1. Introduction
Thundersnow events are rare phenomena that are said to occur when snow-bearing storms produce lightning and thunder. Because lightning in winter storms is less common and may seem counterintuitive, such events may take forecasters and the public by surprise and may pose an unexpected hazard. Though people tend to spend less time outdoors in winter months, lightning-related incidents have occurred at ski resorts, which host a variety of outdoor activities (e.g., Berger 1998; Zook 2014). Electrified winter clouds can pose an aviation hazard, as highly electrified areas have a high probability of causing aircraft- or helicopter-triggered lightning (e.g., Mäkelä et al. 2013; Wilkinson et al. 2013). Additionally, other winter weather hazards such as heavy snow have been associated with thundersnow events (Crowe et al. 2006). Thus, better understanding these unique storms could serve to mitigate potential risks.
Schultz and Vavrek (2009) provide a good review of historical observations of thundersnow events, which are limited because of their rarity. In the United States, early climatological studies of thundersnow events used surface reports. For example, Curran and Pearson (1971) investigated 76 thundersnow cases in the United States and found that only 1.3% of cool-season thunderstorms (i.e., those occurring between October and May) produced snow, and that only 0.07% of snowfall observations were associated with lightning or thunder. Further, Market et al. (2002) analyzed 3-hourly METARs to construct a climatology of thundersnow events. They reported only three incidences of thundersnow in northern Colorado over a 30-yr period from 1961 to 1990, and an average annual occurrence of only 6.3 events across the conterminous United States (CONUS). Their climatology was not meant to be exhaustive and, instead, probably better characterizes the broader synoptic distribution of thundersnow events, as the use of surface reports for thundersnow detection is limited by the spatial coverage and resolution of surface observing stations, population density, etc.
These studies were performed before data from nationwide lightning detection networks were readily available. These networks provide high-resolution spatiotemporal information on lightning events across the CONUS (e.g., Orville 1991; Cummins et al. 1998; Orville and Huffines 2001; Holle 2014). More recent studies have made use of such lightning detection network data to investigate various aspects of thundersnow events (e.g., Pettegrew 2008; Market and Becker 2009; Rauber et al. 2014). For example, Pettegrew (2008) analyzed Vaisala’s National Lightning Detection Network (NLDN) data from 14 central United States thundersnow cases between October 2006 and April 2007, and found that only 1.4% of flashes during these events were associated with winter precipitation.
Such CONUS networks mostly detect cloud-to-ground (CG) lightning, which only makes up a fraction of all lightning discharges (Cummins et al. 1998; Boccippio et al. 2000). In recent years, however, the detection efficiency of total lightning from these networks has been improving (e.g., Rudlosky and Fuelberg 2010). Additionally, the growing availability of regional total lightning detection systems has enhanced our detection capabilities. Such systems include the Lightning Mapping Array (LMA), developed by the New Mexico Institute of Mining and Technology (NMIMT), which detects both CG and in-cloud (IC) flashes with high detection efficiency and accuracy (Rison et al. 1999; Krehbiel et al. 2000; Thomas et al. 2004; Lang et al. 2004) within a 100–200-km range. Thus, with these systems it is possible to document events that previously may have gone undetected.
Previous studies have also explored the necessary atmospheric ingredients for thundersnow storms. Schultz and Vavrek (2009) point out that the same conditions needed for warm-season thunderstorms must be present in thundersnow events: moisture, lift, and an unstable temperature profile. Curran and Pearson (1971) showed that the mean environment of their thundersnow cases was supportive of elevated convection, with a stable boundary layer topped by a near-neutral thermal profile. Market et al. (2006) found similar results for thundersnow events in the central United States, with the most-unstable level roughly 30–50 hPa above the top of the low-level temperature inversion. Rauber et al. (2014) found that lightning in midwestern winter cyclones tended to occur in regions of convective instability in the comma-head region, associated with dry upper-tropospheric air overrunning moister air from the Gulf of Mexico. They suggest that conditional symmetric instability (CSI) likely did not play a role in driving the convection. In contrast to typical summer convective storms, thundersnow storms are found to have lower convective available potential energy (CAPE) (e.g., Market et al. 2006; Mäkelä et al. 2013) and, consequently, weaker updrafts and much lower vertical cloud extent.
In addition to an environment supporting convection, cold (<0°C) air throughout the precipitation-bearing layer is required to produce snow at the surface. The composite sounding in Market et al. (2006) was cold enough throughout the lower troposphere to support snow. The surface temperature in thundersnow events often is found to be very near 0°C (e.g., Schultz 1999; Hunter et al. 2001; Stuart 2001). Market et al. (2002) found the mean surface temperature in their thundersnow cases to be about −1°C. Model soundings from the long-lived thundersnow event studied by Market et al. (2007) were consistent with these previous findings, as well.
Thunderstorm electrification leading to lightning is typically associated with storms exhibiting a strong updraft and a robust mixed-phase region (Williams and Lhermitte 1983; Dye et al. 1988; Zipser and Lutz 1994; Wiens et al. 2005; Deierling et al. 2008). Moreover, collisions between hydrometeors such as graupel and ice crystals in the presence of supercooled liquid water (SLW) is the basis for significant charging to take place via the noninductive charging mechanism that is thought to play a dominant role in lightning production (Takahashi 1978; Saunders 1993; Saunders and Peck 1998; MacGorman and Rust 1998; Takahashi and Miyawaki 2002). However, in the absence of appreciable SLW, noninductive charging may also occur (e.g., Dye and Willett 2007; Kuhlman et al. 2009), albeit at lesser charging rates. Furthermore, other charging mechanisms including inductive charging may also contribute to cloud electrification (e.g., MacGorman and Rust 1998). As thundersnow cases documented in previous studies are associated with lower CAPE, yielding lower updraft strength and vertical cloud extent, one expects that the availability of SLW, charging, and thus resultant flash rates should also be lower compared to their warm-season convective storm counterparts. For reference, flash rates in warm-season convective storms may range from several flashes per minute to hundreds of flashes per minute in high-end severe cases (e.g., Williams et al. 2005; Deierling et al. 2008; Fuchs et al. 2015). In contrast, total lightning flash rates in a variety of thundersnow storms have yet to be quantified.
