Abstract

In this study, it is hypothesized that the mesoscale environment can indirectly control the cloud-to-ground (CG) lightning polarity of severe storms by directly affecting their structural, dynamical, and microphysical properties, which in turn directly control cloud electrification and ground flash polarity. A more specific hypothesis, which has been supported by past observational and laboratory charging studies, suggests that broad, strong updrafts and associated large liquid water contents in severe storms lead to the generation of an inverted charge structure and enhanced +CG lightning production. The corollary is that environmental conditions favoring these kinematic and microphysical characteristics should support severe storms generating an anomalously high (>25%) percentage of +CG lightning (i.e., positive storms) while environmental conditions relatively less favorable should sustain storms characterized by a typical (≤25%) percentage of +CG lightning (i.e., negative storms). Forty-eight inflow proximity soundings were analyzed to characterize the environment of nine distinct mesoscale regions of severe storms (4 positive and 5 negative) on 6 days during May–June 2002 over the central United States. This analysis clearly demonstrated significant and systematic differences in the mesoscale environments of positive and negative storms, which were consistent with the stated hypothesis. When compared to negative storms, positive storms occurred in environments associated with a drier low to midtroposphere, higher cloud-base height, smaller warm cloud depth, stronger conditional instability, larger 0–3 km AGL wind shear, stronger 0–2 km AGL storm relative wind speed, and larger buoyancy in the mixed-phase zone, at a statistically significant level. Differences in the warm cloud depth of positive and negative storms were by far the most dramatic, suggesting an important role for this parameter in controlling CG lightning polarity. In this study, strong correlations between the mesoscale environment and CG lightning polarity were demonstrated. However, causality could not be verified due to a lack of in situ observations to confirm the hypothesized microphysical, dynamical, and electrical responses to variations in environmental conditions that ultimately determined the dominant CG polarity. Future observational field programs and cloud modeling studies should focus on these critical intermediary processes.

1. Introduction

Although the overwhelming majority (i.e., about 90%) of ground flashes lower net negative charge across the contiguous United States (CONUS; Orville and Huffines 2001), a few severe storms can generate positive cloud-to-ground (+CG) flash rates, densities, and percentages comparable to those typically observed for negative cloud-to-ground (−CG) flashes in active thunderstorms (e.g., MacGorman and Burgess 1994; Stolzenburg 1994; Carey and Rutledge 1998; Lang and Rutledge 2002; Carey et al. 2003a).

The large majority (about 80%) of warm-season severe storms throughout the CONUS generate mostly (>75%) negative ground flashes (i.e., so-called negative storms) while a minority (about 20%) produce an anomalously high (>25%) percentage of positive ground flashes (i.e., so-called positive storms; Carey et al. 2003b, hereafter CRP03).1 The frequency of these “anomalous” positive storms varies regionally (Fig. 1). Positive storms account for less than 10%–20% of all warm-season severe storms in the eastern and southern United States. In the central and northern plains from the Texas panhandle northeastward to Minnesota, 30%–90% of all severe storms are positive storms.

Fig. 1.

Percentage (color shaded as shown) of severe storm reports associated with >25% +CG lightning during the warm season (April–September) during 1989–98. The boxed region (dotted) indicates the IHOP_2002 experimental domain. The ellipse (dashed) marks the area of interest for this study where both positive and negative polarity storms are common. The significant overlap between the two regions highlights the suitability of the IHOP_2002 dataset for this study. (Adapted from CRP03.)

Fig. 1.

Percentage (color shaded as shown) of severe storm reports associated with >25% +CG lightning during the warm season (April–September) during 1989–98. The boxed region (dotted) indicates the IHOP_2002 experimental domain. The ellipse (dashed) marks the area of interest for this study where both positive and negative polarity storms are common. The significant overlap between the two regions highlights the suitability of the IHOP_2002 dataset for this study. (Adapted from CRP03.)

The geographic preference of positive storms for the central and northern plains over the eastern United States and southern plains suggests that such storms may be linked to specific meteorological conditions that are more prevalent in the favored regions. Several past studies have noted that severe storms passing through similar mesoscale regions on a given day exhibit similar CG lightning behavior (Branick and Doswell 1992; MacGorman and Burgess 1994; Smith et al. 2000), leading to the hypothesis that the local mesoscale environment indirectly influences CG lightning polarity by directly controlling storm structure, dynamics, and microphysics, which in turn control storm electrification (e.g., MacGorman and Burgess 1994; Gilmore and Wicker 2002, hereafter GW02; CRP03; Williams et al. 2005, hereafter W05).

2. Background

a. Thunderstorm charging and cloud-to-ground lightning polarity

The main electrical dipole present in most thunderstorms (i.e., negative storms) is composed of an upper-level positive over a low-level negative charge (Wilson 1920). Of course, it is now well known that the electrical structure in deep convection is more complex with three (tripole = upper positive, lower negative, and lower positive charges; Williams 1989) to four or five significant layers of charge in the vertical (Stolzenburg et al. 1998).

Recent observations with balloon-borne electric field mills (Rust and MacGorman 2002) and (very high frequency) VHF-based total lightning mapping networks (Lang et al. 2004; Wiens et al. 2005) suggest that positive storms are characterized by an inverted-dipole or inverted-polarity vertical electric field structure, meaning that the vertical configuration of the inferred charge polarity structure is nearly opposite to the arrangement found in the more typical negative storms. In inverted-polarity storms, the lowest significant charge layer of the thunderstorm tripole is negative, the midlevel charge is positive, and the main upper-level charge is negative (i.e., opposite to the conventional tripole).2 In one hypothesis (e.g., GW02; CRP03; W05), broad, intense updrafts and associated high liquid water contents in positive storms lead to positive charging of graupel and hail and negative charging of ice crystals during rebounding collisions via the noninductive charging (NIC) mechanism (e.g., Takahashi 1978; Jayaratne et al. 1983; Saunders et al. 1991; Saunders and Peck 1998). The differential fall speeds of ice particles and an accelerating updraft result in the storm-scale separation of the positively charged graupel and hail and the negatively charged ice crystals and the formation of an inverted dipole (i.e., negative over positive charge). This suggestion is sometimes broadly referred to as the “inverted-dipole” hypothesis. The more common NIC process, which is the negative charging of graupel and hail and positive charging of ice crystals, is hypothesized to occur in more typical storms with moderate updrafts and liquid water contents, resulting in the typical thunderstorm dipole and −CG lightning. The tilted dipole hypothesis (Brook et al. 1982) suggests that the upper positive charge of a typical dipole may become exposed to the ground through differential advection of charged hydrometeors in the presence of strong deep-layer shear, thus allowing +CG lightning to emanate from anvil level. The interested reader is referred to Williams (2001) for a more complete review of various hypotheses that seek to explain how storm dynamics and microphysics might control CG lightning polarity, particularly in positive storms. More recent discussion and hypotheses regarding the control of CG lightning polarity in severe storms can be found in GW02, Carey et al. (2003a), CRP03, Knupp et al. (2003), Wiens et al. (2005), and W05, among others.

b. The meteorological environment and cloud-to-ground lightning polarity

Given various hypotheses connecting cloud electrification to microphysical and dynamical processes, a handful of studies have explored the detailed relationship between the local meteorological environment and the CG lightning polarity of severe storms (Reap and MacGorman 1989; Curran and Rust 1992; Smith et al. 2000; GW02). From a comparison of numerical model output with radar and lightning data, Reap and MacGorman (1989) found that freezing level height and deep-layer wind shear were not as important as boundary layer fields such as moisture convergence, cyclonic relative vorticity, and strong upward vertical motions in determining the location of positive lightning activity. Using multiple proximity soundings for a single +CG lightning-producing tornadic storm, Curran and Rust (1992) found that strong deep-layer wind shear may be a necessary, but not sufficient, condition for the production of positive ground flashes. In a study of three tornadic outbreaks, Smith et al. (2000) used hourly surface airways observation data to show that positive (negative) storms occurred in a strong (weak) gradient region of the surface equivalent potential temperature (θe), upstream (over or downstream) of a surface θe maximum or ridge. A transition from positive to negative polarity CG lightning occurred when initially positive storms crossed the center of the θe ridge into decreasing values of θe. Assuming surface θe is well correlated to convective available potential energy (CAPE; e.g., Williams and Renno 1993) and hence to the potential maximum updraft speed via parcel theory, Smith et al. (2000) suggest that dominant +CG lightning polarity may be associated with rapid updraft intensification brought about by an increase in the buoyancy of low-level inflow air. Inversely, dominant −CG polarity may be associated with rapid updraft weakening brought about by precipitation loading and a decrease in inflow buoyancy. In a study of a tornado outbreak using mobile mesonet and proximity soundings, GW02 found that boundary-crossing supercells transitioned from dominant −CG to dominant +CG lightning when the storms experienced enhanced CAPE (below the in-cloud freezing level), boundary layer mixing ratios, and low-level (0–3 km) vertical wind shear on the immediate cool side of the boundary. Larger CAPE and 0–3-km shear in positive storms would result in both stronger thermodynamic and dynamic forcing of the updraft and more intense updrafts in the mixed-phase zone (e.g., Weisman and Klemp 1982, 1984, 1986; McCaul and Weisman 2001). Rotunno et al. (1988) present a theory that squall-line strength and longevity were most sensitive to the magnitude of the low-level (0–3 km AGL) vertical wind shear perpendicular to squall-line orientation (i.e., the RKW theory). An ‘‘optimal’’ state was proposed by RKW based on the relative strength of the circulation associated with the storm-generated cold pool and low-level shear. The deepest lifting and most effective convective retriggering occurred when these circulations were in near balance. RKW theory was recently reexamined and supported by additional 2D and 3D model simulations of Weisman and Rotunno (2004). RKW theory may explain why 0–3-km shear could affect the intensity and longevity of squall-line (multicell) convection and hence cloud electrification and lightning.

A few studies have drawn inferences regarding the relationship between CG lightning polarity and the meteorological environment over the CONUS using large-scale climatological data (Knapp 1994; CRP03; W05). Knapp (1994) noticed that regions in the CONUS where the troposphere tends to be relatively moist (e.g., east, southeast, southern plains) experienced significantly fewer positive storms. Using 10 yr of CONUS National Lightning Detection Network (NLDN) and the National Centers for Environmental Prediction (NCEP) reanalysis data, CRP03 showed that locations of the monthly frequency maxima of severe storms that produced predominantly −CG and +CG lightning were systematically offset with respect to the climatological monthly position of the surface θe ridge on severe outbreak days. Positive storms generally occurred west and northwest of the θe ridge in the upstream θe gradient region. Negative storms were primarily located southeast of the positive storm maximum, closer to the axis of the θe ridge in higher mean values of θe. The northward migration of this pattern from spring to summer months results in the well-known southwest-to-northeast (from the Texas panhandle to Minnesota) anomaly in the percentage of +CG lightning in annual maps (e.g., Orville and Huffines 2001) in addition to the geographic distribution of positive storms shown in Fig. 1 (CRP03; Carey and Rutledge 2003). Using climatological wet bulb potential temperature (θw) (as a proxy for CAPE) and cloud-base height (CBH) data, W05 suggest an important role for CBH in controlling the efficiency of the transfer of CAPE to updraft kinetic energy in thunderstorms. According to W05 “an elevated CBH may enable larger cloud water concentrations in the mixed-phase region, a favorable condition for the positive charging of large ice particles that may result in thunderclouds with a reversed polarity of the main cloud dipole.” This condition is also conducive to very large hail production (e.g., Knight and Knight 2001). W05 note the geographical coincidence of the climatological positive storm, very large hail, dry line, and low precipitation (LP) supercell signals over the central United States. Hence, W05 state that “the combined requirements of [large] instability and [high] CBH serve to confine the region of [positive severe storms] to the vicinity of the ridge in moist entropy in the western Great Plains.” It is important to point out that CRP03 found only a weak positive correlation between hail size and +CG fraction. In their study, hail size explained only a minor amount of the variance in the +CG fraction over the United States. Regional variability of +CG fraction, regardless of hail size, was dominant. Thus, one should conclude that the physical and hence environmental conditions associated with large hail and +CG lightning production, while potentially similar, are not uniquely related except in specific circumstances that have yet to be determined.

There are two underlying arguments associated with the role of CBH in the W05 hypothesis—one dynamical and one microphysical. Fundamentally important to both arguments is the observation that water loading and entrainment can significantly reduce the updraft velocity below what would be realized by CAPE from parcel theory, as was shown by Jorgensen and LeMone (1989) for tropical oceanic convection. W05 suggest a potential role for CBH in modulating water loading and entrainment and hence updraft strength. Others (e.g., Lucas et al. 1994, 1996; Michaud 1996, 1998; Williams and Stanfill 2002; Zipser 2003) have made similar suggestions. In the broad hypothesis, the diameter of the convective updraft is assumed to scale with the CBH or cumulus-topped boundary layer depth (e.g., Kaimal et al. 1976). Higher CBH is hypothesized to result in a larger updraft diameter, less entrainment (e.g., McCarthy 1974), more efficient processing of CAPE, stronger updrafts, and ultimately larger liquid water content (LWC) in the mixed-phase zone. It is interesting to note that Lang and Rutledge (2002) determined that a broad, strong updraft was a necessary ingredient for the production of enhanced +CG lightning. A higher CBH also implies a lesser warm cloud depth (WCD). A smaller WCD reduces the depth through which the warm rain (i.e., collision–coalescence) process can occur in the updraft, leaving a greater relative portion of cloud liquid water or greater (cloud water)/(rainwater) ratio. All else being equal, a higher CBH should result in larger supercooled liquid water content in the mixed-phase and charging zone (0° to −40°C) because of subsequent rainout or freezing of large rain drops (Rosenfeld and Woodley 2003).

