Radar and Lightning Observations of the 3 June 2000 ...olympic.atmos.colostate.edu/pdf/Tessendorf-2007-1.pdfJune. The polarimetric data were edited, gridded, and used in a fuzzy-logic

Radar and Lightning Observations of the 3 June 2000 Electrically Inverted Stormfrom STEPS

SARAH A. TESSENDORF

Department of Atmospheric Science, Colorado State University, Fort Collins, Colorado

KYLE C. WIENS

Los Alamos National Laboratory, Los Alamos, New Mexico

STEVEN A. RUTLEDGE

Department of Atmospheric Science, Colorado State University, Fort Collins, Colorado

(Manuscript received 11 July 2006, in final form 12 December 2006)

ABSTRACT

This study addresses the kinematic, microphysical, and electrical evolution of an isolated convectivestorm observed on 3 June 2000 during the Severe Thunderstorm Electrification and Precipitation Studyfield campaign. Doppler-derived vertical velocities, radar reflectivity, hydrometeor classifications frompolarimetric radar, and Lightning Mapping Array (LMA) charge structures are examined over a nearly 3-hperiod. This storm, characterized as a low-precipitation supercell, produced modest amounts of hail, de-termined by fuzzy-logic hydrometeor classification as mostly small (�2 cm) hail, with one surface report oflarge (�2 cm) hail. Doppler-derived updraft speeds peaked between 20 and 25 m s�1, and reflectivity wasnever greater than 60 dBZ. The most striking feature of this storm was its total lack of cloud-to-ground(CG) lightning. Though this storm was electrically active, with maximum flash rates near 30 per minute, noCG flashes of either polarity were detected. The charge structure inferred from the LMA observations wasconsistent with an inverted dipole, defined as having a midlevel positive charge region below upper-levelnegative charge. Inverted charge structures have typically been considered conducive to producing positiveCG lightning; however, the 3 June storm appeared to lack the lower negative charge layer below theinverted dipole that is thought to provide the downward electrical bias necessary for positive CG lightning.

1. Introduction

Based on cloud-to-ground (CG) lightning climatolo-gies, CG lightning lowering negative charge to groundis far more common than positive CG lightning acrossthe majority of the United States. However, stormsdominated (�50%) by CG lightning that lower posi-tive charge to ground are indeed observed and appearto be most frequent in the high plains of the UnitedStates (Orville and Huffines 2001; Carey and Rutledge2003). While several hypotheses have emerged to ex-plain the possible charge structure of positive CG-dominated storms (Brook et al. 1982; Seimon 1993;

Carey and Rutledge 1998; Lang and Rutledge 2002),few studies have had three-dimensional lightning ob-servations available to infer the general charge struc-ture of such storms. The Severe Thunderstorm Electri-fication and Precipitation Study (STEPS; Lang et al.2004) field campaign took place between 17 May and 20July 2000 in eastern Colorado and western Kansas. Agoal of the STEPS campaign was to identify relation-ships among microphysics, dynamics, and electrificationin severe storms on the high plains, and in particular, toinvestigate positive CG lightning production. A com-prehensive network of observing systems was deployed[see Lang et al. (2004) for a complete listing], includingtwo dual-polarization Doppler radars and a three-dimensional Lightning Mapping Array (LMA; Rison etal. 1999).

Until recently, hypotheses offered to explain positiveCG-dominated storms and positive CG lightning in

Corresponding author address: Sarah A. Tessendorf, NOAA/Earth System Research Laboratory, DSRC R/CSD3, 325 Broad-way, Boulder, CO 80305.E-mail: [email protected]

VOLUME 135 M O N T H L Y W E A T H E R R E V I E W NOVEMBER 2007

DOI: 10.1175/2006MWR1953.1

© 2007 American Meteorological Society 3665

MWR3451

general [e.g., the tilted dipole or inverted dipole out-lined in detail in Williams (2001)] do not discuss therole of a lower negative charge layer below the lowestpositive charge region. The charge structure typicallyassociated with negative CG-producing storms is oftenreferred to as a “normal” tripole, consisting of a mainmidlevel negative charge region below an upper-levelpositive charge layer, with a small lower positive chargelayer situated below the negative region (Simpson andScrase 1937; Krehbiel 1986; Williams 1989; Stolzenburget al. 1998). Several studies (e.g., Jacobson and Krider1976; Williams et al. 1989) suggest that in these normaltripole storms, the lower positive charge region is re-quired to produce negative CG lightning. The modelsimulations of storm electrification by Mansell et al.(2002, 2005) also suggest that lower negative charge re-gions may be necessary for positive CG flashes, consis-tent with the observations of Wiens et al. (2005).Hence, lower negative charge may play a role in theproduction of positive CG flashes similar to the roleplayed by lower positive charge in the production ofnegative CG flashes.

Several studies have already emerged from theSTEPS dataset, primarily concerning the 29 June 2000supercell that produced predominantly positive CGlightning (MacGorman et al. 2005; Tessendorf et al.2005; Wiens et al. 2005; Kuhlman et al. 2006). The goalof this study is to document properties of a STEPSstorm in which no CG flashes were detected, to be usedas comparison to the other cases in which CG flasheswere observed. This analysis is similar to that presentedin Tessendorf et al. (2005), Wiens et al. (2005), andTessendorf et al. (2007). By studying a broad spectrumof cases we hope to learn more about why some stormsare dominated by positive CG lightning.

2. Data and methods

a. Radar data and processing

The triple-Doppler radar network in STEPS wascomposed of three S-band (10–11-cm wavelength) ra-dars: the Colorado State University (CSU)–Universityof Chicago–Illinois State Water Survey (CHILL) pola-rimetric Doppler radar (hereafter CHILL), theNational Center for Atmospheric Research (NCAR)S-Pol polarimetric Doppler radar, and the Goodland,Kansas, National Weather Service Weather Surveil-lance Radar-1988 Doppler (KGLD). The three radarswere arranged in a nearly equilateral triangle with ap-proximately 60-km sides (Tessendorf et al. 2005, theirFig. 1). The 3 June storm developed north of the S-Polradar, near the northwestern edge of the eastern S-Pol–KGLD dual-Doppler lobe, and then tracked to the

southeast. It remained within the eastern S-Pol–KGLDdual-Doppler lobe throughout its lifetime.

Wind field syntheses were completed for 25 volumescans during the period 2301 UTC 3 June–0121 UTC 4June.1 The radar data were interpolated onto a Carte-sian grid with 0.5-km horizontal and vertical resolutionusing NCAR’s Sorted Position Radar Interpolator(SPRINT; Mohr and Vaughn 1979). After the grid in-terpolation, the velocity data from S-Pol and KGLDwere globally unfolded by means of NCAR’s CustomEditing and Display of Reduced Information in Carte-sian Space software (CEDRIC; Mohr et al. 1986). Inthe wind synthesis, the data from each radar were ad-vected to a common time using a manually calculatedstorm motion vector, and the vertical velocities wereobtained using a variational integration of the continu-ity equation (O’Brien 1970).

