Holle et al. 1994). Warm season

DMCSs often form in environments conducive to the

development of extensive CG lightning including very

warm, humid air at the surface and generous amounts of

convective available potential energy (CAPE; Johns and

Hirt 1987; Bentley and Mote 1998; Evans and Doswell

2001; Coniglio and Stensrud 2004). Current theories on the

formation of lightning focus on the separation of electric

charges and generation of an electric field within deep,

moist convection (Rakov and Uman 2003). Evidence sug-

gests that ice, hail and graupel are collocated in the

cumulonimbus to optimize charge separation and subse-

quent lightning formation (Rakov and Uman 2003). The

synoptic environment associated with warm season, pro-

gressive derechos is particularly favorable for graupel

formation given the availability of low-level moisture,

updraft strength promoted by high CAPE and the

development of a rear-inflow jet that further enhances

charge separation (Bentley et al. 2000; Coniglio and

Stensrud 2004; Weisman et al. 2013). The updrafts trans-

port small water droplets many kilometers above the

freezing level. Simultaneously, downdrafts transport hail

and ice from the upper frozen reaches of the thunderstorm.

Where these branches interface, the water droplets freeze

and release heat keeping the surface of the hail and ice

particles slightly warmer than the ambient environment and

create graupel or ‘‘soft hail’’ (Rakov and Uman 2003).

When the graupel collides with additional water and ice,

negatively charged electrons are sheared off of the

ascending particles and accumulate on the descending ones

leading to a negatively charged cloud base and positively

charged anvil (Rakov and Uman 2003). Evidence suggests

that the liquid water content of the cumulonimbus can lead

to more graupel; likewise, the presence of a rear-inflow jet

can increase the collision zone leading to enhanced charge

separation, with both processes leading to a greater

potential for lightning (Rakov and Uman 2003).

This investigation examines CG flash characteristics of

20 DMCSs that occurred throughout the Central and

Eastern United States from 2003 through 2013. Nearly two

million flashes were incorporated into the dataset to detail

the spatial and temporal distributions of these events.

2 Data and methods

Four primary data sources were utilized to perform the

analyses:

• Cloud-to-ground flashes and associated metrics

acquired from the National Lightning Detection Net-

work (NLDN) owned and maintained by Vaisala. The

NLDN data were archived from 2003 through 2013 for

the contiguous United States and obtained from the

Internet Data Distribution (IDD) from Unidata. The

sensitivity of the NLDN sensors allows it to detect CG

flashes as well as some cloud strokes (Cummins et al.

1998). Therefore, to eliminate non-CG flashes, all

flashes between 0 and 15 kA were removed (Cummins

et al. 1998; Cummins and Murphy 2009). The flash

metrics contained in the NLDN dataset include loca-

tion, time, peak current, polarity and multiplicity.

Improvements to the NLDN have occurred twice in

the span of this investigation, once in the early 2000s

and another in 2013 (Cummins et al. 2006; Holle 2014).

Flash detection is approximately 90–95 % and median

locational accuracy of at least 500 m (Holle 2014).

• Radar archive of WSR-88D imagery of DMCSs. These

data were acquired from the University Corporation for

Atmospheric Research (UCAR) image archive (http://

M. L. Bentley et al.

123

www2.mmm.ucar.edu/imagearchive/). Base reflectivity

was utilized to track the movement, intensity, and size

of the DMCS and to identify radar signatures associated

with severe MCSs such as bow echoes, rear-inflow

notches, bookend vortices and steep leading edge

reflectivity gradients.

• The Storm Prediction Center’s (SPC) online archive of

damaging wind reports (http://www.spc.noaa.gov/

climo/online/). The wind reports were layered over

the NLDN data to assist in determining flashes to

associate with the DMCS.

• The North American list of derecho events (Wikipedia

contributors 2014). This list of derechos include many

particularly intense events and was utilized initially to

begin building our database. The list encompasses

derechos documented in the scientific literature, from

other derecho lists (i.e., SPC noteworthy derecho events

website) and local National Weather Service studies.

Twenty-seven, warm season events were identified

during our time span with seven events discarded after

careful evaluation using the radar and wind damage

data. These events were discarded due to ambiguity as

to whether the event met derecho criteria, multiple

events occurring simultaneously making flash identifi-

cation problematic, or missing data from one of the

utilized datasets.