Using idealized soundings and numerical cloud simulations, McCaul and Cohen (2002) demonstrated that when the level of free convection (LFC), which is equal to the lifting condensation level (LCL) in much of their study, is within an optimal range of 1.5 to 2.5 km above ground level (AGL), maximum convective updraft overturning efficiency can approach 100% of parcel theory, while for lower LFC = LCL heights the overturning efficiency is reduced significantly. The increased overturning efficiency with LFC height was associated with a larger diameter updraft and mean θe of the low-level updraft. In a sensitivity study, McCaul and Cohen (2002) found that the strongest updraft case was associated with a high LFC and low LCL scenario. From their study, one would expect the most efficient processing of CAPE, strongest updrafts, and least dilution of cloud water from high LFC and especially high LFC/low LCL environments. The inverted-dipole hypothesis discussed above would thus predict that high LFC (and low LCL) would be associated with positive storms, all else being equal. In a related study, McCaul et al. (2005) demonstrated that stronger peak updraft speeds occurred in lower precipitable water (PW) environments, all else being equal. The stronger peak updrafts were apparently the result of less rainwater loading and the lower altitudes at which the latent heat of freezing and deposition commences in relatively lower PW environments. Hence, PW could possibly modulate updraft strength independent of CBH.

Since it is difficult to obtain representative convective inflow soundings, only a few studies have analyzed representative thermodynamic and dynamic characteristics of the mesoscale environment in conjunction with CG lightning properties. Clearly, more field observations and further study is warranted in order to test and, if necessary, refine the above hypotheses linking environmental conditions to updraft intensity, supercooled cloud water contents, electrification, and ultimately CG lightning polarity. Therefore, this study seeks to investigate the relationship between CG lightning polarity and the immediate meteorological environment of severe storms, thereby providing further insight into why only some severe storms are dominated by +CG flashes, and in particular, what conditions lead to this +CG dominance. A determination of whether environmental conditions are systematically related to CG lightning polarity, and if so, what these conditions are, is a crucial step in determining the reliability of using NLDN real-time flash polarity data for the short-term prediction of severe weather. Furthermore, determining the relationship between certain environmental conditions and CG lightning polarity will lead to an improved understanding of cloud electrification mechanisms, especially in positive storms, which remains speculative at this time (Williams 2001).

3. Data and methodology

Using data from the International H20 Project (IHOP_2002; Weckwerth et al. 2004), we explored the relationship between the local meteorological environment and the CG lightning polarity of severe storms. In particular, we wished to determine if there were significant and systematic differences in the mesoscale environments of negative and positive severe storms that are consistent with the hypothesis discussed above. In a more general sense, our objective was to test if there was a significant relationship between CG lightning polarity and meteorological parameters that are known to affect storm dynamics and microphysics, and hence possibly cloud electrification and lightning. IHOP_2002 was conducted from 13 May to 25 June 2002 across the Southern Great Plains (Kansas, Oklahoma, and the Texas panhandle—see boxed area in Fig. 1). The IHOP_2002 area also happens to be ideal for our scientific objectives because it represents a transition region between the prevalence of positive and negative severe storms in the central CONUS (see elliptical area in Fig. 1), allowing us to carefully quantify the meteorological environment of both negative and positive storms. The main goal of IHOP_2002 was to obtain more accurate and reliable measurements of moisture in the air, in an attempt to improve quantitative precipitation forecasts and increase understanding of convective initiation (Weckwerth et al. 2004). Thus, detailed measurements of the mesoscale environment (both wind and thermodynamic parameters) in both the horizontal and vertical were obtained during IHOP_2002. Of particular interest to this study is the multitude of environmental soundings taken during IHOP_2002.

The relationship between CG lightning polarity and the local mesoscale environment was investigated for 6 different days (23, 24 May 2002; 4, 12, 15, 19 June 2002) during IHOP_2002. The chosen days were selected based on the prevalence of positive and/or negative storms (Table 1), the availability of proximity sounding data to characterize the local mesoscale environments of these storms (Table 2), and the occurrence of severe weather (Table 3). During these 6 days, nine distinct mesoscale regions were identified based on the CG lightning polarity characteristics of the storms occurring within them. Four of these regions contained positive storms and the remaining five were associated with negative storms as defined earlier (Table 1). All of the regions contained significant numbers of severe storm reports except for the negative mesoscale region on 23 May 2002, which contained only one large hail report (Table 3).

Table 1.

Definition of mesoscale regions within the IHOP_2002 domain and characterization of CG lightning. The overall percentage of +CG lightning, mean total CG lightning flash density, and number of positive and negative CG flashes are given.

Definition of mesoscale regions within the IHOP_2002 domain and characterization of CG lightning. The overall percentage of +CG lightning, mean total CG lightning flash density, and number of positive and negative CG flashes are given.
Definition of mesoscale regions within the IHOP_2002 domain and characterization of CG lightning. The overall percentage of +CG lightning, mean total CG lightning flash density, and number of positive and negative CG flashes are given.
Table 2.

Total number of soundings used to characterize each mesoscale region, and the number of these soundings that were full and truncated. The pressure level of truncation (hPa) is also provided, with the number in parentheses following the pressure level indicating the number of soundings truncated at that respective level (unless only one sounding was truncated at the given level). Sounding numbers and levels of truncation are also provided for the total number of positive (negative) soundings from all positive (negative) mesoscale regions combined.

Total number of soundings used to characterize each mesoscale region, and the number of these soundings that were full and truncated. The pressure level of truncation (hPa) is also provided, with the number in parentheses following the pressure level indicating the number of soundings truncated at that respective level (unless only one sounding was truncated at the given level). Sounding numbers and levels of truncation are also provided for the total number of positive (negative) soundings from all positive (negative) mesoscale regions combined.
Total number of soundings used to characterize each mesoscale region, and the number of these soundings that were full and truncated. The pressure level of truncation (hPa) is also provided, with the number in parentheses following the pressure level indicating the number of soundings truncated at that respective level (unless only one sounding was truncated at the given level). Sounding numbers and levels of truncation are also provided for the total number of positive (negative) soundings from all positive (negative) mesoscale regions combined.
Table 3.

Description of storm type and severity within each mesoscale region. Number of severe wind (≥26 m s−1), hail (diameter ≥1.9 cm), and tornado (F0–F5) reports during the analysis time period listed in Table 1 for each mesoscale region are given.

Description of storm type and severity within each mesoscale region. Number of severe wind (≥26 m s−1), hail (diameter ≥1.9 cm), and tornado (F0–F5) reports during the analysis time period listed in Table 1 for each mesoscale region are given.
Description of storm type and severity within each mesoscale region. Number of severe wind (≥26 m s−1), hail (diameter ≥1.9 cm), and tornado (F0–F5) reports during the analysis time period listed in Table 1 for each mesoscale region are given.

a. Cloud-to-ground lightning data

Cloud-to-ground lightning data analyzed in this study were collected by the NLDN (Cummins et al. 1998), which is owned and operated by Vaisala (Tucson, Arizona). The NLDN records the time, location, polarity, peak current, and multiplicity (number of strokes per flash) of CG lightning flashes. The time, location, polarity, and peak current reported for a flash are those measured for the first return stroke of the flash. Over most of the continental United States, including the IHOP_2002 domain (Fig. 1), the NLDN has a median location accuracy of 0.5 km and a flash detection efficiency of 80%–90% for strokes with peak currents greater than 5 kA (Cummins et al. 1998). Since +CG flashes with peak currents less than 10 kA were likely associated with misidentified cloud flashes (Cummins et al. 1998; Wacker and Orville 1999a, b), they were removed from the data sample as suggested by Cummins et al. (1998).

Maps of the +CG percentage over the IHOP_2002 domain (not shown) were visually inspected to define mesoscale regions across which either positive or negative storms prevailed. The overall percentage of positive flashes within each of these defined regions was then calculated to confirm that the mesoscale region was indeed characterized by positive storms (i.e., overall percent positive flashes for the mesoscale region >25%; termed positive mesoscale regions) or negative storms (i.e., overall percent positive flashes for the mesoscale region ≤25%; termed negative mesoscale regions). The maps were also visually inspected to ensure that the CG lightning polarity was consistent across the entire mesoscale region. Maps of flash density were used as checks to insure that the associated storms generated a significant number of flashes on a consistent basis across the defined mesoscale region. Through this method, four positive (23, 24 May; 15, 19 June) and five negative (23, 24 May; 4, 12, 15 June) mesoscale regions were identified (Table 1). As expected from CRP03 and Fig. 1, negative mesoscale regions were found generally east of positive mesoscale regions on days when both were present. The positive (negative) storms were characterized by a mean +CG percentage of 50% (9%) and a range of 32%–72% (7%–17%).3 Even though negative storms had generally larger CG flash densities and rates, the mean CG flash density in each region was comparable (i.e., same order of magnitude) because the negative storm systems were typically much more widespread over a larger geographic area than positive storm systems (Table 1).

b. Sounding data

Data from several different sounding platforms operating during IHOP_2002 were used to characterize the nine mesoscale regions of interest. Sounding platforms included National Weather Service (NWS) upper-air sites, Atmospheric Radiation Measurement Clouds and Radiation Testbed (ARM CART) sites, the National Center for Atmospheric Research/Atmospheric Technology Division (NCAR/ATD) Integrated Sounding System (ISS) facility, the National Severe Storms Laboratory (NSSL) Mobile Cross-chain Loran Atmospheric Sounding System (MCLASS) facility, and NCAR/ATD Mobile GPS/Loran Atmospheric Sounding System (MGLASS) facilities. Dropsondes launched from a Flight International (FI) Learjet were also used. Sounding data from these platforms were quality controlled and interpolated to a constant vertical resolution of 5 hPa by the University Corporation for Atmospheric Research (UCAR) Joint Office for Science Support (JOSS). See the appendix for details regarding the selection of proximity soundings.

The National Centers Advanced Weather Interactive Processing System (AWIPS) Skew-T Hodograph Analysis and Research Program (NSHARP; Hart and Korotky 1991) was used for sounding display and analysis, including the calculations of most sounding-derived parameters. NSHARP includes a virtual temperature correction (Doswell and Rasmussen 1994) in the calculations of thermodynamic parameters to account for the effects of water vapor. A mean-layer parcel, using mean temperature and dewpoint in the lowest 100 hPa (approximately 1 km in depth), was used to calculate thermodynamic parameters. Craven et al. (2002a) determined that a mean-layer parcel is more representative of the actual parcel associated with convective cloud development than is a surface-based parcel, and thus recommended using a mean-layer parcel in the calculations of thermodynamic parameters. Thompson et al. (2003) also found a mean-layer parcel to be superior to a surface-based parcel in accurately calculating thermodynamic parameters in convective storm environments. Equivalent potential temperature (θe) was computed from the recommended formula in Bolton (1980). CAPE is traditionally defined and calculated from the LFC to the equilibrium level (EL; Moncrieff and Miller 1976). For our comparison with CG lightning, CAPE for a specific temperature layer was calculated by modifying the soundings input to NSHARP. For example, to calculate “CAPE between the −10°C and −40°C levels,” we truncated the sounding input to NSHARP at −40°C, repeated the process with a sounding truncated at −10°C, and then subtracted the calculated CAPE value of the latter from the former NSHARP run.

In calculating environmental parameters from the sounding data, obviously only those soundings that contained the necessary data and extended through the necessary depth to accurately calculate each respective parameter were used. For instance, CAPE (i.e., LFC to EL) was not calculated from soundings truncated below the EL. As another example, storm system motion, which is estimated from the mean wind in the surface to 6 km AGL layer, could not be estimated by NSHARP for soundings that were truncated below 6 km AGL. As a result, any parameters dependent on storm system motion (e.g., storm relative winds) were also discarded from such soundings. Some of the mobile (MCLASS, MGLASS) soundings launched during IHOP_2002, as well as the dropsondes, did not extend through the full depth of the troposphere, since IHOP_2002 investigators were primarily concerned with measuring low-level moisture fields. Hence, some IHOP_2002 soundings could not be used for calculating parameters that require measurements through a significant depth of the troposphere [e.g., total CAPE, storm motion and dependent parameters, deep-layer (0–6 km AGL) shear, PW, etc.]. Table 2 displays the total number of soundings used to characterize each mesoscale region, the number of these soundings that were full (extended through the depth of the troposphere), the number that were truncated (did not extend through the depth of the troposphere), and their truncation levels. As seen in Table 2, all mesoscale regions included at least one full sounding so that all environmental parameters could be calculated for each region. In the positive (negative) regions, there were an average of 6.0 (4.8) total and 2.5 (4.4) full tropospheric soundings per region. This is a rather small number of soundings per region for a statistically significant comparison of environmental conditions between individual mesoscale regions. As a result, we combined all 24, including 10 full, positive region soundings into one distinct “positive group” and all 24, including 22 full, negative region soundings into another distinct “negative group” that could then be compared statistically.