Polarimetric data were available from either CHILLor S-Pol between 2210 UTC 3 June and 0121 UTC 4June. The polarimetric data were edited, gridded, andused in a fuzzy-logic hydrometeor classification scheme(FHC) adapted from Liu and Chandrasekar (2000) andStraka et al. (2000), as in Tessendorf et al. (2005). TheFHC algorithm used a temperature sounding from the0013 UTC 4 June National Severe Storms LaboratoryElectric Field Meter (EFM) balloon flight. As in Tes-sendorf et al. (2005), hydrometeor echo volumes werecalculated for each radar scan time by multiplying thenumber of grid points that satisfied the FHC categoryof interest by the volume of a grid box (0.125 km3).

b. Lightning data and processing

The two sources of lightning data used in this studyare the National Lightning Detection Network (NLDN;Cummins et al. 1998) and the New Mexico Institute ofMining and Technology (NMIMT) LMA (Rison et al.1999). The NLDN provides the location, polarity, mul-tiplicity, and peak current of CG flashes. The LMAprovides measurements of the time and three-dimensional location of very high-frequency (VHF) ra-diation sources emitted by lightning discharges. For agiven lightning flash, the LMA may locate hundreds tothousands of such sources resulting in detailed maps ofthe total lightning activity. The LMA may suggest adownward-propagating leader to ground, but leaders toground are poorly resolved by the LMA, particularly inthe case of positive CG flashes. Furthermore, the ver-tical location accuracy is very poor for sources near the

1 The S-Pol radar went down for 20 min prior to the 0026 UTCvolume scan and therefore syntheses could not be performed dur-ing that period.

3666 M O N T H L Y W E A T H E R R E V I E W VOLUME 135

ground (Thomas et al. 2004). Hence, CG return strokescannot be identified with any confidence using onlyLMA data; thus, CG detection in this study was basedon the NLDN data only.

To determine total [CG plus intracloud (IC)] flashrates, we used a sorting algorithm developed at NMIMT(Thomas et al. 2003) that sorts the LMA sources intodiscrete flashes using a 10-source minimum thresholdfor the flash grouping. As shown in Wiens et al. (2005),calculating quantitative flash rates with this sorting al-gorithm is somewhat ambiguous, especially for stormswith intense lightning activity. Thus, we will focus ourdiscussion on the trends and the order of magnitude ofthe flash rates.

To infer charge structure, we analyzed the LMA dataon a flash-by-flash basis using the bidirectional dis-charge model (Kasemir 1960; Mazur and Ruhnke1993), as in Wiens et al. (2005). For example, we as-sumed that flashes initiate in strong electric fields be-tween regions of opposite net charge and propagatebidirectionally into the charge regions on either side.

The negative breakdown component of a lightning flashis higher in VHF power and thus is more readily de-tected by the LMA compared with the positive break-down component. Assuming that negative (positive)breakdown traverses regions of net positive (negative)charge, we infer the qualitative structure of the chargeregions involved in the flash based on the temporalevolution of the flash and on the relative number ofLMA sources on either side of the flash initiation loca-tion. Most IC flashes reveal distinct vertically separated“layers” of charge, with many more LMA sources inthe inferred positive layer (see Figs. 7, 8).

3. Overview

a. Environmental conditions

The environment on 3 June 2000 was characterizedby strong south-southwesterly surface winds between 8and 10 m s�1 (gusts to near 13 m s�1) ahead of a surfaceboundary (not shown). Weaker, northwesterly flowprevailed behind (west of) the boundary. Surface tem-

FIG. 1. M-GLASS sounding at 0012 UTC 4 Jun 2000 near Bird City, KS.

NOVEMBER 2007 T E S S E N D O R F E T A L . 3667

peratures ahead of the boundary were near 32°C, anddewpoints were near 12°C. Behind the boundary thetemperatures were similar, yet the dewpoints were aslow as 0°C. Around 1700 UTC, the surface boundaryappeared as a convergence line in the radial velocityfield and as a weak thin line echo in radar reflectivity(not shown). The boundary was oriented from south-west to northeast and propagated southeastward. The

0012 UTC Mobile GPS/Loran Atmospheric SoundingSystem (M-GLASS) sampled the inflow environment(within a few kilometers of the storm; Fig. 1) and indi-cated marginal CAPE (700 J kg�1) and notable dryingabove 500 hPa, and the upper-level winds were westerlynear 25 m s�1.

Just after 2200 UTC, two small cells, referred to as Aand B, were observed in southwestern Nebraska along

FIG. 2. Swath of (a) KGLD composite reflectivity (dBZ ) color contours for the period2210–0121 UTC with a black contour overlaid indicating the regions with updrafts greaterthan 10 m s�1, and (b) maximum updraft color contours during the dual-Doppler analysisperiod with a thick black contour overlaid indicating the regions with vertical vorticity greaterthan 10�2 s�1. Thin black contours at 20 and 40 dBZ from (a) are overlaid onto (b) forreference. Radar locations are denoted with a plus symbol.


Fig 2 live 4/C

the northern end of this boundary (Fig. 2a). Around2240 UTC, IC lightning was first detected in cell A. At2301 UTC (the first dual-Doppler analysis time avail-able), mesocyclonic strength vertical vorticity (�10�2

s�1; Moller et al. 1994) was observed in cell A (Fig. 2b).By 2330 UTC, cell B merged into the forward left flankof cell A. A visible split in the upper-level radar reflec-tivity echo was observed at 2331 UTC, and the left-moving cell began to diminish soon thereafter (Fig. 2).By 0030 UTC, the maximum updraft in cell A had de-clined to near 5 m s�1, and after that the radar reflec-tivity echo continued to decrease over time until thestorm had completely dissipated by 0121 UTC.

b. Kinematics and microphysics

At the time of the first dual-Doppler observations at2301 UTC, updraft speeds were near 20 m s�1 and by2350 UTC reached a maximum of �25 m s�1 (Fig. 3).During the gap in dual-Doppler scans, the maximumupdraft speed(s) measured by the South Dakota Schoolof Mines and Technology (SDSMT) T-28 aircraft (inpass 3) was 12.5 m s�1 (Holm 2005), though this may bean underestimate if the aircraft did not penetrate thestrongest updraft core. When dual-Doppler observa-tions were available again at 0026 UTC, peak Doppler-derived updrafts were near 13 m s�1 and soon thereaf-ter declined to 5 m s�1 and steadily decreased beyondthat. However, the T-28 aircraft measured updraftspeeds as high as 18 m s�1 between 0034 and 0037 UTC(pass 7). This discrepancy could be the result of thescale of the aircraft observations, which was smaller

than the resolution of the dual-Doppler analysis or dueto uncertainties in the dual-Doppler derived winds(Holm 2005).