Several authors have also investigated the precipitation structure of thundersnow events using radar. Pettegrew et al. (2009) used data from a single-polarization WSR-88D to investigate a thundersnow event over eastern Iowa and north-central Illinois. The echo top of the storm in their case study never exceeded ~3.7 km AGL, and its maximum values of reflectivity factor at horizontal polarization 
However, 
Comprehensive information about the finescale electrical and microphysical structure of thundersnow is not available to date. In this paper, we report on four events that occurred in northern Colorado over a 5.5-month period during the 2012/13 cold season. For the first time, the electrical and microphysical thundersnow events are analyzed using lightning data from both the LMA and CONUS networks, and using dual-polarization WSR-88D data. During these four events, a total of 16 individual storms produced lightning. Though unclear whether the 2012/13 season was anomalous, it does hint at the possibility that thundersnow events are more common than previously documented [i.e., only 6.3 events on average annually in the CONUS; Market et al. (2002)], as observations from regional lightning detection networks such as the LMA detect CG and IC (i.e., total) lightning more comprehensively than do the CONUS networks used in the previous studies mentioned above.
This study aims to answer the following questions:
- How often are these storms detected by CONUS versus LMA networks?
- What types of dual-polarization radar signatures in these Colorado thundersnow storms may be relevant to operational forecasters?
- What are the electrical properties of these storms, and how do they compare to ordinary warm-season convective storms?
The next section provides an overview of the instrumentation and data used to analyze thundersnow events in this study. Section 3 presents an overview of the cases, and an analysis of the LMA and CONUS lightning data, followed by the detailed polarimetric radar data analysis in section 4. The paper finishes with a discussion and summary of the main conclusions in section 5.
2. Instrumentation and data
a. Dual-polarization WSR-88Ds
The National Weather Service WSR-88D network recently has undergone an upgrade to have dual-polarization capabilities. In addition to the conventional moments of 
Data from the WSR-88D polarimetric stations near Denver, Colorado (KFTG), and Cheyenne, Wyoming (KCYS), are used in this study. In addition to the polarimetric variables, the output of the operational hydrometeor classification algorithm (HCA) is used. The current HCA combines the informative contents of each polarimetric radar variable and determines the type of scatterer dominating the returned signals in each radar sampling volume (Park et al. 2009). Currently, 1 of 10 possible classes is assigned: light-to-moderate rain, heavy rain, big drops, rain mixed with hail, graupel, wet snow, dry snow aggregates, ice crystals, biological scatterers, and ground clutter and/or anomalous propagation.
b. Colorado Lightning Mapping Array
c. CONUS networks
Lightning captured by lightning detection networks covering the CONUS are used herein as well. These include Earth Networks Total Lightning Network (ENTLN), the United States Precision Lightning Network (USPLN) owned by Weather Services International (WSI), and Vaisala’s NLDN. Generally these networks detect a high percentage of CG lightning and a lower percentage of IC lightning, as they mostly operate in the low-frequency (LF) band. Specifically, the NLDN consists of a combination of time-of-arrival sensors and wideband magnetic direction finder (MDF) sensors, collectively referred to as the Improved Accuracy through Combined Technology (IMPACT) sensors. They detect the electromagnetic radiation with frequencies around 10 kHz emitted from long transient current events such as CG lightning return strokes. The NLDN also detects a lower percentage of IC lightning. The two-dimensional, horizontal lightning locations (x, y coordinates) are retrieved using a generalization of the least squares 
All of these networks measure parts of lightning flashes (such as return strokes of CG flashes, etc.), which we will call lightning events. Each of the networks classifies an event as a CG or IC part of a lightning flash, and possibly what events belong to the flash. CONUS lightning data used in this study include the date, time, polarity, signal strength, and type of a lightning event. Flash information was also available from the NLDN and ENTLN. We determined flashes from USPLN-measured lightning events by applying the criteria described in Cummins et al. (1998). Note that the detection efficiency, classification of IC and CG events, and flash determination can vary between the networks depending on factors such as network station density and the way detected lightning signals are processed.
3. The thundersnow events
a. Environment and overview of the cases
The synoptic conditions for each thundersnow case reveal similarities (Fig. 1). On 11 November 2012, 28 January 2013, and 9 April 2013, a large-scale, positively tilted 500-hPa trough was located to the west of Colorado, over the Rocky Mountains. On 25 October 2012, the trough axis had just passed through Colorado and was lifting out. A ridge was located over the eastern United States in all four cases. In each case, a surface low was located to the south or east of the COLMA domain, and a cold frontal passage had just occurred, providing surface temperatures very near 0°C. The surface features provided northerly or northeasterly upslope flow for the Colorado Front Range and Palmer Divide (e.g., Rasmussen et al. 1995; Kumjian et al. 2014; Schrom et al. 2015), overlaid by larger-scale southwesterly flow aloft. The proximity of the upper-level trough would support large-scale ascent across the region, whereas topographically forced ascent was likely for some events, as discussed below.
Synoptic overview of each thundersnow case. The 500-hPa heights (dkm; gray curves) are overlaid by selected surface features including the 0°C isotherm (dashed green curves), location of surface low pressure centers, and surface fronts. Analyses are adopted from the National Centers for Environmental Prediction/Hydrometeorological Prediction Center (NCEP/HPC). Upper-air (surface) analyses are valid at (a) 0000 (0300) UTC 25 Oct 2012, (b) 0000 (0000) UTC 11 Nov 2012, (c) 0000 UTC 29 Jan (2100 UTC 28 Jan) 2013, and (d) 1200 (0700) UTC 9 Apr 2013.