In characterizing the mesoscale environments of positive and negative storms, special emphasis was placed on those thermodynamic and wind parameters that allowed testing of the hypothesis discussed earlier. The hypothesis states that the local mesoscale environment indirectly influences CG lightning polarity by directly controlling storm structure, dynamics, and microphysics, while the associated corollary more specifically states that broad, intense updrafts and associated high LWCs in positive storms lead to positive (negative) charging of graupel and hail (ice crystals) in mixed-phase conditions via the NIC mechanism, an inverted-polarity charge region, and increased frequency of +CG lightning. Based on this hypothesis and corollary, those environmental parameters that strongly influence storm organization, updraft intensity, and associated cloud LWCs were emphasized. To assess updraft intensity from buoyancy, several environmental parameters were investigated, including CAPE, normalized CAPE (NCAPE), and lapse rates computed through various layers of the atmosphere, especially those relevant to cloud electrification (i.e., between 0° and −40°C). NCAPE is the mean parcel acceleration associated with buoyancy in a layer, assuming parcel theory (Blanchard 1998). By dividing CAPE by the depth over which buoyancy is integrated, NCAPE can account for variations in the shape of the vertical profile of CAPE (e.g., Lucas et al. 1994) and provide a better estimate of updraft velocity (Blanchard 1998). CAPE, low-level (0–3 km) and deep-layer (0–6 km) vertical wind shear, and the bulk Richardson number (BRN) were investigated with regard to storm organization (e.g., Weisman and Klemp 1982, 1984, 1986) and dynamic forcing of the updraft (e.g., Weisman and Klemp 1982; McCaul and Weisman 2001). Depth of the warm cloud layer can affect the amount of cloud liquid water available to the mixed-phase region, and hence is hypothesized to affect storm electrification and CG lightning polarity as discussed earlier. Warm cloud layer depth was calculated by subtracting the LCL height (a measure of CBH) from the height of the 0°C level. (A complete list of calculated sounding parameters can be found in Tables 4 –8). Formally defining all of these environmental variables and describing their impact on the dynamics and microphysics of convection is beyond the scope of this paper. The interested reader is referred to Weisman and Klemp (1986), Houze (1993), and to more recent papers relating sounding parameters to severe convective weather (e.g., Rasmussen and Blanchard 1998; Moller 2001; Rasmussen 2003) for an overview.

Table 4.

Mean (median) sounding properties of grouped negative and positive mesoscale regions. The number of soundings in each negative (N−) and positive (N+) parameter sample is also shown. For each parameter, the statistical significance level of the difference in means according to the procedure described in section 3 was very highly significant (99.9% level, p ≤ 0.001).

Mean (median) sounding properties of grouped negative and positive mesoscale regions. The number of soundings in each negative (N−) and positive (N+) parameter sample is also shown. For each parameter, the statistical significance level of the difference in means according to the procedure described in section 3 was very highly significant (99.9% level, p ≤ 0.001).
Mean (median) sounding properties of grouped negative and positive mesoscale regions. The number of soundings in each negative (N−) and positive (N+) parameter sample is also shown. For each parameter, the statistical significance level of the difference in means according to the procedure described in section 3 was very highly significant (99.9% level, p ≤ 0.001).
Table 8.

Same as in Table 4, but the difference in each mean (median) sounding parameter was insignificant at the 90% level (p > 0.1).

Same as in Table 4, but the difference in each mean (median) sounding parameter was insignificant at the 90% level (p > 0.1).
Same as in Table 4, but the difference in each mean (median) sounding parameter was insignificant at the 90% level (p > 0.1).

The samples associated with positive and negative storm environments for each parameter were explored using standard statistical techniques (e.g., Wilks 1995). In particular, the samples for each parameter in each type of environment were compared for location using the arithmetic mean and median. Significance testing of the difference in location was accomplished using a rigorous approach. The two samples (i.e., associated with positive and negative storm environments) were first checked for normality using the Anderson–Darling test (Stephens 1974) at the 95% significance level. If both samples for a given parameter were found to come from a normal distribution, then a two-sample heteroscedastic (i.e., assuming unequal variances) two-tailed Student’s t test was performed to test the null hypothesis that the two population distribution functions corresponding to the two random samples are identical against the alternative hypothesis that they differ by location (i.e., their mean and medians are different). If one or both of the samples for a given parameter were found to be nonnormal, then an attempt was made to normalize them using a suitable Box–Cox (or power) transformation (Wilks 1995). If the samples could be reexpressed to normality, as judged by an Anderson–Darling normality test at the 95% significance level, then a two-sample heteroscedastic two-tailed Student’s t test was performed on the transformed data. Otherwise, the nonparametric Wilcoxon–Mann–Whitney rank sum test (Wilks 1995) was applied to the nonnormal samples to test the null hypothesis that the two population distribution functions corresponding to the two random samples are identical against the alternative hypothesis that they differ by location. To conduct these analyses, we used a combination of Microsoft Excel, the ITT Corporation Interactive Data Language (IDL), and the National Institute of Standards and Technology (NIST) Dataplot (Heckert and Filliben 2003) data visualization and statistical analysis packages.

4. Results

a. Overall mean environmental conditions

The overall mean (and median) environmental parameters for negative and positive mesoscale regions are presented in Tables 4 –8, as ranked by the highest significance level achieved with either the Student’s t test or the Wilcoxon–Mann–Whitney rank sum test as described in section 3. The mean (median) values listed for each parameter represent the parameter mean (median) for all soundings characterizing negative storm environments (i.e., all five negative mesoscale regions grouped together) and all soundings characterizing positive storm environments (i.e., all four positive mesoscale regions grouped together).

In agreement with Knapp (1994), negative storms occurred in a moister environment than positive storms as indicated by significantly higher mean low-level mixing ratio, PW, and surface dewpoint, along with a significantly lower mean surface dewpoint depression. The difference in means for all of the above moisture parameters was very highly significant (99.9% level) as shown in Table 4. Negative storms also occurred in regions of noticeably higher midlevel relative humidity, although the differences in the means were significant only at the 90% level (Table 7).

Table 7.

Same as in Table 4, but the difference in each mean (median) sounding parameter was insignificant at the 95% level but significant at the 90% level (0.05 < p ≤ 0.1).

Same as in Table 4, but the difference in each mean (median) sounding parameter was insignificant at the 95% level but significant at the 90% level (0.05 < p ≤ 0.1).
Same as in Table 4, but the difference in each mean (median) sounding parameter was insignificant at the 95% level but significant at the 90% level (0.05 < p ≤ 0.1).

The positive mesoscale regions were characterized by a significantly higher mean LCL as suggested by W05. In fact, the mean LCL for positive regions (2080 m AGL) was 1.9 times greater than that for negative regions (1120 m AGL). A higher LCL in combination with a slightly lower mean freezing level (and wet-bulb zero height) resulted in a much shallower mean WCD in positive mesoscale regions as also postulated by W05. More specifically, the WCD was 1.3 km deeper in negative storms (2950 m) as compared to positive storms (1700 m). The difference in mean LCL and WCD between the two regions was very highly significant (99.9% level; Table 4). By comparison, the LFC in positive and negative storm environments was not statistically different (mean and median of about 3 km for both; Table 8).

Mean lapse rates in the low to midtroposphere (850–500 and 700–500 hPa) were steeper in positive regions (Tables 4 and 6). The mean surface temperature was greater in positive regions as well (Table 5). The mean EL was higher in negative regions, and thus, despite little difference in the LFC between negative and positive regions, the depth of the free convective layer (EL − LFC) was greater in negative regions (Table 6).

Table 6.

Same as in Table 4, but the difference in each mean (median) sounding parameter was significant (95% level, 0.01 < p ≤ 0.05).

Same as in Table 4, but the difference in each mean (median) sounding parameter was significant (95% level, 0.01 < p ≤ 0.05).
Same as in Table 4, but the difference in each mean (median) sounding parameter was significant (95% level, 0.01 < p ≤ 0.05).
Table 5.

Same as in Table 4, but the difference in each mean (median) parameter was highly significant (99% level, 0.001 < p ≤ 0.01).

Same as in Table 4, but the difference in each mean (median) parameter was highly significant (99% level, 0.001 < p ≤ 0.01).
Same as in Table 4, but the difference in each mean (median) parameter was highly significant (99% level, 0.001 < p ≤ 0.01).

Interestingly, there was no significant difference in the mean total CAPE (LFC–EL) or mean lifted index (LI) for positive and negative storms (Table 8). The mean CAPE was moderate (roughly 2000 J kg−1) and the mean lifted index was low (−6° to −7°C), suggesting very unstable air masses and ample instability for strong updraft development in both regions. There was also no significant difference in the mean deep-layer (0–6 km AGL) vertical wind shear for positive and negative storms (Table 8), as also found by Reap and MacGorman (1989) and Curran and Rust (1992). Therefore, it is not surprising that the BRN did not differ significantly between positive and negative mesoscale regions either (Table 8). However, the mean convective inhibition (CIN) was significantly greater for negative regions (Table 6). Given recent interest in the relationship between dominant CG lightning polarity and position of the θe ridge (e.g., Smith et al. 2000; CRP03), it is interesting (but perhaps not surprising as discussed in section 5b) to note that mean θe values did not differ significantly between negative and positive regions, and were in fact very similar (Table 8).

Mean CAPE and NCAPE were calculated for several different vertical layers. In general, there was more CAPE at warmer temperatures (LFC to −10°C) in negative regions and more CAPE at colder temperatures (−10° to −40°C) in positive regions (Tables 5 –7). Inconsistent with the results from GW02, mean CAPE between the LFC and 0°C level and between the LFC and the −10°C level were significantly greater for negative storms (Tables 7 and 5, respectively). However, mean NCAPE between the LFC and −10°C level was not significantly different between negative and positive regions (Table 8), indicating that the greater CAPE values for negative storms in these lower layers was due to deeper mean LFC to 0°C and LFC to −10°C layers in negative regions, rather than to greater mean parcel buoyant acceleration within these layers. Mean CAPE between the −10° and −40°C levels and mean NCAPE between the LFC and −40°C levels were both 26% greater in positive regions, with the difference in means for both parameters significant at the 95% level (Table 6). Mean NCAPE between the −10° and −40°C levels was also greater for positive storms but was only significant at the 90% level (Table 7). Mean total NCAPE (LFC–EL) was greater for positive storms but the difference in means was only significant at the 90% level (Table 7). Similar to total CAPE, CAPE between the LFC and the height of the −40°C level was not significantly different between negative and positive storms (Table 8). In summary, CAPE and more precisely mean buoyancy acceleration (NCAPE) was larger for positive storms within the mixed-phase zone (0° to −40°C), which is critical for NIC and lightning production.

Consistent with the results of GW02, the 0–3 km AGL vertical wind shear was significantly stronger in positive (14.7 m s−1) versus negative (10.7 m s−1) mesoscale regions and hence could imply stronger dynamical updraft forcing of positive storms. The difference in the mean 0–3-km shear for the positive and negative regions was highly significant (99% level; Table 5). However, as noted above, 0–6 km AGL shear did not differ significantly between negative and positive storms, nor did 0–2 km AGL shear (Table 8). Low-level (0–2 km AGL) storm-relative wind speed was significantly higher in positive regions (Table 6). Low-level storm-relative wind speeds are a proxy for low-level inflow strength. Modeling studies have shown that the low-level outflow strength is detrimental to supercell maintenance and intensity when it is too strong relative to the low-level inflow (e.g., Weisman and Klemp 1982; Brooks et al. 1994a) because it undercuts the warm inflow into the updraft thereby weakening the convection. As a result, stronger inflow may allow the sustenance of positive storms in the presence of strong outflow. Interestingly, storm-relative wind speeds in the mid- and upper troposphere (4–6, 6–10, and 9–11 km AGL; EL) were fairly similar between negative and positive regions (Table 8), despite playing a hypothesized role in supercell microphysics and dynamics (e.g., Brooks et al. 1994a; Rasmussen and Straka 1998). The mean 0–3 km AGL storm relative helicity (SREH) was over twice as high in positive regions than in negative regions. However, due to a large variation in 0–3-km SREH values, this difference in means was only significant at the 90% level (Table 7). SREH provides a means of assessing the tendency for mesocyclone formation in supercells (Davies-Jones et al. 1990) and hence dynamical forcing of the updraft (e.g., Weisman and Klemp 1982). The 0–3-km shear and SREH results suggest that positive storms apparently experienced stronger dynamical forcing of the updraft than negative storms. Since total CAPE was so similar for positive and negative regions, the energy helicity index (EHI; using 0–3 km AGL SREH) did not differ significantly between positive and negative regions (Table 8).

b. Variability of sounding parameters

Relative frequency histograms of select sounding parameters that characterize the mesoscale environments of negative and positive storms from individual soundings (shown in Figs. 2 –11) highlight the significant (i.e., significance level ≥95% or p ≤ 0.05) differences in these variables and yet assess the degree of overlap in environmental conditions between the two types of regions. Finally, the histograms demonstrate the range of variability in each parameter associated with both negative and positive storms.

Fig. 2.

Relative frequency histogram of LCL (m, AGL) for individual soundings associated with negative and positive storms. Labels along the horizontal axis represent the maximum value for the bin.

Fig. 2.

Relative frequency histogram of LCL (m, AGL) for individual soundings associated with negative and positive storms. Labels along the horizontal axis represent the maximum value for the bin.

Fig. 11.

Same as in Fig. 2, but for NCAPE (m s−2) between the LFC and −40°C level.

Fig. 11.

Same as in Fig. 2, but for NCAPE (m s−2) between the LFC and −40°C level.

In agreement with the mean value comparisons and significance tests the histograms (shown in Figs. 2 –11) make it readily apparent that LCL (Fig. 2) and especially WCD (Fig. 3) differed much more between positive and negative mesoscale regions than most other parameters investigated. Although the populations were not completely distinct, there was relatively little overlap, especially for WCD. The modal LCL for positive and negative storms was 2000 and 1000 m AGL, respectively. The WCD for positive storms was somewhat bimodal with relative maxima around 1300 m, which was the primary peak, and 2100 m while the mode for negative storms was 3100 m. There was one outlier sounding launched in the vicinity of negative storms that was characterized by an LCL of approximately 2500 m AGL and a WCD of about 1100 m. While high LCL and shallow WCD were not exclusively associated with positive storms, the degree of separation between positive and negative regions evident in the LCL and WCD of individual soundings was noteworthy. As such, these results strongly support a role for the LCL and WCD in influencing the CG lightning polarity of severe storms in the central CONUS as first suggested by W05.