The storm exhibited mesocyclonic strength verticalvorticity between 2301 UTC and 0026 UTC (Fig. 2b).At 2301 UTC, this vorticity was rather shallow, con-fined between 7 and 8 km, but by 2310 UTC it loweredin altitude to as low as 4 km and persisted for at leastanother 20 min based on our dual-Doppler observa-tions (not shown). Notice that the updraft and cyclonicvorticity were collocated with the main reflectivity corefor the duration of the dual-Doppler analysis (Fig. 2).Implications of this observation will be discussed fur-ther in section 5. Graupel was first detected by the FHCalgorithm near 2235 UTC in the midlevels of the storm(near 8 km) and steadily increased in echo volume(EV) until near 2320 UTC, at which time graupelamounts leveled off until near 2350 UTC (Figs. 3, 4).After this time, the graupel EV attained its maximumvalue of near 1000 km3 at 0002 UTC, most of which wascentered near 7 km MSL (Fig. 3).2

Total hail EV was minimal and confined to 7–9 kmbetween 0000 and 0020 (Fig. 3). According to NationalClimatic Data Center storm data, there was only onelarge hail (�2 cm) report associated with this storm.The report was at 0015 UTC, immediately after theFHC-inferred hail aloft. The hail EV contours in Fig. 3

2 All heights hereafter will be in kilometers above mean sealevel (MSL).

FIG. 3. Time–height contours of total graupel echo volume(solid black contours) and total hail echo volume (grayscale), andmaximum updraft time series (dashed black line; values on rightaxis) for 3 Jun 2000.

FIG. 4. Time series of updraft volume greater than 10 m s�1

(dashed black line; values on left axis), total graupel echo volume(solid black line; values on left axis), and the total flash rate fromthe LMA data (solid gray line; values on right axis) for 3 Jun 2000.


do not explicitly show this hail falling out, perhaps be-cause the temporal resolution of the radar data was toocoarse or the hail partially melted on its descent andbecame classified as graupel near the surface. However,it should be noted that most of the FHC-inferred low-level hail EV (i.e., that shown in the contours between2320 and 0026 UTC that appear constrained to near 3km) is suspect because of its horizontal stratificationjust below the melting level. It is possible that thesehydrometeors were water-coated graupel (i.e., in theprocess of melting) that was misclassified as small hail(SH; see section 4b). One other point to note is that thecalculated total hail EV is composed of the FHC smallhail and large hail (LH) categories, and for this storm,large hail was scarcely detected.

Because of the limited availability of dual-Dopplerscans, the trends in the volume of updraft greater than10 m s�1 (hereafter, UV10) cannot be fully resolved.However, during the portion of the time series withUV10 data, peaks in UV10 were coincident with orpreceded those in graupel and hail EV. Shortly after2310 UTC, UV10 rose sharply and then peaked at 2316UTC with a value of 97 km3, just before the first peakin graupel EV (Fig. 4). By 2331 UTC, UV10 began torise sharply again and then achieved its maximum valueof 140 km3 at 2344 UTC. The absolute maximum ingraupel EV happened within 15 min of the last resolvedUV10 peak, but during a period when UV10 was un-available, and thus it is unknown if UV10 increased,decreased, or remained steady during this gap.

This storm could be considered a supercell based onthe single criterion in Moller et al. (1994) that define asupercell as having a deep mesocyclone that persists onthe order of tens of minutes. This storm was most simi-lar to a low-precipitation (LP) supercell, however, be-cause its radar echo was relatively small and it did notexhibit a low-level hook echo, nor did it have strong(�60 dBZ) reflectivity (Fig. 2), indicating that it hadless overall precipitation than a “classic” supercell(Bluestein and Parks 1983). Average radar-derived rainrates (using an R–Kdp relationship; Cifelli et al. 2002)on 3 June were a factor of 2 lower than those in the 29June 2000 STEPS classic supercell during its intensephase (Fig. 5).

c. Charge structure

The relationship between total lightning and graupelEV trends in this storm clearly reinforces the impor-tance of graupel production to the electrification pro-cess. There was no lightning detected by the LMA inthis storm prior to the initial radar-inferred presence ofgraupel around 2240 UTC (Fig. 4), nor was there anystrong radar reflectivity (nothing greater than 35 dBZ)

prior to this time (not shown). Furthermore, the trendin total lightning flash rate closely followed that ofgraupel EV (Fig. 4). This temporal relationship wasconfirmed quantitatively by Wiens (2005), who found astatistical correlation coefficient of 0.81 between thetime series of total flash rate and graupel EV. Themaximum flash rate in this storm was near 36 perminute and occurred at 0002 UTC when the graupelEV reached its peak.

Throughout the duration of the 3 June storm, the vastmajority of lightning flashes occurred near the precipi-tation core of the storm, initiated near 9–11-km alti-tude, and extended downward producing relativelylarge numbers of LMA sources below the flash originand relatively few LMA sources above (Figs. 6, 7). Thissituation describes what could be termed an inverteddipole, with a negative charge region near 10–12-kmaltitude (T � �40°C) and positive charge below (Kreh-biel et al. 2000; Williams 2001; Rust and MacGorman2002). Some of these “inverted” IC flashes remainedvertically confined to the upper part of the storm, withthe positive charge centered near 10 km (T � �30°C)within strong (�30 dBZ) lofted echo. However, most ofthe inverted flashes extended to much lower altitudes,with the positive charge sloping downward east of theupdraft, apparently following the descent of the pre-cipitation (see section 4). Hence, the lower positivecharge may have consisted of multiple layers or simplyone deep charge layer. There were no flashes that in-dicated an intervening negative charge region within

FIG. 5. The average precipitation rate calculated at 3 km MSLfor 3 Jun 2000 (solid black) and 29 Jun 2000 (dashed black) foreach sequential radar volume in their respective analysis period.The analysis period for 3 Jun was 2210–0121 UTC (with 3–5-minspacing) and for 29 Jun was 2130–0115 UTC (with 5–7-min spac-ing).


the positive charge. As the time–height contours of to-tal LMA sources in Fig. 6b indicate, the bulk of theLMA sources were constrained between 5- and 10-kmaltitude, which is the same altitude range that we con-sistently identified as the positive charge region of aninverted dipole in our flash-by-flash analysis (see sec-tion 4). The LMA source density contours also closelyresemble the graupel EV contours in Fig. 3, furtheremphasizing the relationship between rimed ice andelectrification.