2.1 Selecting DMCSs and associated flashes

for analysis

To investigate CG flashes during DMCS episodes, the

research was limited to 2003 through 2013; the temporal

extent of the archived NLDN data. Additionally, we

identified warm season (May through August) DMCS

environments as they have the greatest probability of

extensive lightning production. Finally, the storm’s dam-

age swath and wind speeds were verified to ensure they

adhered to previous derecho identification criteria (Johns

and Hirt 1987). Using the list of North American derechos,

we identified 20 DMCS episodes warranting further anal-

ysis (Table 1). We utilized the radar imagery to assist in

determining the path and duration of the DMCSs. Starting

and ending times for each event were determined based on

when the DMCS had organized into a bow echo signature

and when the bow echo had effectively dissipated.

Damaging wind reports archived by SPC were imported

into QGIS 2.8 and used along with the WSR-88D imagery

to determine the extent of the damage swath and beginning

and ending times of each DMCS. These data were espe-

cially helpful for determining which lightning flashes were

to be included in each episode by examining the wind

damage swath and radar reflectivity along the path of the

DMCS. A wind damage layer was created that consists of

only data within the event’s duration and path, which was

determined by observing the evolution of the system from

the radar imagery.

Utilization of the lightning flash data constituted the

bulk of the GIS analysis. A definition query in QGIS was

employed to remove all lightning flashes occurring outside

the event’s temporal and spatial bounds. The wind report

data were then overlaid onto the flash data to aid inter-

pretation and refine the selection of DMCS flashes. To

create the main flash data layers for each event, the radar

imagery and wind report data were referenced to determine

the flashes that emanated from within the damaging wind

swath and along the path of the DMCS. A new vector

polygon layer was created that included all flashes identi-

fied as part of each DMCS episode. Geo-processing tools

were used to create a new point layer for all flashes within

the polygon. Although the wind data and radar imagery

were closely referenced to resolve the DMCS’s overall

damage swath and bow echo extent to ensure the archival

of relevant flashes, in several particularly active events,

multiple bow echoes and areas of convection were ongoing

within the spatial and temporal bounds of the primary

DMCS. These events contain elevated flash rates due to the

additional convective activity (Table 1). The flashes cap-

tured for analyses were from convection that occurred

along the spatial path and within the temporal bounds of

the primary DMCS that included convection within the

warm advection wing, leading edge cell mergers, and the

trailing stratiform region. The polygon created by outlining

the flashes identified with the spatial and temporal bounds

of each DMCS were then used to calculate areal extent and

flash metrics.

2.2 Analysis of DMCS flash densities

Heat maps were constructed using QGIS 2.8 to determine

coherent patterns in the flash distribution of DMCS envi-

ronments (Wilkinson and Friendly 2009). The heat maps

were calculated using kernel density estimation, a non-

parametric technique to describe the probability density

function of a random variable. The data matrices used in

rendering the heat maps were generated with a differing

number of columns depending on the geographical cover-

age for analysis. All matrices contained 2000 rows with the

number of columns automatically determined by QGIS.

When calculating heat maps, a search radius of 3.385 km

was employed for monthly and single event analysis,

meaning a summation of flashes falling within this circular

region were placed in a grid cell at the center. The heat

map of the entire DMCS distribution utilized a flash search

radius of 4.514 km. The search radius varied to ensure a

resolution appropriate to the number of flashes culled into

Lightning characteristics of derecho producing mesoscale convective systems

123

each grid cell as well as the areal extent of the domain. The

binned flashes within the search radii were given equal

weight and the heat maps were generated using the

Epanechnikov kernel. The Epanechnikov kernel is one of

the most commonly used estimators for kernel density and

is optimal in minimizing mean squared errors (Epanech-

nikov 1969). An analysis of flash density maxima was

conducted utilizing the raster output of the heat maps. Peak

value pixels were identified by polygonizing the raster

images produced from the heat map analyses. The resulting

vector layer was then analyzed for pixels containing flash

densities in the upper 10 % of the distribution. The iden-

tification of flash density maxima enabled easier interpre-

tation of flash density and are overlaid onto the gridded

flash data. The flash densities were constructed by creating

a 4 9 4 km fishnet grid over the domain which varied in

geographical size depending on the overall flash distribu-

tion (i.e., monthly, all flashes, etc.). For single event

analysis, a 2 9 2 km fishnet grid was utilized. Flashes

were binned into this grid and then mapped using the Jenks

natural breaks classification scheme to categorize the flash

densities through reducing in-class but maximizing

between class variance (Jenks 1967).