Fig. 3.

Same as in Fig. 2, but for WCD (m).

Fig. 3.

Same as in Fig. 2, but for WCD (m).

Figures 4 –6 illustrate once again that negative storms occurred in more moist environments than positive storms similar to the suggestion by Knapp (1994). The moister negative storm environments were noticeable in measurements of surface moisture (Fig. 4), low-level moisture (Fig. 5), and low- to midtropospheric moisture (Fig. 6). The moisture parameters exhibited more overlap of individual sounding values between negative and positive storms than did LCL and WCD, but the respective modes for positive and negative storms of all three moisture parameters were clearly distinct. The same can be said for 0–3 km AGL shear (Fig. 7). The modes were obviously different (7 m s−1 for negative storms versus 15 m s−1 for positive storms), with the positive sounding population occupying the higher end of the parameter spectrum, and the negative sounding population occupying the lower end, confirming the findings of GW02.

Fig. 4.

Same as in Fig. 2, but for surface dewpoint depression (°C).

Fig. 4.

Same as in Fig. 2, but for surface dewpoint depression (°C).

Fig. 6.

Same as in Fig. 2, but for PW (cm) between the surface and 400 hPa.

Fig. 6.

Same as in Fig. 2, but for PW (cm) between the surface and 400 hPa.

Fig. 5.

Same as in Fig. 2, but for mean mixing ratio (g kg−1) in the lowest 100 hPa.

Fig. 5.

Same as in Fig. 2, but for mean mixing ratio (g kg−1) in the lowest 100 hPa.

Fig. 7.

Same as in Fig. 2, but for low-level (0–3 km AGL) vertical wind shear (m s−1).

Fig. 7.

Same as in Fig. 2, but for low-level (0–3 km AGL) vertical wind shear (m s−1).

Looking at 0–2 km AGL storm-relative wind speed (Fig. 8), the separation between sounding populations of negative and positive regions was less defined, as is expected based on the decreasing significance level of the difference in means from Tables 4 to 6. The negative sounding population was bimodal, with one mode (9.5 m s−1) greater than and the other mode (3.5 m s−1) less than the positive sounding mode (6.5 m s−1). The positive sounding population was skewed toward higher storm-relative wind speed values up to 17 m s−1. While there was more overlap for this parameter, the samples for the positive and negative regions were still distinct.

Fig. 8.

Same as in Fig. 2, but for 0–2 km AGL storm-relative wind speed (m s−1).

Fig. 8.

Same as in Fig. 2, but for 0–2 km AGL storm-relative wind speed (m s−1).

Finally, parameters related to buoyant or conditional instability were considered. The 850–500-hPa lapse rate sample for each region was distinct with negative (positive) region values skewed toward lower (higher) values (Fig. 9). Although there was some overlap in the lapse rate samples between 6.9° and 8.4°C km−1, an overwhelming majority of negative (positive) region lapse rates were less (greater) than 7.2°C km−1. Relative frequency histograms of CAPE between −10° and −40°C and NCAPE between the LFC and −40°C are presented in Figs. 10 and 11, respectively. There was noticeable overlap between negative and positive region CAPE values calculated between the −10° to −40°C levels with close but distinct modes (negative: 1100 J kg−1; positive: 1300 J kg−1). There was less overlap in the samples of NCAPE from the LFC to the −40°C level for the two regions (negative mode: 0.21 m s−2; positive mode: 0.26 m s−2). CAPE and NCAPE frequency histograms calculated for other layers of the atmosphere considered in this study (not shown but see Tables 7 and 8 for means) were characterized by significantly more overlap than Figs. 10 and 11. Indeed, the samples of traditional (i.e., LFC–EL) CAPE calculated from soundings in the negative and positive mesoscale regions were indistinguishable. In summary, positive storms generally formed in regions characterized by larger 850–500-hPa lapse rates and larger NCAPE (i.e., mean parcel acceleration associated with buoyancy; Blanchard 1998) up to the top of the mixed-phase zone (i.e., −40°C).

Fig. 9.

Same as in Fig. 2, but for 850–500-hPa lapse rate (°C km−1).

Fig. 9.

Same as in Fig. 2, but for 850–500-hPa lapse rate (°C km−1).

Fig. 10.

Same as in Fig. 2, but for CAPE (J kg−1) between the −10° and −40°C levels.

Fig. 10.

Same as in Fig. 2, but for CAPE (J kg−1) between the −10° and −40°C levels.

c. Median environmental conditions associated with the mesoscale regions

To carefully investigate the hypothesis that regional CG lightning polarity is controlled by systematic differences in mesoscale environmental conditions (e.g., MacGorman and Burgess 1994), median values of select parameters, which were found to differ significantly between negative and positive regions in the overall comparisons (Tables 4 –6), were calculated for each of the nine mesoscale regions separately (Table 9 and Figs. 12 and 13). As was the case with the overall comparisons, WCD and LCL were the strongest discriminators between negative and positive mesoscale regions. The LCL values for all individual positive and negative regions were distinct with values greater than and less than 1600 m AGL, respectively (Table 9 and Fig. 12). The median WCD values for individual positive regions were all less than 2000 m, while the corresponding median values for negative regions were all greater than 2500 m (Table 9 and Fig. 13). Negative mesoscale regions were once again noticeably moister than positive mesoscale regions, as indicated by the median mixing ratio in the lowest 100 hPa, the PW between the surface and 400 hPa, and the surface dewpoint depression (Table 9). Median PW in negative regions was >3.3 cm while it was <3.0 cm in positive regions. Median surface dewpoint depression was consistently <11°C in negative regions and ≥11°C in positive regions, consistent with lower and higher LCLs in each region, respectively (W05). There was slight overlap in the median low-level mixing ratios between the two types of regions. Lapse rates between 850 and 500 hPa proved to be consistently higher in positive mesoscale regions. Median lapse rates in the 850–500-hPa layer were greater (less) than 7.5°C km−1 for all individual positive (negative) mesoscale regions (Table 9 and Fig. 13). Although grouped positive mesoscale regions were generally characterized by larger 0–3 km AGL shear, 0–2 km AGL storm-relative wind speed, NCAPE (LFC to −40°C), and CAPE (−10° to −40°C) when compared to grouped negative mesoscale regions (Tables 5 and 6), there was considerable overlap between the median parameter values of individual negative and positive mesoscale regions (Table 9 and Figs. 12 and 13).

Table 9.

Median parameter values for the nine individual mesoscale regions investigated. Parameters listed were found to differ significantly (i.e., significance level ≥95% or p ≤ 0.05) between negative and positive regions in the overall group comparisons (Tables 4 –6). However, only median values are directly compared here due to the small sample size for each individual region.

Median parameter values for the nine individual mesoscale regions investigated. Parameters listed were found to differ significantly (i.e., significance level ≥95% or p ≤ 0.05) between negative and positive regions in the overall group comparisons (Tables 4 –6). However, only median values are directly compared here due to the small sample size for each individual region.
Median parameter values for the nine individual mesoscale regions investigated. Parameters listed were found to differ significantly (i.e., significance level ≥95% or p ≤ 0.05) between negative and positive regions in the overall group comparisons (Tables 4 –6). However, only median values are directly compared here due to the small sample size for each individual region.
Fig. 12.

Scatterplot of the median NCAPE between the LFC and −40°C level vs the height of the LCL for each of the nine mesoscale regions. The size of each bubble in the scatterplot is proportional to the median 0–2 km AGL storm-relative wind speed in each region, which is also indicated by the label on each bubble. The 23 May negative mesoscale region, which was the only region not characterized by widespread severe storms, is labeled as “nonsevere.”

Fig. 12.

Scatterplot of the median NCAPE between the LFC and −40°C level vs the height of the LCL for each of the nine mesoscale regions. The size of each bubble in the scatterplot is proportional to the median 0–2 km AGL storm-relative wind speed in each region, which is also indicated by the label on each bubble. The 23 May negative mesoscale region, which was the only region not characterized by widespread severe storms, is labeled as “nonsevere.”

Fig. 13.

Same as in Fig. 12, but for median 850–500-hPa lapse rate vs median WCD for each individual mesoscale region. The bubble size (and labels) represent the median 0–3 km AGL shear for each region.

Fig. 13.

Same as in Fig. 12, but for median 850–500-hPa lapse rate vs median WCD for each individual mesoscale region. The bubble size (and labels) represent the median 0–3 km AGL shear for each region.

The contribution of each parameter to modulating updraft strength and supercooled liquid water content in each region may not be equal, possibly explaining some of the overlap. One parameter may compensate for another. For example, on 15 June the positive region LCL (WCD) was only slightly higher (smaller) than the in the negative region on the same day. However, the positive region CAPE between −10° and −40°C (NCAPE from LFC to −40°C) was 2.9 (2.5) times larger than in the negative region on 15 June such that the updraft was apparently still stronger. Finally, it is important to recall that the sample size for each individual region is relatively small so that regional differences (or lack thereof) must be viewed with some caution.

A key finding of earlier case studies was the observation that storms on the same day passing over similar mesoscale regions produced similar CG lightning behavior (e.g., Branick and Doswell 1992; MacGorman and Burgess 1994; Smith et al. 2000). To investigate this issue, we compared the environmental conditions in positive and negative storm regions occurring adjacent to each other on the same day (e.g., we compared the parameters of the 23 May positive region to those of the 23 May negative region shown in Table 9). Opportunities for a daily comparison of positive and negative mesoscale regions were available on 23 and 24 May and 15 June. The median positive and negative region values of WCD, LCL, PW, dewpoint depression, and 850–500-hPa lapse rate were significantly different on a daily basis (Table 9). The daily regional differences for these parameters were also consistent with the overall grouped results (e.g., Tables 4 –6). When considering the rest of the environmental parameters in Table 9, the daily comparison of positive and negative mesoscale regions was somewhat mixed and not always consistent with the overall grouped results. For example, the median low-level mixing ratio was somewhat larger and the 0–3-km shear was somewhat smaller in the positive region on 15 June. Similarly, the median 0–2-km storm-relative wind speed, NCAPE (LFC to −40°C), and CAPE (−10° to −40°C) were all larger in the negative mesoscale region on 24 May. On the other hand, all of the medians listed in Table 9 for 23 May were distinctly different between the positive and negative mesoscale regions and consistent with the overall grouped results.

Sometimes the daily differences in medians between the positive and negative regions were quite dramatic (Table 9). On 23 and 24 May, the median WCD was 2.6 and 1.7 times smaller, the median LCL was 2.1 and 2.2 times higher, the median surface dewpoint depression was 2.2 and 2.9 times larger, and the median 0–3-km shear was 1.7 and 2.0 times larger in the positive mesoscale region, respectively. The median 0–2-km storm-relative wind speed was 2.0 times larger in the positive region on 23 May. On 15 June, the median NCAPE (LFC to −40°C) and CAPE (−10° to −40°C) values were 2.5 and 2.9 times larger in the positive region, respectively.

d. Intraregional variation of environmental condition—The positive region of 24 May 2002

Dropsondes released from the FI Learjet on 24 May permitted the intraregional investigation of a positive storm environment at a very high spatial and temporal resolution. The dropsondes were released in the eastern Texas panhandle and southwest Oklahoma (Fig. 14). The dropsonde line was approximately perpendicular to the surface boundaries associated with convective initiation (CI) and near-surface θe ridge, and roughly parallel to subsequent storm motion (not shown).4 The numbers of positive and negative CG lightning flashes during the 6-h period were accumulated in 25 km2 grid boxes over the region of interest. From this gridded data, we computed the fraction of positive CG flashes over all CG flashes. Since the storms initiated just to the east of the dry line and generally moved eastward, Fig. 14 essentially provides a view of the evolution of the positive CG fraction during the life cycle of the event. As shown in Fig. 14, CG lightning polarity transitioned from positive to negative as storms in the vicinity of the 24 May dropsondes moved from west to east across Oklahoma and away from the dryline. Although the dropsondes were dropped entirely in the positive mesoscale region (Fig. 14), the +CG percentage generally decreased from west to east within the identified positive region (Figs. 14 and 15).

Fig. 14.

Percent positive flashes (%, shaded as shown) for 2000 UTC 24 May 2002–0200 UTC 25 May 2002. Black diamonds indicate dropsonde locations (from left to right, dropsonde release times were 2022, 2025, 2031, 2034, 2037, 2043, and 2046 UTC). Dropsondes released at 2028 and 2040 UTC contained bad data and thus were not used. Orange rectangles (north–south dimension = 80 km, east–west dimension = 27.4 km) centered on dropsonde locations denote the areas for which CG flash characteristics (number of negative flashes, number of positive flashes, and percent of positive flashes) were determined. The area to the north and west of the dotted line is the positive mesoscale region. The plot is centered over the eastern Texas panhandle, north-central Texas, and southwest Oklahoma.

Fig. 14.

Percent positive flashes (%, shaded as shown) for 2000 UTC 24 May 2002–0200 UTC 25 May 2002. Black diamonds indicate dropsonde locations (from left to right, dropsonde release times were 2022, 2025, 2031, 2034, 2037, 2043, and 2046 UTC). Dropsondes released at 2028 and 2040 UTC contained bad data and thus were not used. Orange rectangles (north–south dimension = 80 km, east–west dimension = 27.4 km) centered on dropsonde locations denote the areas for which CG flash characteristics (number of negative flashes, number of positive flashes, and percent of positive flashes) were determined. The area to the north and west of the dotted line is the positive mesoscale region. The plot is centered over the eastern Texas panhandle, north-central Texas, and southwest Oklahoma.

Fig. 15.