Citation: Weather and Forecasting 30, 6; 10.1175/WAF-D-15-0007.1
Figure 2 shows the nearest soundings from each of the four cases. The lowest levels in the first three cases (Figs. 2a–c) exhibited temperatures >0°C; however, strong cold advection associated with the frontal passage soon decreased temperatures throughout the atmosphere to below 0°C. For example, at 0000 UTC 25 October 2012, the surface temperature near Denver was about 10°C (Fig. 2a), whereas near Cheyenne it was already below −1°C (not shown). By 0200 UTC, Denver was reporting snow and a surface temperature of 0.5°C (not shown). The soundings are quite consistent with those associated with thundersnow cases presented in previous studies (e.g., Curran and Pearson 1971; Market et al. 2006, 2007), with a low-level inversion or stable layer overtopped with nearly pseudoadiabatic lapse rates and small dewpoint depressions. In each of the soundings, layers of potential instability (i.e., where the vertical gradient of equivalent potential temperature 
Radiosonde measurements of temperature (blue curves) and dewpoint temperature (dashed orange curves) observed at (a) 0000 UTC 25 Oct 2012, (b) 0000 UTC 11 Nov 2012, (c) 0000 UTC 29 Jan 2013, and (d) 0820 UTC 9 Apr 2013. The data in (a)–(c) are from the operational soundings taken from DNR, whereas (d) shows a sounding taken from the Marshall Field Site near Boulder during the Front Range Orographic Storms experiment (FROST; Kumjian et al. 2014).
Citation: Weather and Forecasting 30, 6; 10.1175/WAF-D-15-0007.1
The analysis domain is shown in Fig. 3. Locations of all LMA sources from each case are overlaid, as are the locations of the COLMA stations, the WSR-88Ds near KFTG and KCYS, and the National Weather Service Denver sounding site (DNR). One immediately sees a preferred location for lightning activity south of the LMA network, seemingly anchored to the terrain feature known as the Palmer Divide. The northerly or northeasterly upslope low-level flow in each case would favor enhanced uplift at the Palmer Divide. This topographically forced ascent could provide a focused lifting mechanism required to release the potential instability. In the absence of other lifting mechanisms, the Palmer Divide then could serve as a focus for convective initiation and subsequent lightning activity. The presence of broader large-scale ascent associated with upper-level troughs in each case could also have helped provide the lift necessary to release the potential instability.
Map of the terrain (m MSL; shaded according to scale) in the study domain. Locations of the operational sounding site (DNR; square green marker) and WSR-88Ds (KFTG and KCYS; red triangle markers) are shown. Colorado LMA stations are shown by circle black markers. Pink dots represent LMA sources for 25 Oct 2012, cyan dots represent source points for 11 Nov 2012, blue dots represent source points for 28 Jan 2013, and goldenrod dots represent source points for 9 Apr 2013.
Citation: Weather and Forecasting 30, 6; 10.1175/WAF-D-15-0007.1
Other features of note in Fig. 3 include linear tracks of static discharges seen by the LMA that are produced by aircraft departing from or arriving at Denver International Airport (i.e., LMA source locations increased or decreased in altitude at rates typical of commercial aircraft ascent and descent rates). The LMA observes VHF sources from sparks caused by planes that get charged (through collisional charging of the aircraft’s fuselage with ice particles) when flying through ice clouds (e.g., Thomas et al. 2004). For the subsequent analyses, these points were removed based on their linear spatial pattern and temporal continuity. Thus, in the event of a real flash in the spatial and temporal proximity of an aircraft track, a small number of real source points may be removed and/or a small number of aircraft sources may be retained, but this was not found to be an important issue for these cases.
b. Lightning analysis and charge structures
LMA sources and CONUS data were used to determine the number of IC and CG lightning events for each case. To filter out noise, LMA VHF sources were kept only if they were detected by a minimum of seven LMA stations, had altitudes <15 km, and had 
Time series of the flash counts determined by the LMA and CONUS networks for each case are shown in Fig. 4. In the first 80 min on 25 October, six different cells produced flashes (Fig. 4a), including one (cell 1) that produced 10 flashes over a period of about 30 min. In contrast, cells 2 and 4 only produced one flash, and cell 6 only produced two flashes. Six additional cells produced lightning flashes over the next 5 h (not shown in Fig. 4), with most of these not producing more than four flashes in their lifetime. One short-lived cell (cell 9) was more active, producing 15 flashes within 15 min. Another (cell 8) produced 19 flashes over about 90 min. The 11 November case (Fig. 4b) had the most electrically active storm of the dataset, with 28 flashes observed over about 1 h. Two cells produced flashes on 28 January (Fig. 4c), and one produced five flashes in about 10 min on 9 April 2013 (Fig. 4d).
Time series of flash counts for each case: (a) 25 Oct 2012, (b) 11 Nov 2012, (c) 28 Jan 2013, and (d) 9 Apr 2013. The same horizontal and vertical axis ranges are used to facilitate comparison. Multiple cells during one event are shown in different color lines.
Citation: Weather and Forecasting 30, 6; 10.1175/WAF-D-15-0007.1
Flash characteristics from each case are summarized in Table 1. The total number of flashes was determined by accumulating lightning data from all networks and matching them in space (within 1–2 km) and time (within 1 ms, the temporal resolution of CONUS network data). IC and CG events were determined based on the available data, including CONUS network classifications (e.g., Cummins et al. 1998) and interpretation of 3D LMA data (e.g., Thomas et al. 2003). The maximum flash rate (based on total flashes) in any 10-min period is also provided. Four cases produced more CG flashes than IC flashes, though one of these cases (cell 1, 25 October 2012) was at the edge of the LMA detection range,1 and thus not all flashes may have been detected. Another flash (cell 2, 25 October 2012) likely was tower initiated, as discussed further in section 4b. Otherwise, the majority of the cases produced more IC than CG lightning. Interestingly, for most cases the three CONUS networks exhibited variability in the number, type, and polarity of flashes detected, owing to differences in their design and performance, especially for cells with high IC-to-CG ratios. Note that there are ambiguities in deciding if flashes were IC or CG events based on the LMA data, particularly for flashes that initiate and progress very close to the ground and thus are limited by the line-of-sight restrictions of the LMA detections. Therefore, we caution that these flash-type classifications contain uncertainty.