3 Results

The DMCSs identified for this investigation were signifi-

cant events with some derechos being particularly long-

lived and intense. This is evident in the 12-h average

duration and 1,909,606 flashes of these events (Table 1).

Three occurred in May, eleven in June, five in July and one

in August. Six DMCSs occurred in 2009, which was a

particularly active year for warm season, progressive

derechos, including a ‘‘super-derecho’’ which occurred on

8 May (Table 1; Evans et al. 2014). On average, the

DMCSs produced 69,379 flashes which yielded a flash

density of 0.23 flashes km-1 and an impressive flash fre-

quency of 5667 flashes h-1 (Table 1). This flash frequency

is considerably higher than the peak frequency of

2700 flashes h-1 found by Goodman and MacGorman

(1986) in a study of MCCs. The most prolific DMCS flash

rate was 10,066 flashes h-1 during 19 June, 2009

(Table 1). Even higher flash rates were calculated; how-

ever, these were likely inflated by ongoing additional

convection within the DMCS spatial and temporal domain.

Total flashes exhibited considerable variance and roughly

corresponded to the total area affected by the DMCS

Table 1 The DMCS events included in this investigation

Event start

(date/GMT)

Duration

(h)

Total CG

flashes

DMCS

area (km2)

Avg. flash density

(flashes km-2)

Flashes (h-1) Positive polarity

flashes

Positive

flashes (%)

5/21/04 14:00 15.0 247,727a 647,464 0.38 16,515 9809 4.0

7/22/03 6:00 16.8 186,368a 527,448 0.35 11,126 4870 2.6

7/11/11 6:54 16.5 163,159a 580,107 0.28 9906 N/A N/A

5/8/09 6:55 16.5 150,962a 576,097 0.26 9160 8535 5.7

6/4/08 9:50 14.2 120,698a 551,265 0.22 8530 3152 2.6

6/12/09 9:54 17.0 128,438 360,910 0.36 7555 4422 3.4

7/13/04 18:30 11.5 105,139 337,100 0.31 9143 2517 2.4

6/19/09 4:57 9.5 95,322 256,183 0.37 10,066 3904 4.1

6/18/10 12:25 15.0 88,564 360,045 0.25 5904 5231 5.9

6/11/12 16:51 14.1 87,030 583,288 0.15 6194 N/A N/A

6/18/09 9:27 11.0 83,076 289,196 0.29 7552 2806 3.4

5/3/09 8:55 14.0 76,836 464,635 0.17 5488 4975 6.5

6/13/13 2:28 13.5 67,643 390,819 0.17 5018 6537 9.7

6/29/12 18:55 10.5 65,951 440,212 0.15 6293 N/A N/A

6/23/10 3:55 12.5 64,080 227,055 0.28 5126 4024 6.3

6/16/09 9:24 13.6 56,582 470,038 0.12 4176 2095 3.7

7/17/06 22:00 12.0 47,368 117,147 0.40 3947 1778 3.8

7/21/08 6:22 10.0 31,645 229,444 0.14 3158 2617 8.3

8/4/08 7:54 9.4 29,529 169,276 0.17 3141 1557 5.3

6/24/13 19:26 6.0 13,489 175,821 0.08 2237 745 5.5

Averages 12.0 69,379 324,744 0.23 5667 3324 4.9

The ‘‘a’’ indicates events where additional convection occurred within the spatial and temporal bounds of the DMCS path, thereby increasing

flash activity. These events were omitted from the calculation of averages in the table

M. L. Bentley et al.

123

(Table 1). The percentage of positive flashes was found to

be 4.9 % for all events with considerable variance ranging

from 2.6 to 9.7 % (Table 1). This is lower than that iden-

tified in previous studies of MCCs/MCSs (Morgenstern

1991; Holle et al. 1994; Makowski et al. 2013) and may be

a result of the high DMCS flash densities.