(a) NCAPE, low-level (0–3 km AGL) shear and (b) WCD, LCL, freezing level, and 850–575-hPa lapse rate calculated from the 24 May 2002 dropsondes as a function of distance along the Learjet flight track. LCL and freezing level heights are AGL. The time of each dropsonde release is indicated and annotated by the dashed gray lines in the vertical. The number of −CG and +CG flashes and the percent of positive polarity for storms within each orange rectangular box in Fig. 14 are shown. The thick vertical dash–dotted line shows the location of CI for deep convection along the Learjet flight track as determined by satellite imagery.

Fig. 15.

(a) NCAPE, low-level (0–3 km AGL) shear and (b) WCD, LCL, freezing level, and 850–575-hPa lapse rate calculated from the 24 May 2002 dropsondes as a function of distance along the Learjet flight track. LCL and freezing level heights are AGL. The time of each dropsonde release is indicated and annotated by the dashed gray lines in the vertical. The number of −CG and +CG flashes and the percent of positive polarity for storms within each orange rectangular box in Fig. 14 are shown. The thick vertical dash–dotted line shows the location of CI for deep convection along the Learjet flight track as determined by satellite imagery.

Parameters that were found to differ significantly and consistently between negative and positive regions in the grouped and regional comparisons above were calculated for all of the good (i.e., containing no bad data) 24 May dropsondes from 2022 to 2046 UTC as shown in Fig. 15a (NCAPE and 0–3 km AGL shear) and Fig. 15b (850–575 hPa lapse rate,5 WCD, LCL, and freezing level). A cold front was located between the locations of the 2025 and 2031 UTC dropsondes. There was also a dryline intersecting the cold front, forming a triple point in the vicinity of the Learjet flight track (not shown). The dryline ran between the 2031 and 2034 UTC dropsondes in Fig. 15, and this is where CI occurred as determined from satellite imagery (not shown).

NCAPE rapidly increased in the vicinity of the dryline, where convection initiated, and peaked at about 0.11 m s−2 just 10–40 km eastward into the warmer and moister air (Fig. 15a). Although low-level shear was stronger rearward (i.e., westward) of the dryline and cold front, a relative maxima in the 0–3 km AGL shear (17.5 m s−1) was located just east of the dryline at 110 km (Fig. 15a). As a result, the developing convection on 24 May experienced initially increasing and near-peak values of NCAPE and low-level shear. The +CG flash percentage also rapidly increased eastward of the dryline to a maximum value of 73% at a point centered just 35 km eastward (i.e., at 135 km along the Learjet flight track), apparently in response to these elevated values of NCAPE and low-level shear. Moving farther eastward, the NCAPE decreased dramatically (up to 36%) and the low-level shear dropped slightly. At the same time, the +CG percentage also decreased significantly from the peak of 73% to only 24%, which is just below the subjective threshold required for positive storm status. The height of the LCL peaked just 20 km westward of the dryline (Fig. 15b). Initial convection on 24 May was associated with an LCL between the maximum of 2200 and 1600 m AGL, which was measured just 10 km eastward of the dryline where convection initiated (i.e., 110 km along the Learjet flight track). The LCL continued to decrease eastward, reaching 1000 m AGL at the last dropsonde located at 217 km along the Learjet flight track (i.e., about 120 km eastward of where convection initiated). The height of the freezing level gradually increased eastward along the flight track. Combining the LCL and freezing level heights on 24 May in Fig. 15b, the WCD increased noticeably eastward of the dryline from a value that was between 1300 and 2000 m where convection initiated to just above 2700 m about 120 km eastward of the point of CI. The 850–575-hPa lapse rates peaked at 8.4°C km−1 just west of where convection initiated, and steadily decreased to the east, reaching a value of 6.8°C km−1 at the last dropsonde location, about 120 km eastward of CI (Fig. 15b). The dramatic decrease in the +CG fraction eastward of the dryline on 24 May was accompanied by a slight increase in the freezing level, a significant lowering of the LCL, an associated noteworthy increase in the WCD, and a modest decrease in the 850–575-hPa lapse rate. These trends are consistent with the relationships found between CG flash polarity and these environmental parameters in the grouped and regional comparisons.

5. Summary and discussion

The analysis of proximity sounding data associated with both positive and negative CG dominant severe storms during IHOP_2002 over the central United States clearly demonstrated significant and systematic differences in their mesoscale environments. When compared to negative storms, positive storms occurred in environments associated with a drier low- to midlevel troposphere (i.e., lower surface dewpoint, mean mixing ratio in the lowest 100 hPa, and PW from the surface to 400 hPa), higher CBH (i.e., higher LCL), smaller (i.e., shallower) WCD, stronger conditional instability (i.e., larger 850–500- and 700–500-hPa lapse rates), larger 0–3 km AGL wind shear, stronger 0–2 km AGL storm-relative wind speed, larger CAPE/NCAPE in the mixed-phase zone (i.e., LFC to −40°C NCAPE, and larger −10° to −40°C CAPE). Differences in the WCD of positive and negative storms were by far the most dramatic, suggesting an important role for this parameter in controlling CG lightning polarity.

These results support the hypothesis that broader, stronger updrafts and larger supercooled liquid water contents in the mixed-phase zone of convection cause the positive charging of graupel and hail via NIC (e.g., Saunders and Peck 1998) and the subsequent formation of a midlevel (i.e., −10° to −20°C) positive charge region and enhanced production of +CG lightning as has been recently observed (Rust and MacGorman 2002; Lang et al. 2004; Wiens et al. 2005). Stronger updrafts in positive storms were apparently generated thermodynamically from larger and more efficiently utilized conditional instability/buoyancy in the mixed-phase zone and dynamically from stronger low-level shear and storm relative inflow, similar to the results of GW02. Higher CBH in positive storms likely resulted in broader updrafts and reduced entrainment, allowing more of the potentially available buoyancy (i.e., CAPE and NCAPE) to be realized and effectively causing stronger updrafts as was first suggested by W05. Additionally, the lower PW content in positive storms was shown to be generally associated with stronger peak updrafts in simulated convection by McCaul et al. (2005) due to reduced water loading, among other effects. The broader and stronger updrafts apparently generated larger supercooled liquid water contents in the mixed-phase zone of positive storms by suppressing coalescence and reducing dry air entrainment, thus explaining how environmental conditions can systematically control CG lightning polarity. Dramatically reduced WCD in positive storms apparently also increased the supercooled liquid water content by drastically reducing the rainout of available cloud water via collision–coalescence as was postulated by W05.

Although McCaul and Cohen (2002) found that LFC and not the LCL controlled overturning efficiency and the degree of entrainment in simulated deep convection, we found no statistically significant difference between the LFCs of positive and negative storm environments (Table 8). The mean LFC in both environments was just above the stated optimal range for overturning efficiency in McCaul and Cohen (2002). In their study, the strongest updraft case actually resulted from high LFC and low LCL (i.e., large LFC–LCL depth) because of the combined influence of increased overturning efficiency (high LFC) and the elimination of outflow dominance (low LCL). In our study, the LFC–LCL depth was significantly larger in the negative (mean of 1.6 km) storms as compared to the positive storms (mean of 0.7 km; Table 5). Of course, this result was primarily associated with the much lower LCL in negative storms since the LFC was comparable. All else being equal, this result would suggest that negative storms have more efficient overturning, stronger updrafts, and less dilution of cloud water and buoyancy according to McCaul and Cohen (2002), which would reject our hypothesis. It is also possible that other factors (e.g., larger NCAPE and 0–3-km shear in positive storms) masked this effect such that positive storms still had stronger, broader updrafts and higher supercooled liquid water contents. In McCaul and Cohen (2002) the LCL and LFC were equivalent in most simulations and the LFC was used as a proxy for the moist layer depth such that all levels beneath the LFC were characterized by equally enhanced θe and hence a most unstable CAPE (MUCAPE) layer. The examined soundings in our study did not share these idealized characteristics as there was typically CIN (i.e., LCL ≠ LFC) and the LFC was on average 200 hPa above the lifted parcel level (LPL) of the most unstable parcel in the lowest 300 hPa from which MUCAPE was calculated. Since the LFC was therefore not an adequate proxy for moist layer depth in our sample, the McCaul and Cohen (2002) conclusions regarding the role of LFC on overturning efficiency may not apply to our study. Since our sample was small (48 soundings), a future study using a larger dataset of real soundings is required to confirm this suggestion.

a. Large supercooled water contents in high based severe storms: An apparent paradox

In our hypothesized scenario there is an apparent paradox in associating larger supercooled liquid water contents with the higher CBH of positive storms. Because of their lower and warmer cloud bases and more moist boundary layers (higher low-level mixing ratio and dewpoint), negative storms have higher adiabatic liquid water and hence the potential for higher actual cloud water contents than positive storms, especially if the cores are nearly undiluted as one might expect in supercell convection.6 Hence, to argue that high cloud base (small WCD) positive storms have more supercooled (T ≤ 0°C) cloud water than negative storms, we must somehow compensate for more adiabatic condensate being available in negative storms in the first place. We (as have others) make the following arguments that might compensate for more available moisture in negative storms:

  1. High CBH (or LCL) in positive storms may translate into broader updrafts with less entrainment (e.g., McCarthy 1974) since the convective bubble scales like the CBH or cumulus-topped boundary layer depth (e.g., Kaimal et al. 1976). Similar suggestions have been made by Michaud (1996, 1998), Lucas et al. (1994, 1996), Williams and Stanfill (2002), Zipser (2003), and W05. Less entrainment would result in less dilution of cloud water and buoyancy in positive storms. Hence, there would also be more efficient processing of CAPE and stronger updrafts in positive storms, all else being equal.

  2. Lower (higher) cloud bases in negative (positive) storms means greater (lesser) depth through which mixing of environmental air can reduce cloud buoyancy and water as pointed out by Michaud (1996, 1998) and discussed by Lucas et al. (1996).

  3. In general CAPE between positive and negative storms was very similar in our study. However, CAPE between −10° and −40°C and NCAPE in nearly all layers was stronger in positive storms providing stronger vertical accelerations in the mixed-phase zone for NIC. As also found in GW02, low-level (0–3 km) shear was larger and hence dynamic pressure forces may be stronger in positive storms (e.g., Weisman and Klemp 1984). In combination with potentially more efficient processing of available CAPE (see above), the updraft in positive storms may be stronger. Postulated stronger updrafts in positive storms would have the effect of suppressing precipitation and increasing the cloud/precipitation fraction as was recently suggested by W05 in a similar context and as has long been known in the severe storm community (e.g., the echo-free vault; Browning 1964, 1965).

  4. Lower cloud base (and to a lesser extent higher freezing height) in negative storms results in significantly larger WCD. Larger WCD would tend to increase the efficiency of warm rain–collision–coalescence processes (e.g., Rosenfeld and Woodley 2003). Increased coalescence results in lowering of the cloud/precipitation fraction. Once precipitation is formed, this condensate is no longer available to be lofted into the mixed-phase zone as cloud water where it can affect storm electrification (Williams et al. 2002; W05). Enhanced condensate in the warm portion of negative storms can also increase water loading and frictional drag and reduce the updraft velocity before the rain falls out, which feeds back into the points above.

  5. Although less convincing from the evidence presented, there was an observed tendency for more supercell characteristics in positive over negative storms (Table 3). From radar, supercell and multicell characteristics are evident in both positive and negative storms. However, inspection of loops of low-level radar reflectivity from individual and regional composite Weather Surveillance Radar–1988 Doppler (WSR-88D) imagery suggested that supercells were more common in positive storms. Limited analysis of WSR-88D Doppler velocity using the Warning Decision Support System–Integrated Information (WDSS-II) software revealed significantly higher mesocyclone frequency and intensity for the 15 June positive storms as compared to the 15 June negative storms (not shown). This finding suggests that supercell characteristics and hence dynamical forcing were stronger and more common in positive storms, at least on this particular day. This result is generally consistent with the environmental data (e.g., lower BRN, larger 0–3-km shear, larger 0–3-km storm relative helicity, larger EHI, and roughly equivalent CAPE and 0–6-km shear), which encourage pressure perturbation dynamics associated with quasi-steady (supercell) forcing of the updraft in positive storms. This dynamical forcing alone could have 1) increased updraft strength and 2) decreased detrimental entrainment of buoyancy and cloud water in positive storms, thus feeding back on several issues above.