Summary of thundersnow storms and their flash characteristics. The date of each event is in the left column, followed by the cell number for each date, the total number of flashes, the numbers of IC and CG flashes, the number of those flashes detected by the LMA, the range (from min to max) of the number of flashes detected by the any of the three CONUS networks, and the max flash rate (flashes per minute) within a 10-min window.
Flash rates were determined from a 10-min moving window throughout each cell’s lifetime. The maximum flash rate in any 10-min window throughout a cell’s lifetime is shown in Table 1. Most cells had maximum flash rates <1.0 min−1; the maximum flash rate for all events was from the 11 November 2012 cell, but was only 1.3 min−1. Compared to flash rates in warm-season thunderstorms (e.g., Williams et al. 2005; Deierling et al. 2008), flash rates in these thundersnow cases were much lower, exemplifying the more marginal nature of these storms. Fuchs et al. (2015) determined that 0-dBZ echo-top heights in Colorado summertime thunderstorms varied between 12 and 18 km, whereas the thundersnow storm echo-top heights varied between 8 and 12 km for most cases (and did not exceed 14 km). This suggests less vigorous upward motion than warm-season convection in this region. In addition, observed temperature profiles suggest no warm-cloud (>0°C) depth associated with these storms. These environmental factors are consistent with the observed lower flash rates (e.g., Williams et al. 2005; Fuchs et al. 2015).
The charge structure of these cells was also examined. As such, the vertical distribution of LMA sources is shown in Fig. 5. The heights of the charge regions varied by case, occurring anywhere between 2 and 10 km AGL. Manual flash-by-flash analysis allows us to infer charge regions based on the propagation of LMA-detected VHF sources (e.g., Wiens et al. 2005). These analyses suggested that, in many of the cases, negative charge was located at lower altitudes and positive charge at higher altitudes in these storms. This positive-over-negative charge structure could be achieved if the faster-falling riming particles (e.g., graupel or rimed aggregates) gained negative charge in these storms at temperatures <0°C and intermediate values of SLW (e.g., Saunders and Peck 1998; Takahashi and Miyawaki 2002). Similar processes contribute to cloud electrification in warm-season thunderstorms; however, warm-season storms often appear to be dynamically more vigorous than these thundersnow events (i.e., greater updraft velocities leading to higher echo-top heights). We caution that the flashes did not always reveal such a simple charge structure, and longer-lived events exhibited evolving structures throughout their lifetimes. Thus, it is difficult to make any broad generalizations about the charge structure of thundersnow storms without a much larger dataset.
LMA source frequency binned by altitude (m MSL) for each case: (a) 25 Oct 2012, (b) 11 Nov 2012, (c) 28 Jan 2013, and (d) 9 Apr 2013. LMA sources are thresholded by altitude and 
Citation: Weather and Forecasting 30, 6; 10.1175/WAF-D-15-0007.1
4. Radar analysis
a. Evolution of HCA fields
The upgraded WSR-88D network employs a hydrometeor classification algorithm (Park et al. 2009) that categorizes each radar pixel as 1 of 10 possible classes. In this section, we show illustrative examples of the HCA evolution for two cases: 11 November and 28 January (a similar evolution was observed for cells on 25 October and for 9 April; for brevity these are not shown). Figure 6 provides the evolution of the HCA output at 2.4° elevation using volume scans from 2350 UTC 10 November through 0007 UTC 11 November 2012, collected by the KFTG WSR-88D. Initially, mainly dry snow aggregates are classified in the echo to the southwest of the radar (Fig. 6a, centered at about x = −35 km, y = −25 km, where the origin of the coordinate system is the KFTG radar). By 2356 UTC, a small area of graupel is identified (Fig. 6b, marked by the arrow). This HCA-indicated graupel region expands in areal extent in the next two volume scans, more than doubling the number of graupel-identified pixels (Figs. 6c,d). By the end of the 0007 UTC volume scan, a flash is detected by the LMA (VHF sources indicated by the black markers in Fig. 6d). A strikingly similar evolution is observed for the 28 January case (Fig. 7). Again, the echo is dominated by dry snow aggregates initially (Fig. 7a). By the 2131:03 UTC scan (Fig. 7b), a few contiguous pixels of graupel are identified. This region expands considerably in the next two scans (Figs. 7c,d). The first LMA sources occurred shortly after 2142 UTC (Fig. 7d).
Output of the operational HCA at (a) 2350 UTC 10 Nov, (b) 2356 UTC 10 Nov, (c) 0002 UTC 11 Nov, and (d) 0007 UTC 11 Nov 2012. Data are collected at the 2.4°-elevation scan. LMA sources that occurred within the volume scan time are indicated as small black triangle markers. The arrow indicates the eruption of HCA-indicated graupel.
Citation: Weather and Forecasting 30, 6; 10.1175/WAF-D-15-0007.1
As in Fig. 6, but for HCA at (a) 2124, (b) 2131, (c) 2135, and (d) 2140 UTC 28 Jan 2013.