Assuming noninductive charging is responsible forelectrification, the implication of this inferred chargestructure is that larger ice particles (e.g., graupel) re-ceived positive charge after rebounding collisions withsmaller ice particles (e.g., ice crystals), and the latterreceived negative charge. Furthermore, according tolaboratory studies that base the sign of the charge trans-ferred on temperature and cloud liquid water content(LWC), effective LWC (a combination of the LWC andcollision efficiency), or rime accretion rate (Takahashi1978; Saunders et al. 1991; Saunders and Peck 1998),this would suggest that LWC or rime accretion rates

were large enough in this storm for the graupel to ac-quire positive charge. The maximum adiabatic LWC(calculated from the M-GLASS sounding in Fig. 1) was3.4 g kg�1 (at 8.9 km). It is well known that entrainmentand mixing effects dilute the LWC from adiabatic val-ues; however, aircraft observations in small cumulushave measured near-adiabatic LWC in the cores of up-drafts (Lawson and Blyth 1998). The SDSMT T-28 ar-mored aircraft measured maximum in situ LWC in therange of 2–3 g m�3 at an altitude near 6 km (Holm2005). Holm (2005) performed an LWC analysis thatcompared adiabatic LWC calculations from a compos-ite EFM/M-GLASS sounding at the height of the air-craft track with the actual SDSMT T-28 LWC measure-ments. Using the ratio of measured LWC to adiabaticLWC, Holm then adjusted the adiabatic LWC for theinferred regions of noninductive charging and foundthat the graupel would acquire negative charge usingthe Takahashi (1978) results but would acquire positivecharge using the Saunders et al. (1991) parameters. Thisreveals the discrepancies between the laboratory stud-ies of noninductive charging, as well as their extremesensitivity to LWC, and perhaps lends support to effec-tive LWC being an important noninductive chargingparameter rather than just LWC alone. Electrificationsimulations by Mansell et al. (2005) and Kuhlman et al.(2006) have also shown that different charging schemes(based on the different laboratory results) can yieldopposite polarity charging, and the rime accretion rateschemes (Saunders and Peck 1998) are more versatileand capable of producing inverted charge structures.Clearly, improvements in our knowledge of noninduc-tive charging parameters are still needed before anyconclusions can be made about how the observedcharge structures of this storm developed.

4. Detailed evolution

Based on the UV10, graupel EV, and lightning flashrate trends in Fig. 4, we have identified 3 main phases ofthe storm’s evolution: a developing phase (2210–2310UTC), a mature phase (2310–0010 UTC), and a dissi-pating phase (0010–0120 UTC). These three phases ofthe storm’s life cycle are similar to the three-stage clas-sification defined by Byers and Braham (1949). We willmake reference to these phases as we discuss the de-tailed observations.

a. Developing phase (2210–2310 UTC)

Near 2210 UTC, the 3 June storm (cell A in Fig. 2)was characterized by a high-based (�10 dBZ below5 km), shallow (�10 dBZ above 9 km) radar echo struc-ture with maximum reflectivity no greater than 30 dBZ

FIG. 6. Time–height contours of (a) LMA flash initiation heightand (b) total LMA sources (grayscale in logarithmic units) withtotal flash rate overlaid in black for 3 Jun 2000.


(not shown). The storm was not in a location for opti-mal dual-Doppler analysis throughout most of the de-veloping phase, so diagnosis of the updraft velocity dur-ing this time was not possible. The echo base (�10 dBZ)lowered to near 1.5 km by 2226 UTC, and reflectivity �30dBZ (inferred to be graupel by FHC) was first observedat 2233 UTC between 7 and 8 km on the west side of thelow-level reflectivity echo (not shown). This perhaps

indicated the presence of a new and stronger updraft onthe west flank of the storm, beneath this lofted echo.After this time, the storm continually exhibited a largervolume of reflectivity �30 dBZ and FHC-inferredgraupel echo (Fig. 4). Soon thereafter, near 2240 UTC,the first lightning flashes were observed by the LMA. Asecond, weaker cell (B in Fig. 2) was first observed onradar at 2239 UTC to the northeast of cell A.

FIG. 7. Lightning mapping of an inverted IC flash at 2302:41 UTC 3 Jun 2000. (top) Altitude of theLMA sources vs time. (middle and bottom) Three different two-dimensional spatial projections alongwith an altitude histogram of the number of sources. The sources are color coded by time from blue tored. Plus and minus symbols indicate inferred ambient charge regions. This flash initiated near 10-kmaltitude and progressed downward into an inferred sloping positive charge region at 4–9-km altitude. Thedelayed sparse sources above the initiation point indicate the inferred negative charge region at 10–11-km altitude.


Fig 7 live 4/C

Dual-Doppler observations of the storm were avail-able by 2301 UTC, near the end of the developingphase. At 2301 UTC, both cells A and B had relativelyweak reflectivity, with cell A still containing reflectivityjust greater than 30 dBZ and some FHC-inferred grau-pel, while cell B did not exhibit any reflectivity greaterthan 30 dBZ (Fig. 9). Cell A had two updraft cores atthis time: one near the west flank of the reflectivitycore, as strong as 20 m s�1, and a shallow and narrow 10m s�1 updraft in the center of the echo (Fig. 9). Withoutdual-Doppler observations prior to this time, we cannot

diagnose the evolution of the second, smaller updraft,but perhaps it was an older updraft that was dissipating,while the stronger updraft on the west flank was anewer, developing updraft. Low-level inflow at thistime was weak and south-southeasterly, with upper-level flow from the northwest (Figs. 9a,b).

The charge structure inferred from the initial flashes,from 2243 to 2255 UTC, is best characterized as aninverted dipole, involving a positive charge region near8–9 km and an upper-level negative charge region at10–11 km (not shown). An example inverted flash at

FIG. 8. As in Fig. 7, but for a normal IC flash at 2302:30 UTC 3 Jun 2000. This flash initiated near9.5-km altitude and progressed upward into an inferred positive charge region at 10–11-km altitude. Thedelayed sparse sources below the initiation point indicate the inferred negative charge region at 9–10-kmaltitude.


Fig 8 live 4/C

2302 UTC is shown in Fig. 7. In addition to the flashesin this persistent inverted dipole, there was a roughly30-min time span (2255–2325 UTC) during which sev-eral flashes initiated upward from 10–12 km to an in-ferred upper positive charge region that lay near theupper radar echo boundary of the storm (Figs. 8–10).Flashes involving the upper positive charge were gen-erally within the anvil, farther downwind (east) of the

core. These upper flashes occurred farther and fartherfrom the core of the storm as time progressed. At 2301UTC, this upper positive charge region was above theinverted dipole and extended downwind into the anvil,centered at a height of 11 km (Figs. 8 and 9). Also atthis time, the LMA sources were only in the vicinity ofthe second, smaller updraft, suggesting that the chargeseparation mechanisms in the stronger (perhaps newer)

FIG. 9. KGLD horizontal radar reflectivity (Zh) at 2301 UTC 3 Jun 2000 at z � (a) 3 and (b) 7 km; vertical crosssections of (c) FHC using polarimetric data from CHILL and (d) KGLD Zh along the slanted black line indicatedin (a) and (b). Updraft velocity contours beginning at 10 m s�1 (with a 10 m s�1 contour interval) are overlaid inblack in (b) and (c) and in blue in (d). LMA sources of 4 representative flashes between 2302:22 and 2302:41 UTC,including the 2 shown in Figs. 7, 8, have been overlaid as small plus symbols onto (d), color coded by inferredambient charge (red � positive, green � negative). The first source of each flash is plotted in (d) as diamonds andtriangles, where the diamonds (triangles) indicate downward (upward) initial flash propagation. Storm-relativewind vectors have been overlaid onto (a) and (b). FHC categories are LH, SH, HG, LG, VI, wet snow (WS), drysnow (DS), rain (R), drizzle (Drz), and unclassified (UC). Note: the LMA sources atop the storm do not appearto be within reflectivity echo, but the 0- and 10-dBZ contours around the periphery of the storm are partiallymissing, either due to scanning/gridding geometry or the editing algorithms deleting the echo in low signal-to-noiseregions.