The spatial distribution of flashes for all DMCS events

exhibits the distribution of derechos across the Eastern and

Central US (Fig. 1; Johns and Hirt 1987; Bentley and Mote

1998; Bentley and Sparks 2003). The highest activity

corridor runs from southern Wisconsin through Ohio with

another beginning in Oklahoma/Kansas and extending to

North Carolina (Fig. 1). Secondary flash density corridors

occur from Iowa southeastward through Tennessee and

from eastern Texas through the Deep South. The flash

density maxima which represent both overlapping DMCS

events as well as prolific lightning production are found

from southern Michigan to Ohio (Fig. 1). Flash densities in

several of these maxima exceed 13 flashes km-1 which is

nearly as high as the annual average flash densities found in

central Florida of 16 flashes km-1 (Holle 2014). The spa-

tial distribution of flash density for the five July events is

similar to all events (Fig. 2). The two primary activity

corridors for derechos are once again apparent as well as

high flash densities throughout southern Michigan and

Ohio. However, the highest July flash density emanates

from maxima in southwestern Tennessee where flash

densities exceed 12 flashes km-1 (Fig. 2). This flash den-

sity maximum was created by several derechos that pro-

gressed through the Tennessee Valley in 2003 and 2004

including the Memphis derecho of 2003 which caused $500

million in damage (McNeil et al. 2003).

Flash densities decreased slightly in June even though

eleven events occurred during the month (Fig. 3). The flash

pattern was more diffuse but covered the two major DMCS

activity corridors with broad swaths of lightning (Fig. 3).

Flash density maxima were located in both corridors with

the highest flash densities found in Arkansas, Mississippi,

and Indiana where over 9 flashes km-1 occurred. The

active northern DMCS corridor produced flash density

maxima from Iowa to Delaware and New Jersey. While

only three DMCS events occurred in May, two of them

produced over 150,000 flashes as additional convective

activity occurred within their spatial and temporal swaths

(Table 1). Each event produced a distinct flash activity

corridor with the greatest flash density maxima emanating

from northeast Indiana into Ohio due to the colocation of

several convective clusters during 21 May 2004 (Fig. 4).

Several of these flash activity maxima contained flash rates

of 9.8 flashes km-1. The 8 May 2009 event was particu-

larly intense, in fact, it was the first DMCS named a ‘‘su-

per-derecho’’ (Przybylinski et al. 2010; Coniglio et al.

2011; Evans et al. 2014). This event moved from Kansas

through Kentucky and produced, on average,

9162 flashes h-1 with the highest flash rates occurring in

Kansas near MCS organization and then through southeast

Missouri and into western Kentucky (Table 1; Fig. 4).

Evidence suggests that flash densities associated with

DMCSs make a significant contribution to the overall

nocturnal lightning climatology in the Central and Eastern

Fig. 1 CG lightning flash

density for all DMCS events.

Flash density classification is

per 16 km2 grid cells. Flash

activity maxima are outlined in

black

Lightning characteristics of derecho producing mesoscale convective systems

123

US (Holle 2014). A total of 491,440 flashes occurred

between 06 and 12 UTC (morning local time) which

accounted for 26 % of the DMCS activity. The overnight

and early morning flash density illustrated that many long-

lived DMCSs in the Central and Eastern US are largely

nocturnal events (Fig. 5). The highest flash densities occur

in northern Arkansas, southeastern Kansas and over Lake

Michigan with maxima of over 6 flashes km-1 (Fig. 5).