Citation: Weather and Forecasting 30, 6; 10.1175/WAF-D-15-0007.1
Such a rapid appearance and expansion of graupel identifications in both cases is suggestive of convective activity and considerable ongoing riming within a region of snow aggregates and ice crystals, which are precursors to electrification. In most of the cases, the operational HCA output classified regions of graupel prior to the first LMA-indicated flash. Most often this change in classification from snow aggregates to graupel is related to the increase in 
From an operational perspective, it appears that a sudden appearance or expansion in the areas classified as graupel should warrant more attention, as conditions are favorable for the development of electrification and possible lightning initiation, as well as potentially heavy snowfall. Crowe et al. (2006) explore correlations between heavy snowfall and thundersnow occurrences using METAR. They report that, if a storm is capable of producing thundersnow, then there is an enhanced likelihood of that system producing heavy snow accumulations. The areal expansion of graupel described above for two of the thundersnow events implies a larger region of enhanced 
However, the appearance of HCA-identified graupel itself is not a necessary or sufficient condition for lightning in the thundersnow cases presented herein. For example, in at least two of the cases (25 October and 11 November) there were other cells in which graupel was identified that did not produce lightning. The 11 November case is investigated in more detail in section 4c below. It is likely that ongoing riming and charging was occurring in these cells. However, the charging apparently was not sufficient to induce a lightning discharge.
In addition to such null cases (i.e., HCA-identified graupel but no lightning), there was one flash on 25 October 2012 that occurred in the absence of HCA-identified graupel. In fact, the flash occurred in a snowband that did not exhibit the typical convective appearance of the other lightning-producing cells. However, as discussed in detail in the next subsection, this case did exhibit a polarimetric signature that could have provided forecasters with information about electrification in the cloud.
b. 25 October 2012 depolarization streak
Here, we present data from the 25 October 2012 case, in which a snowband produced a flash (the northernmost flash in Fig. 3) in the absence of any obvious convective structure in the radar data. Despite not having a clear convective structure apparent in the radar data or large 
The 0.5° PPI scan from KCYS at 0039 UTC, showing (a) 
Citation: Weather and Forecasting 30, 6; 10.1175/WAF-D-15-0007.1
Consecutive fields of 
Citation: Weather and Forecasting 30, 6; 10.1175/WAF-D-15-0007.1
Figure 10 provides an average of radial traces of 
Traces of azimuthally averaged profiles of 
Citation: Weather and Forecasting 30, 6; 10.1175/WAF-D-15-0007.1
Notably, 
The vertical structure of the storms can also be informative. A vertical cross section along the azimuth nearest to the flash (Fig. 11) shows enhanced vertical structure in 
Reconstructed RHI scan of ZH (dBZ; shaded according to scale) from constant-elevation angle PPI scans in the volume beginning at 0035 UTC 25 Oct 2012 from KCYS. The azimuth is 108°, corresponding to the location CG flash detected at 0039 UTC. The flash was located at a range of ~89 km from the radar. Note the vertically extensive region of enhanced 
Citation: Weather and Forecasting 30, 6; 10.1175/WAF-D-15-0007.1
Figure 12 reveals the three-dimensional structure of the 0039 UTC CG flash and its evolution. The LMA sources indicate the flash initiated below 2 km MSL (within a few hundred meters above ground level) and progressed southward. The second part of the flash is accompanied by an upward-moving leader to the north again, which also initiates very close to the ground. The initial low-level sources (gray markers, near the ground) are within 1–2 km of several wind turbines and a communications tower as observed on Google satellite imagery. Given the uncertainty in VHF source positions by the LMA at this range, it is plausible that this flash was triggered by a man-made object. This would explain the isolated nature of this flash, as well as the seeming lack of a convective appearance in the radar data. Man-made objects such as towers can serve to locally enhance the electric field, possibly to a level sufficient to cause a discharge. In fact, Rakov and Uman (2003) suggest that the large horizontal extent of low-level charge regions common in winter storms favor lightning triggered by such towers.
Multipanel view of LMA sources associated with the 0039 UTC 25 Oct 2012 stratiform flash: (a) time (s; past 0000 UTC) vs height (m MSL), (b) lon (°) vs height (m MSL), (c) lon (°) vs lat (°) underlaid with a map, and (d) height (m MSL) vs lat (°). The color of the markers indicates the time of the source, binned into every 0.1 s starting at 2340:40 UTC, as indicated in the legend above (d).
Citation: Weather and Forecasting 30, 6; 10.1175/WAF-D-15-0007.1
Similar observations of flashes in the absence of notable convective structures were made by Bech et al. (2013) in a thundersnow case over Catalonia, Spain. Those authors hypothesized the importance of tall telecommunication towers in leading to CG lightning flashes. Tower-initiated flashes have also been observed in northern Alabama winter cases (C. Schultz 2014, personal communication) and over tall buildings in Chicago, Illinois (Warner et al. 2014). Thus, tall towers may be sufficient to produce a discharge in otherwise submarginal conditions for thundersnow.
c. 10–11 November 2012: A tale of two cells
During 10–11 November 2012, there were numerous isolated convective cells that produced 
Conical graupel particle observed near Boulder at 2201 UTC 10 Nov 2012. The largest graupel particles observed at this time approached 5 mm in diameter at their base.
Citation: Weather and Forecasting 30, 6; 10.1175/WAF-D-15-0007.1
To facilitate a comparison, data from sampling volumes subjectively identified as being associated with each cell are selected. The data from the PPI sweeps in which the cell is observable are used. At each volume scan time, the data are binned as follows: from 0 to 50 dBZ in 2-dB increments for 
The results are shown in Fig. 14. The active cell is shown in Fig. 14 (left), with asterisks above indicative of minutes in which lightning occurred. Note the jump in relative frequency of the larger 
Normalized frequency distributions of (a),(b) 
Citation: Weather and Forecasting 30, 6; 10.1175/WAF-D-15-0007.1
The distributions of 
The distributions of 
In summary, the active cell had a broader distribution of 
Because the mixture of graupel and pristine ice is a key ingredient for electrification, the combined use of 
In addition to the polarimetric radar variables, other investigators have explored the use of radar-derived products for the detection of storms capable of producing lightning. Figure 15 shows two of these products: volume averaged height-integrated radar reflectivity (VAHIRR; Krider et al. 2006) and a proxy for ice mass (e.g., Carey and Rutledge 2000; Mosier et al. 2011). These are products derived from gridded radar volume scans that have been used in previous lightning forecasting studies. One should not ascribe too much importance to individual values, but rather the relative differences in the two cases are what is important. VAHIRR values in the active cell are larger than those of the inactive cell, which do not exceed ~25 dBZ km. The most striking contrast is found in ice mass, however. Whereas the ice mass in the inactive cell does not exceed about 3.5 g m−3, values >4 g m−3 are consistently found in the active cell throughout the period in which it is producing lightning. In fact, values in excess of 8 g m−3 are found during a period of heightened lightning activity. The increased ice mass is consistent with the larger 
As in Fig. 14, but for derived quantities of (top) VAHIRR (dBZ km) and (bottom) ice mass (g m−3). The ice mass frequency distributions are shown in logarithmic scale.