Fig 9 live 4/C

updraft had not yet advanced to the point of generatinglightning. The charge regions sloped downward awayfrom the updraft core, and most of the midlevel positivecharge was in a region of FHC-inferred graupel, whilethe negative charge aloft was in FHC-inferred snow andvertically oriented ice crystals (VI; Fig. 9). The extremeupper positive charge region was also in FHC-inferredsnow and ice crystals.

b. Mature phase (2310–0010 UTC)

During the mature phase of cell A, maximum updraftvelocities were near 20 m s�1 and UV10 reached 100 km3

(Figs. 3, 4). By 2325 UTC, cell B had merged into thenorthern flank of cell A and three 10 m s�1 updraftcores were resolved at 7 km (as seen in Figs. 10a,b). The

low-level inflow at this time was still south-south-easterly with northwesterly upper-level flow. The twolarger updraft areas (to the west and south) became onebroad updraft region by 2331 UTC (not shown), whilethe northern updraft core remained separate and evendeveloped its own upper-level reflectivity core by 2331UTC, creating a split in the storm reflectivity echo (alsoseen at 2344 UTC in Fig. 11b). Because of the enhancedcyclonic vorticity collocated with the updraft during themature phase of this storm, there was some cycloniccurvature in the flow around the south side of thesouthern, larger updraft (Figs. 10a,b). The location ofthis curvature in the flow relative to the updraft, how-ever, likely carried growing particles in the updraftaround to the downwind and northern side of the up-

FIG. 10. As in Fig. 9, but at 2325 UTC. The vertical cross sections in (c) and (d) are along the slanted line shownin (a) and (b), and the FHC used polarimetric data from S-Pol. LMA sources are from 4 representative flashes from2326:03 to 2326:33 UTC.


Fig 10 live 4/C

draft where they fell out away from the storm inflow,thus preventing them from reentering the updraft forfurther growth.

The reflectivity and FHC fields showed a modestoverhang at 2325 UTC (Figs. 10c,d), with graupel in-ferred in the upper levels of the updraft. The small hailclassified by FHC in Fig. 10c near 3 km was likely amisclassification of melting graupel. Note that in thisfigure, the hydrometeors directly above the region clas-sified as SH are high-density graupel (HG) and low-density graupel (LG). As the LG particles fall throughthe melting level at 4.5 km, the meltwater on their sur-faces promotes higher radar returned power (i.e., radarreflectivity; similar to a bright band), and due to theincreasing radar reflectivity, the LG particle is classifiedas HG, and then as SH, based on the radar variable

thresholds in the FHC algorithm (Tessendorf et al.2005).

The charge structure in the mature phase of thisstorm was still characterized as an inverted dipole near-est the precipitation core (Fig. 10d). Most of the in-ferred positive charge was again in the region of FHC-inferred graupel, while the inferred negative charge wasin the upper levels where snow and ice crystals wereidentified by FHC. The single flash in the eastern anvilin Fig. 10 was the last flash that clearly involved theupper positive charge, and it is much farther downwindfrom the precipitation core than were previous flashesthat involved the upper positive charge. Both the nega-tive and upper positive charge layers of this flash werein regions of FHC-inferred snow and ice crystals in theanvil.

FIG. 11. As in Fig. 9, but at 2344 UTC. The vertical cross sections in (c) and (d) are at y � 71 km, and the FHCused polarimetric data from S-Pol. LMA sources are from 3 representative flashes from 2344:23 to 2344:40 UTC.


Fig 11 live 4/C

By 2344 UTC, the updraft reached its absolute maxi-mum intensity (based on the available dual-Dopplerobservations) in the main updraft along the southwestflank of the storm (Fig. 11). The northern flank updrafthad a distinct reflectivity core that was diverging fromthe southern updraft over time and responsible for thenorthern branch of the V-shaped low-level reflectivity.Persistent south-southeasterly inflow and upper-levelnorthwesterly flow was still evident, as well as somecyclonic curvature around the south side of the mainupdraft (Figs. 11a,b). The persistent inverted dipole(now without the upper positive charge layer) was stillthe dominant charge structure and, within the updraftregion, FHC-inferred graupel (ice crystals) was (were)observed where the LMA indicated positive (negative)charge.

c. Dissipating phase (0010–0120 UTC)

Though the dual-Doppler observations were notavailable for the 20-min period prior to 0026 UTC, it isapparent that the storm entered its dissipating phaseduring this time. The graupel EV and total lightningflash rates rapidly diminished near 0010 UTC, and by0026 UTC dramatically weaker maximum updraftspeeds and UV10 were observed (Figs. 3, 4). At 0026UTC, two distinct cores were observed in the low-levelreflectivity field, each corresponding to the northernand southern updraft cores previously discussed at 2344UTC (Figs. 12a,b). The upper-level reflectivity in thenorthern core had greatly diminished by 0026 UTC,while the southern upper-level reflectivity core was still�30 dBZ. The updraft in the southern core was still

FIG. 12. As in Fig. 9, but at 0026 UTC. The vertical cross sections in (c) and (d) are at y � 53 km, and the FHCused polarimetric data from S-Pol. LMA sources are from 2 representative flashes from 0026:42 to 0027:08 UTC.


Fig 12 live 4/C

10 m s�1 at 0026 UTC, but quickly weakened to near 5m s�1 six minutes later. The low-level inflow was nowmore southerly, and the upper-level flow was west-southwesterly (Figs. 12a,b).

The charge structure during this phase was still rep-resentative of an inverted dipole, with a very deep mainpositive charge region situated below negative charge(Fig. 12d). The charge regions were elevated within theupdraft core and sloped downward away from the up-draft into the precipitation core. As in the previousfigures, FHC-inferred graupel was observed in regionswith positive charge, and ice crystals were inferred aloftin regions of negative charge (Fig. 12). This inverteddipole structure persisted up until the very last ob-served flash at 0046 UTC (not shown).