The above suggestions are merely hypotheses based on plausible physical mechanisms and not observed facts. Nonetheless, the presented environmental data is consistent with these hypotheses. In the meantime, they wait further testing with in situ observations and numerical cloud models to evaluate whether these compensating factors increase (decrease) the (cloud water)/(precipitation water) and the (actual supercooled water content)/(adiabatic supercooled water content) fractions in positive (negative) storms sufficiently to result in larger absolute supercooled cloud water contents in positive storms. To understand how much of a compensating effect might be necessary, it is worthwhile to consider differences in the mean adiabatic liquid water content for the examined positive and negative storms. The mean adiabatic liquid water contents at 0°, −20°, and −40°C for negative storms were 6.0, 10.5, and 12.9 g kg−1, respectively, while for positive storms they were 3.3, 7.0, and 9.8 g kg−1, respectively. Note that the difference decreases significantly from the bottom (0°C) to the top (−40°C) of the mixed-phase zone. In the middle of the mixed-phase zone (−20°C) where NIC is the most effective, the mean adiabatic liquid water content in positive storms is about two-thirds of the magnitude in negative storms, providing some estimate of the required compensating effects. Also note that the results confirm that potential water loading effects in the warm portion of the cloud are significantly higher in negative storms, as expected.

b. The relative roles of the various environmental factors in controlling CG lightning polarity

It is important to note that not all of the environmental factors summarized above were always equal in importance for explaining why each individual positive (negative) storm was apparently associated with stronger/broader (weaker/narrower) updrafts and larger (smaller) liquid water contents in the mixed-phase zone. Sometimes one environmental factor seemingly compensated for another (Table 9 and Figs. 12 and 13), as would be expected, since updraft forcing and the production of large supercooled liquid water contents can come from any combination of the mechanisms discussed above. Nonetheless, based on the observational evidence from IHOP_2002, the low-level moisture, LCL, and WCD must be highlighted as likely the most important environmental factors for determining the dominant CG polarity in severe storms. Furthermore, the crucial role of the WCD may help explain the apparent paradox that most severe storms actually produce predominantly −CG lightning (CRP03) despite also likely being associated with vigorous vertical drafts. As pointed out by W05, the climatological overlap of shallow WCD (i.e., high cloud base) and large instability likely defines the geographic distribution of positive storms shown in CRP03 and repeated in Fig. 1. Based on our results, we generally agree with this view but suggest that the occurrence of strong 0–3 km AGL wind shear be added to this union of environmental factors, since dynamic can equal or exceed thermodynamic forcing of the updraft within severe supercell storms over the central United States. In summary, sufficient moisture, instability, and deep-layer shear in combination with enhanced NCAPE at heights below −40°C, enhanced dynamical forcing as seen in enhanced low-level shear and SREH, and a more efficient processing of CAPE and moisture up to and in the supercooled mixed-phase zone associated with higher LCL and smaller WCD is the likely explanation for positive storm occurrence in the central United States. As such, it should be clear that LCL or WCD alone cannot cause the conditions favorable for strong, broad updrafts and high supercooled water contents. For example, in the dry conditions of the desert Southwest CBH is very high and WCD very shallow or perhaps zero but yet positive storms are not common there. Clearly, this is because the other necessary conditions such as sufficient moisture and CAPE are not met. It is interesting to note in Fig. 1 and CRP03 that a few localized areas in the desert Southwest and Intermountain West experience a large fraction of positive storms. The sample size of severe storm reports there is small so the result may not be significant. Nonetheless, it would be interesting to conduct a similar study in those areas under the proper conditions to see if our results are confirmed.

It is important to note that several of the analyzed environmental parameters are highly correlated (i.e., or not independent such as surface dewpoint temperature, low-level mixing ratio, LCL, and WCD). Hence it could be difficult to assign causality to any one parameter individually. In a future study, it would be worthwhile to construct a multiple-linear regression model and analysis of variance to determine the relative importance of several parameters simultaneously and to eliminate redundant variables. Furthermore, other potential hypotheses explaining CG lightning polarity independent of the LCL and WCD but related in some other fashion to the role of low-level moisture on cloud electrification and lightning, which are coupled closely to mixed-phase cloud microphysical processes, are possible.

As pointed out by CRP03, it is not entirely clear from Smith et al. (2000) why storms should be positive dominant in the θe gradient region. We suggest that the θe gradient region (on the western and northern sides) of the ridge in moist entropy is a preferred area for positive storms (Smith et al. 2000; CRP03) not simply because of a difference in θe or CAPE values,7 but rather because the combination of environmental parameters we found to be favorable for +CG production are likely to be found in unison there, including sufficient low-level moisture, ample instability that tends to be more concentrated in the supercooled mixed-phase zone, high LCL, shallow WCD, and large 0–3-km shear. It is important to note that this combination of variables is consistent with the possibility of stronger updrafts and supercooled liquid water contents in positive storms, as required by the inverted-dipole hypothesis.

c. The potential role of aerosols

Because of insufficient aerosol data, we have not explored the potential effect of cloud condensation nuclei (CCN) on cloud electrification and CG lightning polarity. According to W05, CCN enhancements in the boundary layer could lead to a reduction in droplet size, the suppression of coalescence (e.g., Williams et al. 2002; Rosenfeld and Woodley 2003), a resulting increase in the supercooled liquid water content, and hence the production of enhanced +CG lightning as discussed above. This mechanism may explain the reported relationship between anomalously high aerosol concentrations associated with biomass burning and the occurrence of positive storms over the southern plains during the spring of 1998 (Lyons et al. 1998; Murray et al. 2000). As pointed out by W05 and shown by Lang and Rutledge (2006), the roles of extraordinary thermodynamic, dynamic, and CCN conditions are often difficult to distinguish since they often occur simultaneously. Nonetheless, Lang and Rutledge (2006) show that +CG lightning enhancements downwind of a forest fire in Colorado were more consistent with a causative role for elevated CBH and reduced WCD than for smoke aerosols. To confirm our results and to evaluate the potential impact of CCN on CG lightning polarity, we suggest a special focus be placed on measuring the simultaneous thermodynamic, wind, and aerosol properties of inflow air into deep convection, including severe storms, in a variety of conditions.

d. Environmental control of supercell type and CG polarity: A comparison of results

There are three supercells types including LP (e.g., Bluestein and Parks 1983), classic (e.g., Browning 1964), and high precipitation (HP; e.g., Doswell and Burgess 1993) that have been identified and defined by the amount of precipitation they produce and where the precipitation is deposited relative to their respective updrafts and mesocyclones (e.g., Doswell and Burgess 1993; Rasmussen and Straka 1998). The distribution and amount of precipitation in supercells is thought to have important implications for storm microphysics that could affect supercell dynamics, including tornadogenesis (e.g., Brooks et al. 1994a) and cloud electrification (e.g., MacGorman and Burgess 1994).

Early observations showed LP supercell storms to be +CG dominant (Curran and Rust 1992; Branick and Doswell 1992) and HP supercells to be −CG dominant (Branick and Doswell 1992), fueling speculation that a potentially exclusive relationship between supercell type and CG lightning polarity might exist. However, Bluestein and MacGorman (1998) present observations of an LP supercell that was dominated by −CG lightning. They also mention unpublished observations of two HP supercells dominated by +CG lightning. Classic supercells appear to be primarily −CG dominant but can also be associated with +CG dominant behavior (MacGorman and Nielsen 1991; MacGorman and Burgess 1994). The HP supercells are typically associated with dominant −CG lightning (e.g., Branick and Doswell 1992; MacGorman and Burgess 1994; Knupp et al. 2003). Transitions from LP to classic (and classic to HP) supercell types are often associated with corresponding transitions from positive to negative CG-dominant lightning behavior (Seimon 1993; MacGorman and Burgess 1994). Dominant negative CG supercells that transition from classic to HP type have also been associated with increasing −CG lightning production (Knupp et al. 2003). Multicellular storms (sometimes with embedded supercells) have been observed to produce primarily +CG lightning (MacGorman and Burgess 1994; Carey and Rutledge 1998) even though most multicell storms produce primarily −CG lightning. In summary, although there is a general tendency for LP (HP) supercells to be positive (negative) CG dominant, there are exceptions to this tendency as noted by Bluestein and MacGorman (1998).

There are two studies on the relationship between environmental conditions and supercell type (LP, classic, HP) that are relevant to our discussion: Bluestein and Parks (1983) and Rasmussen and Straka (1998). Bluestein and Parks (1983) found that the LCL was significantly higher in LP (1.8 km AGL) than classic (1.4 km AGL) supercells. Mean mixing ratio in the lowest kilometer was correspondingly lower in LP (11.9 g kg−1) versus classic (13.5 g kg−1) supercells. The PW was lower in LP (2.8 cm) than classic (3.3 cm) supercells. All of these results are consistent with the fact that LP supercells are most common near the dryline as were two of the four positive storms regions in this study (23 and 24 May). On the other hand, Rasmussen and Straka (1998) did not find statistically significant differences in the LCL or low-level mixing ratio between the three supercell types. They did find that PW was higher in HP versus LP and classic supercells. Rasmussen and Straka (1998) also found that upper-level storm relative flow at 9–10 km is stronger in LP storms and comparatively weak in HP storms with classic supercells falling in between. They speculate that precipitation efficiency is relatively lowered (raised) in LP (HP) storms due to the decreased (increased) recirculation of hydrometeors into the supercell updraft associated with the stronger (weaker) anvil-level winds.

Given the general tendency for LP (classic) supercells to be associated with positive (negative) storms, our results are in good agreement with the Bluestein and Parks (1983) findings regarding the meteorological environments of LP and classic supercells. A comparison with Rasmussen and Straka (1998) is less encouraging as only the PW results appear to be consistent with our study (i.e., high PW is associated with HP supercells and negative storms as expected). As a matter of fact, we inspected upper-level storm relative wind speed based on their results and the known tendencies between supercell type and CG lightning polarity. We found no statistically significant difference between the 9–11-km storm relative wind speed of positive and negative storms during IHOP_2002. Reasons for the discouraging comparison with Rasmussen and Straka (1998) are unknown but they noted that their soundings may not have adequately sampled the low-level moisture conditions of the supercells. We also note that there are known exceptions regarding the general tendencies of CG polarity with supercell type so any comparison between our study and supercell environment studies must be viewed with some caution.

Nonetheless, the fairly consistent relationship between supercell type and CG polarity can apparently be explained by at least one common link: precipitation efficiency. By definition, LP supercells are extremely precipitation inefficient, implying that a relatively large fraction of available condensate remains in cloud form (both water and ice) and is not converted to precipitation. As suggested earlier, the large implied (cloud water)/(precipitation water) fraction in LP supercells could allow for the positive NIC of precipitation ice, the generation of a midlevel positive charge layer, and positive lightning. Based on our results and Bluestein and Parks (1983), we hypothesize that high CBH and shallow WCD in LP storms is likely an important causal factor for the precipitation inefficiency, associated LP structure, and +CG lightning, although other factors likely play a role. The HP supercell is by definition a relatively efficient converter of cloud water to precipitation and hence would be characterized by a relatively smaller (cloud water)/(precipitation water) fraction. Although inconsistent with Rasmussen and Straka (1998), we suggest that relatively low CBH and deep WCD might be preferentially associated with HP supercells and hence could be an important causal mechanism for their precipitation efficiency and −CG lightning production. Of course, other physical factors, such as those highlighted by Rasmussen and Straka (1998), may also be important for both phenomena.

e. Tornadoes, CG lightning polarity reversals, and the LCL

As discussed earlier, CG lightning polarity reversals have been documented in some supercell storms (e.g., Seimon 1993; MacGorman and Burgess 1994; Perez et al. 1997; Smith et al. 2000). The polarity reversals documented in these studies are typically from dominant positive to dominant negative CG lightning. Although not common, an abrupt polarity shift from mostly positive to mostly negative CG lightning is sometimes associated with tornadogenesis and a significant (F2 or greater) tornado on the ground (Seimon 1993; MacGorman and Burgess 1994; Perez et al. 1997). Although the potential causative factors for both significant tornadoes and CG lightning polarity are many, complex, and the subject of current debate, our study in combination with recent studies on the environmental conditions associated with significant tornadic supercells may help explain the occasional coincidence between tornadogenesis and CG lightning polarity reversals. Among other factors, Rasmussen and Blanchard (1998) found that LCL was one of the best environmental discriminators between supercells that produce significant tornadoes (tornadic) and those that do not (nontornadic). An LCL ≤ 800 m was associated with significant tornadoes and ≥ 1200 m was associated with decreasing likelihood of significant tornadoes. More recent studies (Craven et al. 2002b; Thompson et al. 2003; Rasmussen 2003) have confirmed the strong correlation between significant tornadoes in supercells and low LCL, although Rasmussen (2003) raises some concerns, which are associated with his methodology for isolating supercells, regarding the causality of the result. Since our study shows that low LCL is highly correlated with dominant negative CG lightning, it is possible that sudden CG polarity shifts from dominant positive to dominant negative are associated with a rapid decrease in the LCL (increase in the low-level humidity) and hence associated with an increased probability of a significant tornado. Clearly, future studies should continue to focus on the correlation between low-level moisture and supercell type, tornado potential, and CG lightning polarity with an emphasis on observing and modeling the physical and dynamical factors that could confirm or reject causal relationships. Based on the strong spatial correlation between high (intracloud) IC–CG lightning ratio and positive CG fraction in the central United States (Boccippio et al. 2001), we would also add total lightning flash rate and the IC–CG lightning ratio to the list of storm properties that may be at least partially controlled by low-level humidity or CBH, as did W05.

f. Rejecting the tilted-dipole hypothesis

Table 8 provides convincing evidence rejecting the tilted-dipole hypothesis (e.g., Brook et al. 1982) for explaining dominant positive CG lightning in severe convection. The lack of a significant difference of the deep-layer (0–6 km) shear and storm-relative wind speeds in the mid- to upper troposphere (4–6, 6–10, 9–11 km AGL; EL) between positive and negative storms makes it difficult to imagine how a positive storm could tilt the thunderstorm dipole more than a negative storm. Although some early studies seem to favor a role for the tilted-dipole hypothesis in supercells (e.g., Curran and Rust 1992; Branick and Doswell 1992) and mesoscale convective systems (MCSs; e.g., Orville et al. 1988; Stolzenburg 1990), early evidence rejecting this hypothesis can be found in Reap and MacGorman (1989), who found no clear relationship between deep-layer shear and CG polarity using observational and model-generated environmental data. Even Curran and Rust (1992) could only claim that strong deep-layer shear was likely a necessary yet not sufficient condition for the production of +CG lightning. Recent in situ/balloon E-field and VHF-based 3D total lightning observations in supercells (e.g., Rust and MacGorman 2002; MacGorman et al. 2005; Wiens et al. 2005) have inferred vertical charge profiles that support the presence of an enhanced midlevel positive charge (in a generally inverted charge structure) as the most common source for positive CG flashes in many High Plains supercells that produce dominant positive CG lightning and not a tilted upper-level positive charge source as originally hypothesized for winter storms by Brook et al. (1982). Using radar, NLDN, and VHF-based 3D total lightning observations, Carey et al. (2005) demonstrated that the bulk of the +CG lightning production and inferred charge structure in an MCS was inconsistent with the basic premise of the tilted dipole hypothesis. Results herein merely add to the accumulating evidence from earlier environmental and more recent field campaign studies that reject the tilted-dipole hypothesis.