Additionally, the spatial pattern of flashes during the early

morning hours appears similar to the diurnal lightning

climatology of the US (Holle 2014). A slight decrease in

activity is noted between 12 and 18 UTC (morning to early

afternoon local time) with flash activity making up 25 % or

487,429 flashes. An eastward shift in flashes and slight

increase in flash densities occurs with maxima of over

9 flashes km-1 in southern Tennessee (Fig. 6). Georgia

Fig. 2 Same as Fig. 1, except

for DMCS events occurring in

July

Fig. 3 Same as Fig. 1, except

for DMCS events occurring in

June

M. L. Bentley et al.

123

and Alabama flash densities increase in this time frame as

many ongoing DMCSs have progressed eastward into the

Southeast. The time period with the highest flash activity is

18 to 00 UTC (late afternoon to evening local time) when

591,719 flashes or 31 % occurred. This late afternoon

period in the Central and Eastern US is when many

DMCSs are organizing, especially in the more active

northern corridor (Fig. 7; Johns and Hirt 1987). Flash

density maxima also show increased flash rates with areas

in Ohio tabulating rates exceeding 11 flashes km-1

(Fig. 7). The 00 to 06 UTC (late evening to early morning

local time) time frame contains the least amount of flash

activity at 332,480 flashes or 17 % of all flashes in the

dataset. Many DMCSs are entering dissipation in these

hours and the overall flash density also decreases, with

flash rates maximized in Arkansas at up to 7 flashes km-1

Fig. 4 Same as Fig. 1, except

for DMCS events occurring in

May

Fig. 5 Same as Fig. 1, except

for DMCS events occurring

between 00 and 0559 UTC and

flash activity maxima are not

denoted

Lightning characteristics of derecho producing mesoscale convective systems

123

(Fig. 8). This is the time period of greatest flash activity in

the Mid-Atlantic and Deep South due to the progression of

DMCS activity into these regions while typically under-

going weakening (Fig. 8).

The observed frequency of positive flashes has varied

widely in MCS investigations (Makowski et al. 2013; Mor-

genstern 1991; Lang et al. 2004). Evidence suggests that the

stratiform regions of MCSs contain the highest percentages

of positive polarity CG flashes and that some MCSs can

exhibit positive flash rates exceeding 20 % of all ground

flashes (Makowski et al. 2013). Due to an error in signal

strength that led to erroneous polarity, events in this inves-

tigation during 2011 and 2012 had to be omitted from

polarity tabulations (Table 1). These events were also

Fig. 6 Same as Fig. 5, except

for DMCS events occurring

between 06 and 1159 UTC

Fig. 7 Same as Fig. 5, except

for DMCS events occurring

between 12 and 1759 UTC

M. L. Bentley et al.

123

omitted when spatially comparing positive flashes to all CG

flashes. The remaining DMCSs yielded results similar to

Morgenstern (1991) in that all events exhibited positive flash

rates of less than 10 % of CG flashes (Table 1). The removal

of three events that occurred during 2011–2012 slightly

decreased flash densities, but did not significantly alter the

spatial distribution of DMCS flash activity (Fig. 9). Similar

flash densitymaxima exist and the three activity corridors are

still prevalent. However, the spatial distribution of positive

flash density shows marked differences to the overall dis-

tribution (Fig. 10). Evidence suggests that DMCSs ema-

nating from the northern activity corridor produce more

positive flashes than those occurring in the Southeast

(Figs. 9, 10). Positive flash densities approach

1.4 flashes km-1 in portions of Indiana and Ohio but remain

very low in Tennessee, Mississippi, and Alabama where

significant flash activity maxima exist (Figs. 9, 10). The

positive charging of large precipitation particles in the

updraft near the 0� isotherm can lead to enhanced positive

flashes in severe convective storms (Williams 2001). The

positive flash production is modulated by the freezing level

and updraft strength. Given the differences in positive flash

production between DMCSs in the northern versus southern

US, it would appear that freezing levels may play an

important role in governing positive flash production

(Fig. 10; Price and Murphy 2002).

The greatest flash activity in a single event occurred on 12

June 2009 when the ‘‘Mid-South’’ DMCS produced 128,438

flashes over its 17 h duration (Table 1). This particular event

prompted the issuance of a Particularly Dangerous Situation

(PDS) severe thunderstorm watch #388 by the Storm Pre-

diction Center. The DMCS produced several tornadoes, over

200 reports of damaging winds and considerable damage

throughout Memphis. The flash activity pattern contains

numerous shifts in flash rates throughout the evolution of the

DMCS (Fig. 11a). The two flash density maxima in northern

Mississippi appear to be associated with cell mergers on the

southern portion of the bow echo. During the second series of

cell mergers near the Alabama border, flash rates reached

13 flashes km-1 (Fig. 11a). Additionally, the flash signa-

tures of embedded cells within the DMCS are apparent in the

flash distribution as the system progressed southeastward

(Fig. 11b). The pulsing of system and convective intensity

within theDMCS is resolvable in the 2 9 2 kmgridded flash

distribution as well as enhanced flash production as a result

of cell mergers (Fig. 11b).