5. Discussion

The 3 June storm had moderate updraft speeds (20–25 m s�1), as well as moderate FHC-inferred graupeland small hail, with limited large hail. Compared withthe 29 June STEPS supercell (Tessendorf et al. 2005),the 3 June storm formed in an environment with lowerCAPE, had half the maximum updraft speeds, and hadnearly an order of magnitude less graupel and hail EV.A few possible reasons for the lack of (large) hail in thisstorm were weaker updrafts, lower UV10, and the col-location of the updraft and cyclonic vorticity (see Fig.2b). In the case of the 29 June storm, Tessendorf et al.(2005) showed that cyclonically curved flow on the rightflank of the updraft was an important ingredient, inaddition to sufficient updraft size and intensity, in theproduction of large (�2 cm) hail. In that storm, theoffset of the strong cyclonic flow from the updraft coreallowed embryonic particles, which had likely fallenfrom the upper-level stagnation zone upwind of the up-draft, to reenter the updraft for continued growth. Withthe cyclonic vorticity collocated with the updraft on 3June, however, the particles grown from scratch werelikely exhausted into the anvil or along the north side ofthe updraft, certainly not in a position to reenter thesoutheasterly inflow for continued growth (see Fig. 10).Nonetheless, the early evolution (first 2 h) of the 3 Junestorm had similar orders of magnitude of UV10, grau-pel EV, and lightning flash rates to the early evolutionof the 29 June storm, prior to the latter’s right turn anddramatic intensification (Tessendorf et al. 2005; Wienset al. 2005). The maximum updraft and hail echo vol-ume on 29 June prior to its right turn, however, wasalready much higher than that on 3 June.

Maximum total flash rates in this storm were near 30flashes per minute, which is below the 60 flashes perminute threshold found by Williams et al. (1999) to

distinguish nonsevere from severe storms in Florida.However, this threshold may be elevated because ofcontamination by single-source flashes (whereas weused a 10-source threshold in flash grouping), so the 3June flash rates may actually have been closer to thoseobserved in severe storms. This storm did have an iso-lated severe hail report, which technically classifies it asa severe storm, but nonetheless, its total flash rateswere somewhat low compared with other severestorms. Furthermore, no CG flashes of either polaritywere detected by NLDN. The lightning activity wassomewhat similar to that in the early (nonsevere) de-velopment of the 29 June supercell. Prior to the rightturn and onset of the severe phase in the 29 June storm,the total flash rates were on the same order of magni-tude as those on 3 June, and the 29 June storm pro-duced only two CG flashes during the first 2 h of light-ning. Once the 29 June storm became severe, flash ratesgained an order of magnitude and frequent positive CGflash activity began (Wiens et al. 2005).

Though the mechanism(s) that generated the in-verted dipole charge structure with an upper positivecharge layer cannot be identified with certainty, we canprovide some plausible assumptions. First, we assumethat most of the charge separating collisions took placein the main updraft and that the largest hydrometeorsresided and fell out closest to the strongest echo. Giventhe observations that positive charge was also con-strained within the strongest echo, the interpretation isthat the electrification processes were granting positivecharge to the larger particles. The high cloud base nearthe main updraft and charging region may explain whythe particles undergoing growth by riming were consis-tently charging positively. Following the arguments ofWilliams et al. (2005), higher cloud base reduces thewarm-rain (collision/coalescence) precipitation growthzone, thereby promoting higher supercooled liquid wa-ter contents and elevated riming rates in the mixed-phase region where charge-separating collisions occur.According to laboratory studies of the noninductive(collision charging) mechanism (Takahashi 1978; Saun-ders et al. 1991), graupel (i.e., the rimer) charges posi-tively under high LWC conditions.

There are a few possible explanations for the ex-treme upper positive charge: 1) it may have been ascreening layer above the inverted dipole, because itwas observed along the upper storm boundary, or 2) itmay have resulted from noninductive charging pro-cesses in a different charging regime (i.e., in regionswith different ambient temperature and LWC from themain updraft). For example, collisions where LWCwould be lower, in either the periphery of the main


updraft or in a weaker, older convective updraft, mayhave resulted in the riming ice receiving negativecharge. The observations are most consistent with thelatter explanation. The flashes involving the extremeupper positive charge propagated away from the stormcore over time, and the last of these flashes, near 2325UTC (Fig. 10), shows the upper positive LMA sourceswithin noteworthy radar echo, suggesting that thischarge layer was perhaps associated with a group ofhydrometeors in the anvil and not just a screeninglayer.

It has been suggested that a lower positive chargeregion locally enhances the electric field, providing im-petus for negative discharges to ground (Jacobson andKrider 1976; Williams et al. 1989; Mansell et al. 2002).Recent studies suggest a similar role for a lower nega-tive charge region in the production of positive CGflashes in inverted polarity storms (Mansell et al. 2002,2005; Wiens et al. 2005). Like the 3 June storm of thisstudy, the 29 June supercell also exhibited an invertedcharge structure. However, unlike the 3 June storm,LMA observations during the severe phase of the 29June storm showed a lower negative charge region be-low the inverted dipole (Wiens et al. 2005). Addition-ally, Kuhlman et al. (2006) simulated the electrificationin the 29 June storm and suggested that the observedand simulated lower negative charge layer was crucialto the production of positive CGs. The LMA datanever indicated the presence of a lower negative chargelayer below the inverted dipole on 3 June. Therefore itwould follow that the lack of a sufficient lower negativecharge layer on 3 June may have inhibited the produc-tion of positive CG flashes.

The reason for the apparent lack of a lower negativecharge layer on 3 June is difficult to pinpoint. In thesimulation of the 29 June supercell by Kuhlman et al.(2006), the lower negative charge formed by a combi-nation of negative noninductive charging of graupeloutside the updraft core, precipitation fallout and recy-cling, and inductive charging. Based on additionalstorm electrification simulations, Mansell et al. (2005)suggested that noninductive charging could account forthe lower charge layer without inductive charging pro-cesses, if the ice crystal concentrations at lower alti-tudes (i.e., warmer temperatures) were high enough(�50 L�1 in their simulations), but for all other cases,inductive charging was deemed important. Guided bythese modeling results, we speculate that the LMA didnot indicate a lower negative charge layer on 3 Junebecause of one (or all) of the following factors:

1) There was a lower negative charge region, but it wastoo weak to initiate a discharge. Since the LMA

cannot reveal a charge region unless that region isinvolved in lightning, there was no LMA evidence ofthe lower negative charge. This explanation is par-tially supported by the 0013 UTC EFM soundingthat indicated a weak lower negative charge layermay have been present (Rust et al. 2005).

2) Inductive charging processes were inhibited, per-haps because of less liquid precipitation in this LPstorm.

3) The lack of precipitation recycling reduced thequantity of riming ice growing at lower altitudes,which suppressed (noninductive or inductive) pre-cipitation-based charge separation processes neededto generate the low-level charge layer.

Though we presume that inverted charge structuresfavor positive CG lightning, provided a lower negativecharge layer is present, negative CGs have been shownto originate from the upper negative charge in an in-verted charge structure (Wiens et al. 2005). We suggestthat negative CG flashes in the 3 June storm were un-likely, however, because of the presence of a deep posi-tive charge layer between the upper negative chargeand the ground. This deep positive charge layer mostlikely made negative CG flashes less energetically fa-vorable than IC flashes between the two charge layers(Marshall and Stolzenburg 2002).