Citation: Weather and Forecasting 30, 6; 10.1175/WAF-D-15-0007.1
We performed a similar comparison between the two cells using the level-III product called the enhanced echo-top height (not shown). There were no significant differences between the two cells, highlighting their similarity in storm depth. This also could mean that the vertical resolution of the WSR-88D volume scan is insufficient to observe such subtle differences.
These analyses offer promising clues of subtle differences between otherwise similar convective cells. However, a much larger dataset is required to determine the robustness of these radar-observed differences. Compositing or averaging over a large number of storms may help reveal reliable bulk microphysical differences between convective snowstorms that produce lightning and those that do not.
5. Discussion and conclusions
The four thundersnow events presented in this study display common features in their synoptic-scale setting, thermodynamic environment, radar presentation, and electrical properties. Each case featured a large-scale environment conducive for ascent, with low-level mesoscale features that provided upslope flow. In particular, the Palmer Divide was found to be a preferred location for electrically active snowstorms in Colorado. In each case, the surface temperature was just under 0°C.
With regard to the questions asked in the introduction, we come to the following conclusions:
- In general, lightning from the four thundersnow days was detected by both the CONUS and LMA networks. Of the six cells without CG flashes, the CONUS network detected IC flashes in four of them. The two cases that were not detected by the CONUS network were only weakly electrically active, producing only one in-cloud flash. The LMA detected flashes in all 16 cells, though individual flashes were missed when the cells were at the edge of the network’s detectability. The relatively prolific number of thundersnow storms in the 2012/13 snow season in northern Colorado strongly suggests that thundersnow is more common than previously reported in published climatologies that used different datasets (i.e., no combination of LMA and CONUS networks). LMA networks allow for the detection of marginal events with a single or a few IC flashes that generally do not get detected as efficiently by CONUS networks. Future research should investigate other LMA networks in different parts of the world to verify if, in fact, thundersnow events indeed are more common than previously thought.
- Many of the flash events were associated with localized high-
- Compared to typical warm-season convective storms that may have from several to hundreds of flashes per minute, the thundersnow events studied herein had much lower flash rates (often <1 min−1). This is not surprising given the lower-CAPE environments (and thus weaker maximum updraft velocities and cloud vertical extent). Flash-by-flash analysis based on LMA data suggests negative charge below an upper positive charge region in many of the cases. This is very similar to weakly electrified warm-season thunderstorm structures with intermediate amounts of supercooled liquid water. Thus, we suggest that thundersnow storms often are simply at the weaker end of the spectrum of ordinary thunderstorms, with an absence of any >0°C temperatures in the lower troposphere.
Aside from being simply a meteorological curiosity, thundersnow events are scientifically advantageous to study because of their marginal nature. Electrification and lightning in deep moist convection are virtually guaranteed in most midlatitude continental storms, whereas thundersnow events are much more marginal. If some electrification threshold exists that may distinguish lightning-producing storms from those that do not produce lightning, thundersnow storms are certainly closer to that threshold than most deep moist convective storms. Electric field measurements inside thundersnow clouds may help better define this threshold, if it exists.
The National Center for Atmospheric Research (NCAR) is sponsored by the National Science Foundation (NSF). Funding for the first author comes from NSF Grant AGS-1143948. Partial support for the first author came from the NCAR Advanced Study Program. We thank Jim Dye and Matthias Steiner (NCAR) for valuable discussions on this research, and Paul Krehbiel and Bill Rison (New Mexico Institute of Mining and Technology) for the Colorado LMA lightning data. We also thank Earth Networks, Vaisala, and WSI for the use of their respective lightning data in our research. The three anonymous reviewers are thanked for their detailed, constructive criticisms and suggestions that greatly improved the manuscript.
REFERENCES
Andrić, J., , Kumjian M. R. , , Zrnić D. S. , , Straka J. M. , , and Melnikov V. M. , 2013: Polarimetric signatures above the melting layer in winter storms: An observational and modeling study. J. Appl. Meteor. Climatol., 52, 682–700, doi:10.1175/JAMC-D-12-028.1.
Bech, J., , Pineda N. , , Rigo T. , , and Aran M. , 2013: Remote sensing analysis of a Mediterranean thundersnow and low-altitude heavy snowfall event. Atmos. Res., 123, 305–322, doi:10.1016/j.atmosres.2012.06.021.
Berger, A., 1998: VITAL SIGNS: Safety: When lightning strikes on the ski slopes. New York Times, 8 December 1998, p. 36. [Available online at: http://www.nytimes.com/1998/12/08/health/vital-signs-safety-when-lightning-strikes-on-the-ski-slopes.html.]
Boccippio, D. J., , Koshak W. , , Blakeslee R. , , Driscoll K. , , Mach D. , , and Buechler D. , 2000: The Optical Transient Detector (OTD): Instrument characteristics and cross-sensor validation. J. Atmos. Oceanic Technol., 17, 441–458, doi:10.1175/1520-0426(2000)017<0441:TOTDOI>2.0.CO;2.