Acknowledgments

We wish to acknowledge fruitful discussions with Earle Williams, Walter Petersen, Kenneth Cummins, Daniel Rosenfeld, Timothy Lang, and Donald MacGorman. We are grateful to Earle Williams for providing early access to ongoing research. Earle Williams, Ted Mansell, and an anonymous reviewer provided meticulous and helpful reviews that improved the manuscript. The NLDN lightning data were obtained from Vaisala, Inc., under the direction of Jerry Guynes of TAMU. We thank Brandon Ely and Scott Steiger for assistance with IDL code. This research was supported by National Science Foundation Grants ATM-0233780 and ATM-0442011.

REFERENCES

REFERENCES
Blanchard
,
D. O.
,
1998
:
Assessing the vertical distribution of convective available potential energy.
Wea. Forecasting
,
13
,
870
877
.
Bluestein
,
H. B.
, and
C. R.
Parks
,
1983
:
A synoptic and photographic climatology of low-precipitation severe thunderstorms in the southern plains.
Mon. Wea. Rev.
,
111
,
2034
2046
.
Bluestein
,
H. B.
, and
D. R.
MacGorman
,
1998
:
Evolution of cloud-to-ground lightning characteristics and storm structure in the Spearman, Texas, tornadic supercells of 31 May 1990.
Mon. Wea. Rev.
,
126
,
1451
1467
.
Boccippio
,
D. J.
,
K. L.
Cummins
,
H. J.
Christian
, and
S. J.
Goodman
,
2001
:
Combined satellite- and surface-based estimation of the intracloud–cloud-to-ground lightning ratio over the continental United States.
Mon. Wea. Rev.
,
129
,
108
122
.
Bolton
,
D.
,
1980
:
The computation of equivalent potential temperature.
Mon. Wea. Rev.
,
108
,
1046
1053
.
Branick
,
M. L.
, and
C. A.
Doswell
III
,
1992
:
An observation of the relationship between supercell structure and lightning ground-strike polarity.
Wea. Forecasting
,
7
,
143
149
.
Brook
,
M.
,
M.
Nakano
,
P.
Krehbiel
, and
T.
Takeuti
,
1982
:
The electrical structure of the Hokuriku winter thunderstorms.
J. Geophys. Res.
,
87
,
1207
1215
.
Brooks
,
H. B.
,
C. A.
Doswell
III
, and
R. B.
Wilhelmson
,
1994a
:
The role of midtropospheric winds in the evolution and maintenance of low-level mesocyclones.
Mon. Wea. Rev.
,
122
,
126
136
.
Brooks
,
H. B.
,
C. A.
Doswell
III
, and
J.
Cooper
,
1994b
:
On the environments of tornadic and nontornadic mesocyclones.
Wea. Forecasting
,
9
,
606
618
.
Browning
,
K. A.
,
1964
:
Airflow and precipitation trajectories within severe local storms which travel to the right of the winds.
J. Atmos. Sci.
,
21
,
634
639
.
Browning
,
K. A.
,
1965
:
Some inferences about the updraft within a severe local storm.
J. Atmos. Sci.
,
22
,
669
677
.
Carey
,
L. D.
, and
S. A.
Rutledge
,
1998
:
Electrical and multiparameter radar observations of a severe hailstorm.
J. Geophys. Res.
,
103
,
13979
14000
.
Carey
,
L. D.
, and
S. A.
Rutledge
,
2003
:
Characteristics of cloud-to-ground lightning in severe and nonsevere storms over the central United States from 1989–1998.
J. Geophys. Res.
,
108
.
4483, doi:10.1029/2002JD002951
.
Carey
,
L. D.
,
W. A.
Petersen
, and
S. A.
Rutledge
,
2003a
:
Evolution of cloud-to-ground lightning and storm structure in the Spencer, SD, supercell of 30 May 1998.
Mon. Wea. Rev.
,
131
,
1811
1831
.
Carey
,
L. D.
,
S. A.
Rutledge
, and
W. A.
Petersen
,
2003b
:
The relationship between severe storm reports and cloud-to-ground lightning polarity in the contiguous United States from 1989–98.
Mon. Wea. Rev.
,
131
,
1211
1228
.
Carey
,
L. D.
,
M. J.
Murphy
,
T. L.
McCormick
, and
N. W. S.
Demetriades
,
2005
:
Lightning location relative to storm structure in a leading-line, trailing-stratiform mesoscale convective system.
J. Geophys. Res.
,
110
.
D03105, doi:10.1029/2003JD004371
.
Craven
,
J. P.
,
R. E.
Jewell
, and
H. E.
Brooks
,
2002a
:
Comparison between observed convective cloud-base heights and lifting condensation level for two different lifted parcels.
Wea. Forecasting
,
17
,
885
890
.
Craven
,
J. P.
,
H. E.
Brooks
, and
J. A.
Hart
,
2002b
:
Baseline climatology of sounding derived parameters associated with deep, moist convection. Preprints, 21st Conf. on Severe Local Storms, San Antonio, TX, Amer. Meteor. Soc., 643–646
.
Cummins
,
K. L.
,
M. J.
Murphy
,
E. A.
Bardo
,
W. L.
Hiscox
,
R. B.
Pyle
, and
A. E.
Pifer
,
1998
:
A combined TOA/MDF technology upgrade of the U.S. National Lightning Detection Network.
J. Geophys. Res.
,
103
,
9035
9044
.
Cummins
,
K. L.
,
J. A.
Cramer
,
C. J.
Biagi
,
E. P.
Krider
,
J.
Jerauld
,
M.
Uman
, and
V.
Rakov
,
2006
:
The U.S. National Lightning Detection Network: Post-upgrade status. Preprints, Second Conf. on the Meteorological Applications of Lightning Data, Atlanta, GA, Amer. Meteor. Soc., CD-ROM, P6.1
.
Curran
,
E. B.
, and
W. D.
Rust
,
1992
:
Positive ground flashes produced by low-precipitation thunderstorms in Oklahoma on 26 April 1984.
Mon. Wea. Rev.
,
120
,
544
553
.
Davies-Jones
,
R. P.
,
D.
Burgess
, and
M.
Foster
,
1990
:
Test of helicity as a tornado forecast parameter. Preprints, 16th Conf. on Severe Local Storms, Kananaskis Park, AB, Canada, Amer. Meteor. Soc., 588–592
.
Doswell
III,
C. A.
, and
D. W.
Burgess
,
1993
:
Tornadoes and tornadic storms: A review of conceptual models. The Tornado: Its Structure, Dynamics, Prediction, and Hazards, Geophys. Monogr., Vol. 28, Amer. Geophys. Union, 161–172
.
Doswell
III,
C. A.
, and
E. N.
Rasmussen
,
1994
:
The effect of neglecting the virtual temperature correction on CAPE calculations.
Wea. Forecasting
,
9
,
619
623
.
Gilmore
,
M. S.
, and
L. J.
Wicker
,
2002
:
Influences of the local environment on supercell cloud-to-ground lightning, radar characteristics, and severe weather on 2 June 1995.
Mon. Wea. Rev.
,
130
,
2349
2372
.
Hart
,
J. A.
, and
W. D.
Korotky
,
1991
:
The SHARP workstation v1. 50 user’s guide. National Weather Service, NOAA, 30 pp. [Available from NWS Eastern Region Headquarters, Scientific Services Division, 630 Johnson Ave., Bohemia, NY 11716.]
.
Heckert
,
N. A.
, and
J. J.
Filliben
,
2003
:
NIST Handbook 148: DATAPLOT reference manual, Volume I: Commands. National Institute of Standards and Technology Handbook Series, June 2003. [Available online at http://www.itl.nist.gov/div898/software/dataplot/refman1/homepage.htm.]
.
Houze
Jr.,
R. A.
,
1993
:
Cloud Dynamics.
Academic Press, 573 pp
.
Jayaratne
,
E. R.
,
C. P. R.
Saunders
, and
J.
Hallett
,
1983
:
Laboratory studies of the charging of soft-hail during ice crystal interactions.
Quart. J. Roy. Meteor. Soc.
,
109
,
609
630
.
Jorgensen
,
D. P.
, and
M. A.
LeMone
,
1989
:
Vertical velocity characteristics of oceanic convection.
J. Atmos. Sci.
,
46
,
621
640
.
Kaimal
,
J. C.
,
J. C.
Wyngaard
,
D. A.
Haugen
,
O. R.
Cote
,
Y.
Izumi
,
S. J.
Caughey
, and
C. J.
Readings
,
1976
:
Turbulence structures in the convective boundary layer.
J. Atmos. Sci.
,
33
,
2152
2169
.
Knapp
,
D. I.
,
1994
:
Using cloud-to-ground lightning data to identify tornadic thunderstorm signatures and nowcast severe weather.
Natl. Wea. Digest
,
19
,
35
42
.
Knight
,
C. A.
, and
N. C.
Knight
,
2001
:
Hailstorms. Severe Convective Storms, Meteor. Monogr., No. 50, Amer. Meteor. Soc., 223–248
.
Knupp
,
K. R.
,
S.
Peach
, and
S.
Goodman
,
2003
:
Variations in cloud-to-ground lightning characteristics among three adjacent tornadic supercell storms over the Tennessee valley region.
Mon. Wea. Rev.
,
131
,
172
188
.
Krehbiel
,
P. R.
,
R. J.
Thomas
,
W.
Rison
,
T.
Hamlin
,
J.
Harlin
, and
M.
Davis
,
2000
:
GPS-based mapping system reveals lightning inside storms.
Eos, Trans. Amer. Geophys. Union
,
81
,
21
25
.
Lang
,
T. J.
, and
S. A.
Rutledge
,
2002
:
Relationships between convective storm kinematics, precipitation, and lightning.
Mon. Wea. Rev.
,
130
,
2492
2506
.
Lang
,
T. J.
, and
S. A.
Rutledge
,
2006
:
Cloud-to-ground lightning downwind of the 2002 Hayman forest fire in Colorado.
Geophys. Res. Lett.
,
33
.
L03804, doi:10.1029/2005GL024608
.
Lang
,
T. J.
, and
Coauthors
,
2004
:
The Severe Thunderstorm Electrification and Precipitation Study.
Bull. Amer. Meteor. Soc.
,
85
,
1107
1126
.
Lucas
,
C.
,
E. J.
Zipser
, and
M. A.
LeMone
,
1994
:
Vertical velocity in oceanic convection off tropical Australia.
J. Atmos. Sci.
,
51
,
3183
3193
.
Lucas
,
C.
,
E. J.
Zipser
, and
M. A.
LeMone
,
1996
:
Reply.
J. Atmos. Sci.
,
53
,
1212
1214
.
Lyons
,
W. A.
,
T. E.
Nelson
,
E. R.
Williams
,
J.
Cramer
, and
T.
Turner
,
1998
:
Enhanced positive cloud-to-ground lightning in thunderstorms ingesting smoke.
Science
,
282
,
77
81
.
MacGorman
,
D. R.
, and
K. E.
Nielsen
,
1991
:
Cloud-to-ground-lightning in a tornadic storm on 8 May 1986.
Mon. Wea. Rev.
,
119
,
1557
1574
.
MacGorman
,
D. R.
, and
D. W.
Burgess
,
1994
:
Positive cloud-to-ground lightning in tornadic storms and hailstorms.
Mon. Wea. Rev.
,
122
,
1671
1697
.
MacGorman
,
D. R.
,
W. D.
Rust
,
P.
Krehbiel
,
W.
Krehbiel
,
W.
Rison
,
E.
Bruning
, and
K.
Wiens
,
2005
:
The electrical structure of two supercell storms during STEPS.
Mon. Wea. Rev.
,
133
,
2583
2607
.
McCarthy
,
J.
,
1974
:
Field verification of the relationship between entrainment rate and cumulus cloud diameter.
J. Atmos. Sci.
,
31
,
1028
1039
.
McCaul
Jr.,
E. W.
, and
M. L.
Weisman
,
2001
:
The sensitivity of simulated supercell structure and intensity to variations in the shapes of environmental buoyancy and shear profiles.
Mon. Wea. Rev.
,
129
,
664
687
.
McCaul
Jr.,
E. W.
, and
C.
Cohen
,
2002
:
The impact on simulated storm structure and intensity of variations in the mixed layer and moist layer depths.
Mon. Wea. Rev.
,
130
,
1722
1748
.
McCaul
Jr.,
E. W.
,
C.
Cohen
, and
C.
Kirkpatrick
,
2005
:
The sensitivity of simulated storm structure, intensity, and precipitation efficiency to environmental temperature.
Mon. Wea. Rev.
,
133
,
3015
3037
.
Michaud
,
L. M.
,
1996
:
Comments on “Convective available potential energy in the environment of oceanic and continental clouds.”.
J. Atmos. Sci.
,
53
,
1209
1211
.
Michaud
,
L. M.
,
1998
:
Entrainment and detrainment required to explain updraft properties and work dissipation.
Tellus
,
50A
,
283
301
.
Moller
,
A. R.
,
2001
:
Severe local storm forecasting. Severe Convective Storms, Meteor. Monogr., No. 50, Amer. Meteor. Soc., 433–480
.
Moncrieff
,
M. W.
, and
M. J.
Miller
,
1976
:
The dynamics and simulation of tropical cumulonimbus and squall lines.
Quart. J. Roy. Meteor. Soc.
,
102
,
373
394
.
Murray
,
N. D.
,
R. E.
Orville
, and
G. R.
Huffines
,
2000
:
Effect of pollution from Central American fires on cloud-to-ground lightning in May 1998.
Geophys. Res. Lett.
,
27
,
2249
2252
.
Orville
,
R. E.
, and
G. R.
Huffines
,
2001
:
Cloud-to-ground lightning in the United States: NLDN results in the first decade, 1989–98.
Mon. Wea. Rev.
,
129
,
1179
1193
.
Orville
,
R. E.
,
R. W.
Henderson
, and
L. F.
Bosart
,
1988
:
Bipole patterns revealed by lightning locations in mesoscale storm systems.
Geophys. Res. Lett.
,
15
,
129
132
.
Perez
,
A. H.
,
L. J.
Wicker
, and
R. E.
Orville
,
1997
:
Characteristics of cloud-to-ground lightning associated with violent tornadoes.
Wea. Forecasting
,
12
,
428
437
.
Rasmussen
,
E. N.
,
2003
:
Refined supercell and tornado forecast parameters.
Wea. Forecasting
,
18
,
530
535
.
Rasmussen
,
E. N.
, and
D. O.
Blanchard
,
1998
:
A baseline climatology of sounding-derived supercell and tornado forecast parameters.
Wea. Forecasting
,
13
,
1148
1164
.
Rasmussen
,
E. N.
, and
J. M.
Straka
,
1998
:
Variations in supercell morphology. Part I: Observations of the role of upper-level storm-relative flow.
Mon. Wea. Rev.
,
126
,
2406
2421
.
Reap
,
R. M.
, and
D. R.
MacGorman
,
1989
:
Cloud-to-ground lightning: Climatological characteristics and relationships to model fields, radar observations, and severe local storms.
Mon. Wea. Rev.
,
117
,
518
535
.
Rosenfeld
,
D.
, and
W. L.
Woodley
,
2003
:
Closing the 5-year circle: From cloud seeding to space and back to climate change through precipitation physics. Cloud Systems, Hurricanes, and the Tropical Rainfall Measuring Mission (TRMM), Meteor. Monogr., No. 51, Amer. Meteor. Soc., 59–80
.
Rotunno
,
R.
,
J. B.
Klemp
, and
M. L.
Weisman
,
1988
:
A theory for strong, long-lived squall lines.
J. Atmos. Sci.
,
45
,
463
485
.
Rust
,
W. D.
, and
D. R.
MacGorman
,
2002
:
Possibly inverted-polarity electrical structures in thunderstorms during STEPS.
Geophys. Res. Lett.
,
29
.
1571, doi:10.1029/2001GL014303
.
Saunders
,
C. P. R.
, and
S. L.
Peck
,
1998
:
Laboratory studies of the influence of the rime accretion rate on charge transfer during crystal/graupel collisions.
J. Geophys. Res.
,
103
,
13949
13956
.
Saunders
,
C. P. R.
,
W. D.
Keith
, and
R. P.
Mitzeva
,
1991
:
The effect of liquid water on thunderstorm charging.
J. Geophys. Res.
,
96
,
11007
11017
.
Seimon
,
A.
,
1993
:
Anomalous cloud-to-ground lightning in an F5-tornado-producing supercell thunderstorm on 28 August 1990.
Bull. Amer. Meteor. Soc.
,
74
,
189
203
.
Smith
,
S. B.
,
J. G.
LaDue
, and
D. R.
MacGorman
,
2000
:
The relationship between cloud-to-ground lightning polarity and surface equivalent potential temperature during three tornadic outbreaks.
Mon. Wea. Rev.
,
128
,
3320
3328
.
Stephens
,
M. A.
,
1974
:
EDF statistics for goodness of fit and some comparisons.
J. Amer. Stat. Assoc.
,
69
,
730
737
.
Stolzenburg
,
M.
,
1990
:
Characteristics of the bipolar pattern of lightning locations observed in 1988 thunderstorms.
Bull. Amer. Meteor. Soc.
,
71
,
1331
1338
.
Stolzenburg
,
M.
,
1994
:
Observations of high ground flash densities of positive lightning in summertime thunderstorms.
Mon. Wea. Rev.
,
122
,
1740
1750
.
Stolzenburg
,
M.
,
W. D.
Rust
, and
T. C.
Marshall
,
1998
:
Electrical structure in thunderstorm convective regions 3. Synthesis.
J. Geophys. Res.
,
103
,
14097
14108
.
Takahashi
,
T.
,
1978
:
Riming electrification as a charge generation mechanism in thunderstorms.
J. Atmos. Sci.
,
35
,
1536
1548
.
Thompson
,
R. L.
,
R.
Edwards
,
J. A.
Hart
,
K. L.
Elmore
, and
P.
Markowski
,
2003
:
Close proximity soundings within supercell environments obtained from the Rapid Update Cycle.
Wea. Forecasting
,
18
,
1243
1261
.
Wacker
,
R. S.
, and
R. E.
Orville
,
1999a
:
Changes in measured lightning flash count and return stroke peak current after the 1994 U.S. National Lightning Detection Network upgrade, 1, Observations.
J. Geophys. Res.
,
104
,
2151
2158
.
Wacker
,
R. S.
, and
R. E.
Orville
,
1999b
:
Changes in measured lightning flash count and return stroke peak current after the 1994 U.S. National Lightning Detection Network upgrade, 2, Theory.
J. Geophys. Res.
,
104
,
2159
2162
.
Weckwerth
,
T. M.
, and
Coauthors
,
2004
:
An overview of the International H2O Project (IHOP_2002) and some preliminary highlights.
Bull. Amer. Meteor. Soc.
,
85
,
253
277
.
Weisman
,
M. L.
, and
J. B.
Klemp
,
1982
:
The dependence of numerically simulated convective storms on vertical wind shear and buoyancy.
Mon. Wea. Rev.
,
110
,
504
520
.
Weisman
,
M. L.
, and
J. B.
Klemp
,
1984
:
The structure and classification of numerically simulated convective storms in directionally varying wind shears.
Mon. Wea. Rev.
,
112
,
2479
2498
.
Weisman
,
M. L.
, and
J. B.
Klemp
,
1986
:
Characteristics of isolated convective storms.
Mesoscale Meteorology and Forecasting, P. S. Ray, Ed., Amer. Meteor. Soc. 331–358
.
Weisman
,
M. L.
, and
R.
Rotunno
,
2004
:
“A theory for strong long-lived squall lines” revisited.
J. Atmos. Sci.
,
61
,
361
382
.
Wiens
,
K. C.
,
S. A.
Rutledge
, and
S. A.
Tessendorf
,
2005
:
The 29 June 2000 supercell observed during STEPS. Part II: Lightning and charge structure.
J. Atmos. Sci.
,
62
,
4151
4177
.
Wilks
,
D. S.
,
1995
:
Statistical Methods in the Atmospheric Sciences.
Academic Press, 467 pp
.
Williams
,
E. R.
,
1989
:
The tripole structure of thunderstorms.
J. Geophys. Res.
,
94
,
13151
13167
.
Williams
,
E. R.
,
2001
:
The electrification of severe storms. Severe Convective Storms, Meteor. Monogr., No. 50, Amer. Meteor. Soc., 527–561
.
Williams
,
E. R.
, and
N.
Renno
,
1993
:
An analysis of the conditional instability of the tropical atmosphere.
Mon. Wea. Rev.
,
121
,
21
36
.
Williams
,
E. R.
, and
S.
Stanfill
,
2002
:
The physical origin of the land-ocean contrast in lightning activity.
C. R. Phys.
,
3
,
1277
1292
.
Williams
,
E. R.
, and
Coauthors
,
2002
:
Contrasting convective regimes over the Amazon: Implications for cloud electrification.
J. Geophys. Res.
,
107
.
8082, doi:10.1029/2001JD000380
.
Williams
,
E. R.
,
V.
Mushtak
,
D.
Rosenfeld
,
S.
Goodman
, and
D.
Boccippio
,
2005
:
Thermodynamic conditions favorable to superlative thunderstorm updraft, mixed phase microphysics and lightning flash rate.
Atmos. Res.
,
76
,
288
306
.
Wilson
,
C. T. R.
,
1920
:
Investigations on lightning discharges and on the electric field of thunderstorms.
Philos. Trans. Roy. Soc. London
,
221A
,
73
115
.
Zipser
,
E. J.
,
2003
:
Some views on “Hot Towers” after 50 years of tropical field programs and two years of TRMM data.
Cloud Systems, Hurricanes, and the Tropical Rainfall Measuring Mission (TRMM), Meteor. Monogr., No. 51, Amer. Meteor. Soc., 49–58
.