We feel it is important to make the distinction be-tween storms studied in the literature that have low-CGbut high-IC flash rates, or high IC:CG ratios (MacGor-man et al. 1989; MacGorman and Burgess 1994; Lang etal. 2000; Lang and Rutledge 2002), and the presentstorm that exhibited no CG flashes, because the rea-sons for each type of behavior could be due to differentmechanisms. For example, the proposed “elevatedcharge” hypothesis (MacGorman et al. 1989), which hasbeen previously used to explain low-CG storms, stillseems to be a plausible reason for keeping CG flashrates low, while maintaining or enhancing IC flash ratesin kinematically intense storms. In fact, even in thisstorm, the LMA sources and inferred charge layerswere observed at higher altitudes nearest the strongestupdraft than in the rest of the storm (see Figs. 9–12).However, we speculate that the absence of a sufficientlower charge layer, both opposite in polarity to thecharge region sending charge to ground and of appro-priate strength to enhance the electric field and providethe impetus for the discharge to ground, is perhaps akey reason for the lack of CG flashes in otherwise elec-trically active storms. This suggestion is based upon thetwo key observations of the electrically active 3 Junestorm: no CG flashes were detected, and there was noLMA-inferred lower negative charge region.


6. Conclusions

The objective of this study was to examine relation-ships among the kinematic, microphysical, and electri-cal aspects of the 3 June 2000 non-CG-producing su-percell. The radar coverage on this day was suitable forstudying the storm structure evolution; however, thedual-Doppler coverage was not optimal, and thereforewe were unable to estimate vertical velocities over partof the storm’s evolution. Nonetheless, the isolated na-ture of this storm, in addition to its modest flash rates,provided a unique opportunity to study the evolution ofcharge structure for an inverted storm using the LMAdata.

No CG flashes of either polarity were detected in thisstorm. It exhibited a persistent inverted dipole chargestructure, but the LMA data never indicated the pres-ence of a lower negative charge region below the in-verted dipole. Much like the lower positive charge re-gion has been deemed important in producing negativeCG flashes, these data support the idea that the lack ofthe lower negative charge layer, which completes theinverted tripole and may locally enhance the electricfield allowing for positive CG discharges, may havebeen a key factor in preventing this storm from pro-ducing positive CG flashes. Certainly, more storms thatproduce IC but not CG lightning need to be examinedto evaluate this claim.

Acknowledgments. We thank L. Jay Miller of NCARfor support with the multiple-Doppler analysis. Thanksto Dr. Larry Carey (Texas A&M) for assistance withthe polarimetric data processing and hydrometeor iden-tification algorithm suggestions. The S-Pol and KGLDradar data were obtained from NCAR. We thank Dr.Paul Krehbiel, Dr. William Rison, Dr. Ron Thomas,Dr. Tim Hamlin, and Jeremiah Harlin of New MexicoTech for the LMA data and software. Reviews by Drs.Timothy Lang, Stephen Nesbitt, and Rob Cifelli werealso much appreciated and contributed to the improve-ment of this manuscript. This research was supportedby the National Science Foundation (NSF) PhysicalMeteorology Program under Grant ATM-0309303. TheCSU–CHILL facility is supported by the NSF and CSU.

REFERENCES

Bluestein, H. B., and C. R. Parks, 1983: A synoptic and photo-graphic climatology of low-precipitation severe thunder-storms in the Southern Plains. Mon. Wea. Rev., 111, 2034–2046.

Brook, M., M. Nakano, P. Krehbiel, and T. Takeuti, 1982: Theelectrical structure of the Hokuriku winter thunderstorms. J.Geophys. Res., 87, 1207–1215.

Byers, H. R., and R. R. Braham, 1949: The Thunderstorm. U.S.Government Printing Office, 287 pp. [NTIS PB-234-515.]

Carey, L. D., and S. A. Rutledge, 1998: Electrical and multipa-rameter radar observations of a severe hailstorm. J. Geophys.Res., 103, 13 979–14 000.

——, and ——, 2003: Characteristics of cloud-to-ground lightningin severe and nonsevere storms over the central UnitedStates from 1989–1998. J. Geophys. Res., 108, 4483,doi:10.1029/2002JD002951.

Cifelli, R., W. A. Petersen, L. D. Carey, S. A. Rutledge, and M. A.F. da Silva Dias, 2002: Radar observations of the kinematic,microphysical, and precipitation characteristics of two MCSsin TRMM-LBA. J. Geophys. Res., 107, 8077, doi:10.1029/2000JD000264.

Cummins, K. L., M. J. Murphy, E. A. Bardo, W. L. Hiscox, R. B.Pyle, and A. E. Pifer, 1998: A combined TOA/MDF technol-ogy upgrade of the U.S. National Lightning Detection Net-work. J. Geophys. Res., 103 (D8), 9035–9044.

Holm, M. L., 2005: A case study of the 3 June 2000 STEPS storm:The electrification of a low precipitation supercell. M.S. the-sis, Department of Atmospheric Sciences, South DakotaSchool of Mines and Technology, Rapid City, South Dakota,87 pp.

Jacobson, E. A., and E. P. Krider, 1976: Electrostatic fieldchanges produced by Florida lightning. J. Atmos. Sci., 33,103–117.

Kasemir, H. W., 1960: A contribution to the electrostatic theory oflightning discharge. J. Geophys. Res., 65, 1873–1878.

Krehbiel, P. R., 1986: The electrical structure of thunderstorms.The Earth’s Electrical Environment, National AcademyPress, 90–113.

——, R. Thomas, W. Rison, T. Hamlin, J. Harlin, M. Stanley, J.Lombardo, and D. Shown, 2000: Inverted polarity lightningin STEPS. Eos, Trans. Amer. Geophys. Union, 81 (Suppl.),Abstract A62D-06.

Kuhlman, K. M., C. L. Ziegler, E. R. Mansell, D. R. MacGorman,and J. M. Straka, 2006: Numerical simulations of the 29 June2000 STEPS supercell: Microphysics, electrification, andlightning. Mon. Wea. Rev., 134, 2734–2757.

Lang, T. J., and S. A. Rutledge, 2002: Relationships between con-vective storm kinematics, precipitation, and lightning. Mon.Wea. Rev., 130, 2492–2506.

——, ——, J. E. Dye, M. Venticinque, P. Laroche, and E. Defer,2000: Anomalously low negative cloud-to-ground lightningflash rates in intense convective storms observed duringSTERAO-A. Mon. Wea. Rev., 128, 160–173.

——, and Coauthors, 2004: The Severe Thunderstorm Electrifi-cation and Precipitation Study. Bull. Amer. Meteor. Soc., 85,1107–1125.

Lawson, R. P., and A. M. Blyth, 1998: A comparison of opticalmeasurements of liquid water content and drop size distribu-tion in adiabatic regions of Florida cumuli. Atmos. Res., 48,671–690.

Liu, H., and V. Chandrasekar, 2000: Classification of hydromete-ors based on polarimetric radar measurements: Developmentof fuzzy logic and neuro-fuzzy systems and in situ verification.J. Atmos. Oceanic Technol., 17, 140–164.