Bringi, V. N., , and Chandrasekar V. , 2001: Polarimetric Doppler Weather Radar. 1st ed. Cambridge University Press, 636 pp.
Carey, L. D., , and Rutledge S. A. , 2000: The relationship between precipitation and lightning in tropical island convection: A C-band polarimetric radar study. Mon. Wea. Rev., 128, 2687–2710, doi:10.1175/1520-0493(2000)128<2687:TRBPAL>2.0.CO;2.
Caylor, I. J., , and Chandrasekar V. , 1996: Time-varying crystal orientation in thunderstorms observed with multiparameter radar. IEEE Trans. Geosci. Remote Sens., 34, 847–858, doi:10.1109/36.508402.
Crowe, C. C., , Market P. , , Pettigrew B. , , Melick C. , , and Podzimek J. , 2006: An investigation of thundersnow and deep snow accumulations. Geophys. Res. Let., 33, L24812, doi:10.1029/2006GL028214.
Cummins, K. L., , and Murphy M. J. , 2009: An overview of lightning locating systems: History, techniques and data uses, with an in-depth look at the U.S. NLDN. IEEE Trans. Electromagn. Compat., 51, 499–518, doi:10.1109/TEMC.2009.2023450.
Cummins, K. L., , Murphy M. J. , , Bardo E. A. , , Hiscox W. L. , , Pyle R. B. , , and Pifer A. E. , 1998: A combined TOA/MDF technology upgrade of the U.S. National Lightning Detection Network. J. Geophys. Res., 103, 9035–9044, doi:10.1029/98JD00153.
Curran, J. T., , and Pearson A. D. , 1971: Proximity soundings for thunderstorms with snow. Preprints, Seventh Conf. on Severe Local Storms, Kansas City, MO, Amer. Meteor. Soc., 118–119.
Deierling, W., , Petersen W. A. , , Latham J. , , Ellis S. , , and Christian H. J. , 2008: The relationship between lightning activity and ice fluxes in thunderstorms. J. Geophys. Res., 113, D15210, doi:10.1029/2007JD009700.
Doviak, R. J., , and Zrnić D. S. , 1993: Doppler Radar and Weather Observations. 2d ed. Academic Press, 562 pp.
Dye, J. E., , and Willett J. C. , 2007: Observed enhancement of reflectivity and the electric field in long-lived Florida anvils. Mon. Wea. Rev., 135, 3362–3380, doi:10.1175/MWR3484.1.
Dye, J. E., , Jones J. , , Weisenheimer A. J. , , and Winn W. P. , 1988: Observations within two regions of charge during initial thunderstorm electrification. Quart. J. Roy. Meteor. Soc., 114, 1271–1290, doi:10.1002/qj.49711448306.
Dye, J. E., and Coauthors, 2007: Electric fields, cloud microphysics, and reflectivity in anvils of Florida thunderstorms. J. Geophys. Res., 112, D11215, doi:10.1029/2006JD007550.
Fuchs, B. R., and Coauthors, 2015: Environmental controls on storm intensity and charge structure in multiple regions of the continental United States. J. Geophys. Res. Atmos., 120, 6575–6596, doi:10.1002/2015JD023271.
Fukao, S., , Maekawa Y. , , Sonoi Y. , , and Yoshino F. , 1991: Dual polarization radar observation of thunderclouds on the last of the Sea of Japan in the winter season. Geophys. Res. Lett., 18, 179–182, doi:10.1029/90GL02669.
Hiscox, W. L., , Krider E. P. , , Pifer A. E. , , and Uman M. A. , 1984: A systematic method for identifying site errors in a network of magnetic direction finders. Proc. Int. Aerospace and Ground Conf. on Lightning and Static Electricity, Orlando, FL, National Interagency Coordination Group, 7-1–7-5.
Holle, R. L., 2014: Diurnal variations of NLDN-reported cloud-to-ground lightning in the United States. Mon. Wea. Rev., 142, 1037–1052, doi:10.1175/MWR-D-13-00121.1.
Hubbert, J. C., , Ellis S. M. , , Dixon M. , , and Meymaris G. , 2010: Modeling, error, analysis, and evaluation of dual-polarization variables obtained from simultaneous horizontal and vertical polarization transmit radar. Part II: Experimental data. J. Atmos. Oceanic Technol., 27, 1599–1607.
Hunter, S. M., , Underwood S. J. , , Holle R. L. , , and Mote T. L. , 2001: Winter lightning and heavy frozen precipitation in the southeast United States. Wea. Forecasting, 16, 478–490, doi:10.1175/1520-0434(2001)016<0478:WLAHFP>2.0.CO;2.
Kennedy, P. C., , and Rutledge S. A. , 2011: S-band dual-polarization radar observations of winter storms. J. Appl. Meteor. Climatol., 50, 844–858, doi:10.1175/2010JAMC2558.1.
Krehbiel, P. R., , Thomas R. J. , , Rison W. , , and Hamlin T. , 2000: Lightning mapping observations in central Oklahoma. Eos, Trans. Amer. Geophys. Union, 81 (3), 21–25, doi:10.1029/00EO00014.
Krider, E. P., and Coauthors, 2006: Natural and triggered lightning launch commit criteria. Preprints, 12th Conf. on Aviation, Range, and Aerospace Meteorology, Atlanta, GA, Amer. Meteor. Soc., 8.3. [Available online at http://ams.confex.com/ams/pdfpapers/104060.pdf.]
Kuhlman, K. M., , MacGorman D. R. , , Biggerstaff M. I. , , and Krehbiel P. R. , 2009: Lightning initiation in the anvils of two supercell storms. Geophys. Res. Lett., 36, L07802, doi:10.1029/2008GL036650.
Kumjian, M. R., 2013a: Principles and applications of dual-polarization weather radar. Part I: Description of the polarimetric radar variables. J. Oper. Meteor., 1, 226–242, doi:10.15191/nwajom.2013.0119.