APPENDIX

Selection of Proximity Sounding Data

As described in detail by Brooks et al. (1994b), obtaining proximity soundings that are representative of the meteorological conditions experienced by convection is not a trivial task. The goal in selecting proximity soundings is to select only those soundings that sampled the inflow air of the storm(s) of interest. However, spatial and temporal variability within the environments of severe storms is the rule, rather than the exception, and it is thus often difficult to obtain representative soundings. Fortunately, the multitude of soundings launched during IHOP_2002 mitigated this problem.

Three issues need to be considered and accounted for when selecting representative proximity soundings: 1) spatial variability of environmental conditions, 2) temporal variability of environmental conditions, and 3) the sampling by soundings of conditions that are not representative of the inflow air of the storm(s) of interest, due to factors such as convective contamination and the presence of boundaries (e.g., fronts, drylines, outflow boundaries; Brooks et al. 1994b). To account for the first and second issues, distance and time constraints of 100 km and 3 h, respectively, were used. To account for the third issue, all soundings satisfying the first and second criteria were individually inspected by hand for signatures of convective contamination (e.g., the lower troposphere cooled and stabilized by outflow, the upper troposphere moistened by anvils, the wind structure altered dramatically) and for signatures that the sounding sampled a different air mass than that in which the storms of interest developed (e.g., the sounding was launched on the opposite side of a front, dryline, or outflow boundary from where the convection developed) using surface observations, radar reflectivity imagery, visible satellite imagery, and ground flash plots. Since the proximity sounding dataset compiled strongly dictates the results of a study, great detail and care were taken in assembling this dataset. From hundreds of soundings launched on the 6 days investigated in this study, 48 soundings were chosen as representative inflow proximity soundings. Half of these (24) represent the mesoscale environments of positive storms and the other half (24) represent the environments of negative storms (Table 2).

In compiling a proximity sounding dataset, competing forces exist between assembling a large dataset by enforcing less stringent requirements on the soundings used, and assembling a dataset truly representative of storm inflow air by enforcing more stringent requirements on the soundings used. Naturally, the latter approach results in a smaller proximity sounding dataset (Brooks et al. 1994b). This study placed greater emphasis on using only those proximity soundings truly representative of inflow air than on the assemblage of a large dataset. Even so, the size of our proximity sounding dataset is sufficiently large to produce statistically significant and robust results. Nonetheless, future studies should strive to compile a larger sample of inflow proximity soundings associated with positive and negative storms in order to confirm the findings herein.

Footnotes

* Current affiliation: NOAA/National Weather Service, Weather Forecast Office, Hastings, Nebraska

Corresponding author address: Dr. Lawrence D. Carey, Department of Atmospheric Sciences, Texas A&M University, 3150 TAMU, College Station, TX 77843-3150. Email: larry_carey@tamu.edu

1

The definition of anomalous positive or positive storms is somewhat arbitrary in the literature with values of +CG fraction ranging from ≥25% to 50%. CRP03 found the geographic distribution of severe storms with this range of +CG fraction to be identical. Hence, we choose to use the less restrictive definition of ≥25% +CG fraction.

2

In ordinary (inverted) polarity storms, the highest charge layer, which is likely a screening layer, is negative (positive) as shown by Stolzenburg et al. (1998) (Rust and MacGorman 2002).

3

Cummins et al. (2006) reported that the recent NLDN upgrade increased the detection of low-amplitude flashes and thus the potential for misclassifying cloud flashes as ground discharges of either polarity. To test the sensitivity of our regional CG polarity classification to peak current thresholding, we removed all NLDN-detected CG flashes with peak currents less than 10, 15, and 20 kA (K. Cummins 2006, personal communication). The mean +CG percentage in + (−) CG polarity regions after removing all CG flashes with peak currents less than 10, 15, and 20 kA was 53%, 66%, and 81% (10%, 9%, and 11%), respectively. Since our CG polarity classification was insensitive to the choice of a peak current threshold, we have chosen to follow Cummins et al. (1998) until ongoing NLDN flash classification improvements are completed.

4

The initiation of deep convection in Figs. 14 and 15 was determined by manual inspection of the Geostationary Operational Environmental Satellite (GOES-8) infrared (IR) and visible imagery. The first CG lightning (including positive flashes) was within roughly 15–30 km eastward of the dryline and CI.

5

Dropsondes were sometimes dropped from pressures higher than 500 hPa. To maintain consistency for all lapse rates calculated from dropsondes, we slightly modified the lower pressure of the 850–500-hPa layer to 575 hPa.

6

Of course, it is common sense that this is not where the bulk of the electrification is occurring since there are not sufficient number concentrations of precipitation ice particles in the core updraft or weak echo region (WER) of a supercell. Verification of this common sense idea is the “lightning hole” or absence of lightning and inferred significant charge in VHF-based lightning observations within the supercell WER (e.g., Krehbiel et al. 2000). Since we are unsure of where the electrification is taking place relative to the draft structure, it would not be safe to assume that the parcel is undiluted where NIC is operative.

7

Recall that there was no significant difference between the mean positive and negative region θe and CAPE.