MacGorman, D. R., and D. W. Burgess, 1994: Positive cloud-to-ground lightning in tornadic storms and hailstorms. Mon.Wea. Rev., 122, 1671–1697.

——, ——, V. Mazur, W. D. Rust, W. L. Taylor, and B. C.Johnson, 1989: Lightning rates relative to tornadic storm evo-lution on 22 May 1981. J. Atmos. Sci., 46, 221–250.


——, W. D. Rust, P. Krehbiel, E. Bruning, and K. Wiens, 2005:The electrical structure of two supercell storms duringSTEPS. Mon. Wea. Rev., 133, 2583–2607.

Mansell, E. R., D. R. MacGorman, C. L. Ziegler, and J. M.Straka, 2002: Simulated three-dimensional branched light-ning in a numerical thunderstorm model. J. Geophys. Res.,107, 4075, doi:10.1029/2000JD000244.

——, ——, ——, and ——, 2005: Charge structure and lightningsensitivity in a simulated multicell thunderstorm. J. Geophys.Res., 110, D12101, doi:10.1029/2004JD005287.

Marshall, T. C., and M. Stolzenburg, 2002: Electrical energy con-straints on lightning. J. Geophys. Res., 107, 4052, doi:10.1029/2000JD000024.

Mazur, V., and L. H. Ruhnke, 1993: Common physical processesin natural and artificially triggered lightning. J. Geophys.Res., 98, 12 913–12 930.

Mohr, C. G., and R. L. Vaughn, 1979: An economical approachfor Cartesian interpolation and display of reflectivity factordata in three-dimensional space. J. Appl. Meteor., 18, 661–670.

——, L. J. Miller, R. L. Vaughn, and H. W. Frank, 1986: On themerger of mesoscale data sets into a common Cartesian for-mat for efficient and systematic analysis. J. Atmos. OceanicTechnol., 3, 143–161.

Moller, A. R., C. A. Doswell III, M. P. Foster, and G. R. Woodall,1994: The operational recognition of supercell thunderstormenvironments and storm structures. Wea. Forecasting, 9, 327–347.

O’Brien, J. J., 1970: Alternative solutions to the classical verticalvelocity problem. J. Appl. Meteor., 9, 197–203.

Orville, R. E., and G. R. Huffines, 2001: Cloud-to-ground light-ning in the United States: NLDN results in the first decade,1989–98. Mon. Wea. Rev., 129, 1179–1193.

Rison, W., R. J. Thomas, P. R. Krehbiel, T. Hamlin, and J. Harlin,1999: A GPS-based three-dimensional lightning mapping sys-tem: Initial observations in Central New Mexico. Geophys.Res. Lett., 26, 3573–3576.

Rust, D. R., and D. R. MacGorman, 2002: Possibly inverted-polarity electrical structures in thunderstorms during STEPS.Geophys. Res. Lett., 29, 1571, doi:10.1029/2001GL014303.

Rust, W. D., and Coauthors, 2005: Inverted-polarity electricalstructure in thunderstorms in the Severe Thunderstorm Elec-trification and Precipitation Study (STEPS). Atmos. Res., 76,247–271.

Saunders, C. P. R., and S. L. Peck, 1998: Laboratory studies of theinfluence of the rime accretion rate on charge transfer duringgraupel/crystal collisions. J. Geophys. Res., 103, 13 949–13 956.

——, W. D. Keith, and R. P. Mitzeva, 1991: The effect of liquidwater on thunderstorm charging. J. Geophys. Res., 96,11 007–11 017.

Seimon, A., 1993: Anomalous cloud-to-ground lightning in an F5-tornado-producing supercell thunderstorm on 28 August1990. Bull. Amer. Meteor. Soc., 74, 189–203.

Simpson, G. C., and F. J. Scrase, 1937: The distribution of elec-tricity in thunderclouds. Proc. Roy. Soc. London, 161A, 309–352.

Stolzenburg, M., W. D. Rust, and T. C. Marshall, 1998: Electricalstructure in thunderstorm convective regions. 3. Synthesis. J.Geophys. Res., 103, 14 097–14 108.

Straka, J. M., D. S. Zrnic, and A. V. Ryzhkov, 2000: Bulk hydro-meteor classification and quantification using polarimetric ra-dar data: Synthesis of relations. J. Appl. Meteor., 39, 1341–1372.

Takahashi, T., 1978: Riming electrification as a charge generationmechanism in thunderstorms. J. Atmos. Sci., 35, 1536–1548.

Tessendorf, S. A., L. J. Miller, K. C. Wiens, and S. A. Rutledge,2005: The 29 June 2000 supercell observed during STEPS.Part I: Kinematics and microphysics. J. Atmos. Sci., 62, 4127–4150.

——, S. A. Rutledge, and K. C. Wiens, 2007: Radar and lightningobservations of normal and inverted polarity multicellularstorms from STEPS. Mon. Wea. Rev., 3682–3706.

Thomas, R., P. Krehbiel, W. Rison, J. Harlin, T. Hamlin, and N.Campbell, 2003: The LMA flash algorithm. Proc. 12th Int.Conf. on Atmospheric Electricity, Versailles, France, ICAE,655–656.

——, P. Krehbiel, W. Rison, S. J. Hunyady, W. P. Winn, T. Ham-lin, and J. Harlin, 2004: Accuracy of the Lightning MappingArray. J. Geophys. Res., 109, D14207, doi:10.1029/2004JD004549.

Wiens, K. C., 2005: Kinematic, microphysical, and electrical struc-ture and evolution of thunderstorms during the Severe Thun-derstorm Electrification and Precipitation Study (STEPS).Ph.D. thesis, Colorado State University, 295 pp.

——, S. A. Rutledge, and S. A. Tessendorf, 2005: The 29 June2000 supercell observed during STEPS. Part II: Lightning andcharge structure. J. Atmos. Sci., 62, 4151–4177.

Williams, E. R., 1989: The tripole structure of thunderstorms. J.Geophys. Res., 94, 13 151–13 167.

——, 2001: The electrification of severe storms. Severe ConvectiveStorms, Meteor. Monogr., No. 50, Amer. Meteor. Soc., 527–561.

——, M. E. Weber, and R. E. Orville, 1989: The relationship be-tween lightning type and convective state of thunderclouds. J.Geophys. Res., 94, 13 213–13 220.

——, and Coauthors, 1999: The behavior of total lightning activityin severe Florida thunderstorms. Atmos. Res., 51, 245–264.

——, V. Mushtak, D. Rosenfeld, S. Goodman, and D. Boccippio,2005: Thermodynamic conditions favorable to superlativethunderstorm updraft, mixed phase microphysics and light-ning flash rate. Atmos. Res., 76, 288–306.


Documents

Radar and Lightning Observations of the 3 June 2000 ...olympic.atmos.colostate.edu/pdf/Tessendorf-2007-1.pdfJune. The polarimetric data were edited, gridded, and used in a fuzzy-logic