4 Conclusions

The overall spatial and temporal distribution of flash

activity during DMCS events is similar to prior research

detailing the climatology of these events (Johns and Hirt

1987; Bentley and Mote 1998; Bentley and Sparks 2003;

Coniglio and Stensrud 2004). Evidence suggests that

DMCSs produce significant flash activity in the overnight

hours and are a contributor to the nocturnal lightning

climatology of the Central and Eastern US (Holle 2014).

The 18 to 00 UTC time period contains the greatest

number of DMCS flashes while there exists a lull

Fig. 8 Same as Fig. 5, except

for DMCS events occurring

between 18 and 2359 UTC

Lightning characteristics of derecho producing mesoscale convective systems

123

between 00 to 06 UTC. Curiously, it appears that

northern activity corridor DMCSs produce considerably

higher amounts of positive polarity flashes compared to

those in the southern US. Evidence suggests, overall

flash densities are higher in DMCSs than MCS/MCCs

and given their duration, they tend to produce much

higher amounts of CG lightning.

Finally, results suggest that the use of high-resolution,

gridded flash data may be a useful tool to incorporate inmeso-

analyses of DMCSs.When combined with radar, satellite and

storm data, lightning data provide a high-resolution, objective

metric of the storm’s intensity and storm-scale interactions

(Metzger andNuss 2013).Theoverall flash patterns utilizing 4

and 2 kmgriddedflash data discern sub-synoptic details of the

Fig. 9 Same as Fig. 1, except

for omitting three DMCS events

occurring in 2011 and 2012

Fig. 10 CG positive polarity

flash density for all DMCS

events except those occurring in

2011 and 2012. Flash density

classification is per 16 km2 grid

cells. Flash activity maxima are

outlined in black

M. L. Bentley et al.

123

evolution of individual cells making up the DMCS. The

morphology of the flash activity produced by the DMCS,

further highlighted through the identification of flash density

maxima, can yield insights into storm-scale evolution and

processes within the parent MCS.

References

Bentley ML, Mote TL (1998) A climatology of derecho-producing

mesoscale convective systems in the Central and Eastern United

States, 1986–95. Part I: temporal and spatial distribution. Bull

Am Meteor Soc 79(11): 2527–2540

(a)

(b)

Fig. 11 a CG lightning flash

density for the 12 June 2009

DMCS with associated

damaging wind gusts

([25 m s-1). Flash density

classification is per 4 km2 grid

cells. b Same as a, exceptoutline of WSR-88D base

reflectivity greater than 40dBZ

overlaid onto flash density.

Arrow location of first cell

merger and immediate

downstream flash enhancement

Lightning characteristics of derecho producing mesoscale convective systems

123

Bentley ML, Sparks JA (2003) A 15 yr climatology of derecho-

producing mesoscale convective systems over the Central and

Eastern United States. Clim Res 24(2):129–139

Bentley ML, Mote TL, Byrd SF (2000) A synoptic climatology of

derecho producing mesoscale convective systems in the north-

central plains. Int J Clim 20(11):1329–1349

Coniglio MC, Stensrud DJ (2004) Interpreting the climatology of

derechos. Weather Forecast 19(3):595–605

Coniglio MC, Corfidi SF, Kain JS (2011) Environment and early

evolution of the 8 May 2009 derecho-producing convective

system. Mon Weather Rev 139(4):1083–1102

Cummins KL, Murphy MJ (2009) An overview of lightning locating

systems: history, techniques, and data uses, with an in-depth look

at the U.S. NLDN. IEEE Trans Electromagn Compat 51(3 PART

1):499–518

Cummins KL, Murphy MJ, Bardo EA, Hiscox WL, Pyle RB, Pifer AE

(1998) A combined TOA/MDF technology upgrade of the US

national lightning detection network. J Geophys Res Atmos

103(D8):9035–9044

Cummins KL, Cramer JA, Biagi CJ, Krider EP, Jerauld J, Uman MA,

Rakov VA (2006) The U.S. national lightning detection network:

post-upgrade status. Preprints, second conf meteorological

applications of lightning data. Atlanta, GA Amer Meteor Soc

Epanechnikov VA (1969) Non-parametric estimation of a multivari-

ate probability density. Theory Probab Appl 14(1):153–158

Evans JS, Doswell CA III (2001) Examination of derecho environ-

ments using proximity soundings. Weather Forecast

16(3):329–342

Evans C, Weisman ML, Bosart LF (2014) Development of an intense,

warm-core mesoscale vortex associated with the 8 may 2009

‘‘super derecho’’ convective event. J Atmos Sci

71(3):1218–1240

Fujita TT, Wakimoto RM (1981) Five scales of airflow associated

with a series of downbursts on July 16, 1980. Mon Weather Rev

109(7):1438–1456

Goodman SJ, Macgorman DR (1986) Cloud-to-ground lightning

activity in mesoscale convective complexes. Mon Weather Rev

114(12):2320–2328

Holle RL (2014) Diurnal variations of NLDN-reported cloud-to-

ground lightning in the United States. Mon Weather Rev

142(3):1037–1052

Holle RL, Watson AI, Lopez RE, Macgorman DR, Ortiz R, Otto WD

(1994) The life cycle of lightning and severe weather in a 3-4

June 1985 PRE-STORM mesoscale convective system. Mon

Weather Rev 122(8):1798–1808

Jenks GF (1967) The data model concept in statistical mapping. Int

Yearb Cartogr 7:186–190

Johns RH, Hirt WD (1987) Derechos: widespread convectively

induced windstorms. Weather Forecast 2:32–49

Lang TJ, Rutledge SA, Wiens KC (2004) Origins of positive cloud-to-

ground lightning flashes in the stratiform region of a mesoscale

convective system. Geophys Res Lett 31(10):L10105 1–4

Makowski JA, Macgorman DR, Biggerstaff MI, Beasley WH (2013)

Total lightning characteristics relative to radar and satellite

observations of oklahoma mesoscale convective systems. Mon

Weather Rev 141(5):1593–1611

McNeil SJ, Howell JL, Garrett GR, Valle DN (2003) The mid south

derecho—22 July 2003. National Weather Service Forecast

Office, Memphis

Metzger E, Nuss WA (2013) The relationship between total cloud

lightning behavior and radar-derived thunderstorm structure.

Weather Forecast 28:237–253

Morgenstern CD (1991) Cloud-to-ground lightning characteristics in

mesoscale convective systems. April–September 1986, MS

Thesis, University of Oklahoma, p 109

Price CG, Murphy BP (2002) Lightning activity during the 1999

Superior derecho. Geophys Res Lett 29(23):2142–2145

Przybylinski RW (1995) The bow echo: observations, numerical

simulations, and severe weather detection methods. Weather

Forecast 10:203–218

Przybylinski RW, Schaumann JS, Cramer DT, Atkins NT (2010) The

08 May 2009 Missouri derecho: radar analysis and warning

implications over parts of southwest Missouri. Preprints, 25th

conf on severe local storms, Denver, CO, Amer Meteor Soc

Rakov VA, Uman MA (2003) Lightning: physics and effects.

Cambridge University Press, UK, p 687

Rotunno R, Klemp JB, Weisman ML (1988) A theory for strong,

long-lived squall lines. J Atmos Sci 45(3):463–485

Rutledge SA, MacGorman DR (1988) Cloud-to-ground lightning

activity in the 10–11 June 1985 mesoscale convective system

observed during the Oklahoma-Kansas PRE-STORM project.

Mon Weather Rev 116(7):1393–1408

Weisman ML (1993) The genesis of severe, Long-lived bow echoes.

J Atmos Sci 50(4):645–670

Weisman ML, Evans C, Bosart L (2013) The 8 May 2009

superderecho: analysis of a real-time explicit convective fore-

cast. Weather Forecast 28:863–892. doi:10.1175/WAF-D-12-

00023.1

Wikipedia contributors (2014) List of derecho events, wikipedia, the

free encyclopedia. https://en.wikipedia.org/wiki/List_of_dere

cho_events. Accessed 10 Sept 2014

Wilkinson L, Friendly M (2009) History corner the history of the

cluster heat map. Am Stat 63(2):179–184

Williams ER (2001) The electrification of severe storms. In: Doswell

CA III (ed) Severe convective storms. American Meteor Soc,

AMS Monographs, Boston, pp 527–561

M. L. Bentley et al.

123