Kumjian, M. R., 2013b: Principles and applications of dual-polarization weather radar. Part II: Warm and cold season applications. J. Oper. Meteor., 1, 243–264, doi:10.15191/nwajom.2013.0120.
Kumjian, M. R., 2013c: Principles and applications of dual-polarization weather radar. Part III: Artifacts. J. Oper. Meteor., 1, 265–274, doi:10.15191/nwajom.2013.0121.
Kumjian, M. R., , Rutledge S. A. , , Rasmussen R. M. , , Kennedy P. C. , , and Dixon M. , 2014: High-resolution polarimetric radar observations of snow-generating cells. J. Appl. Meteor. Climatol., 53, 1636–1658, doi:10.1175/JAMC-D-13-0312.1.
Lang, T. J., and Coauthors, 2004: The Severe Thunderstorm Electrification and Precipitation Study. Bull. Amer. Meteor. Soc., 85, 1107–1125, doi:10.1175/BAMS-85-8-1107.
Liu, C., , and Heckman S. , 2011: Using total lightning data in severe storm prediction: Global case study analysis from North America, Brazil, and Australia. Proc. 2011 European Geosciences Union General Assembly, Vienna, Austria, EGU, 20–24, doi:10.1109/SIPDA.2011.6088433.
MacGorman, D. R., , and Rust W. D. , 1998: The Electrical Nature of Storms. 1st ed. Oxford University Press, 422 pp.
Maekawa, Y., , Fukao S. , , and Yoshino F. , 1992: Dual polarization radar observations of anomalous wintertime thunderclouds in Japan. Geosci. Remote Sens., 30, 838–844, doi:10.1109/36.158881.
Maekawa, Y., , Fukao S. , , Sonoi Y. , , and Yoshino F. , 1993: Distribution of ice particles in wintertime thunderclouds detected by a C band dual polarization radar. J. Geophys. Res., 98, 16 613–16 622, doi:10.1029/93JD01357.
Mäkelä, A., , Saltikoff E. , , Julkunen J. , , Gregow E. , , and Niemelä S. , 2013: Cold-season thunderstorms in Finland and their effect on aviation safety. Bull. Amer. Meteor. Soc., 94, 847–858, doi:10.1175/BAMS-D-12-00039.1.
Market, P. S., , and Becker A. E. , 2009: A study for lightning flashes attending periods of banded snowfall. Geophys. Res. Let., 36, L01809, doi:10.1029/2008GL036317.
Market, P. S., , Halcomb C. E. , , and Ebert R. , 2002: A climatology of thundersnow events over the contiguous United States. Wea. Forecasting, 17, 1290–1295, doi:10.1175/1520-0434(2002)017<1290:ACOTEO>2.0.CO;2.
Market, P. S., and Coauthors, 2006: Proximity soundings of thundersnow in the central United States. J. Geophys. Res., 111, D19208, doi:10.1029/2006JD007061.
Market, P. S., , Ebert-Cripe L. , , and Bodner M. , 2007: Case study of a long-lived thundersnow event. Natl. Wea. Dig., 31 (2), 103–119.
Mosier, R. M., , Schumacher C. , , Orville R. E. , , and Carey L. D. , 2011: Radar nowcasting of cloud-to-ground lightning over Houston, Texas. Wea. Forecasting, 26, 199–212, doi:10.1175/2010WAF2222431.1.
Orville, R. E., 1991: Lightning ground flash density in the contiguous United States—1989. Mon. Wea. Rev., 119, 573–577, doi:10.1175/1520-0493(1991)119<0573:LGFDIT>2.0.CO;2.
Orville, R. E., , and Huffines G. R. , 2001: Cloud-to-ground lightning in the United States: NLDN results in the first decade, 1989–98. Mon. Wea. Rev., 129, 1179–1193, doi:10.1175/1520-0493(2001)129<1179:CTGLIT>2.0.CO;2.
Park, H. S., , Ryzhkov A. V. , , Zrnić D. S. , , and Kim K.-Y. , 2009: The hydrometeor classification algorithm for the polarimetric WSR-88D: Description and application to an MCS. Wea. Forecasting, 24, 730–748, doi:10.1175/2008WAF2222205.1.
Pettegrew, B. P., 2008: Analysis of cloud and cloud-to-ground lightning in winter convection. University of Missouri, Ph.D. dissertation, 104 pp.
Pettegrew, B. P., , Market P. S. , , Wold R. A. , , Holle R. L. , , and Demetriades N. W. S. , 2009: A case study of severe winter convection in the Midwest. Wea. Forecasting, 24, 121–139, doi:10.1175/2008WAF2007103.1.
Pruppacher, H. R., , and Klett J. D. , 1997: Microphysics of Clouds and Precipitation. 2d ed. Kluwer, 954 pp.
Rakov, V. A., , and Uman M. A. , 2003: Lightning: Physics and Effects. Cambridge University Press, 687 pp.
Rasmussen, R. M., , Bernstein B. C. , , Murakami M. , , Stossmeister G. , , and Reisner J. , 1995: The 1990 Valentine’s Day Arctic outbreak. Part I: Mesoscale and microscale structure and evolution of a Colorado Front Range shallow upslope cloud. J. Appl. Meteor., 34, 1481–1511, doi:10.1175/1520-0450-34.7.1481.
Rauber, R. M., and Coauthors, 2014: Stability and charging characteristics of the comma head region of continental winter cyclones. J. Atmos. Sci., 71, 1559–1582, doi:10.1175/JAS-D-13-0253.1.
Rison, W., , Thomas R. J. , , Krehbiel P. R. , , Hamlin T. , , and Harlin J. , 1999: A GPS-based three-dimensional lightning mapping system: Initial observations in central New Mexico. Geophys. Res. Lett., 26, 3573–3576, doi:10.1029/1999GL010856.
Rudlosky, S. D., , and Fuelberg H. E. , 2010: Pre- and post-upgrade distributions of NLDN reported cloud-to-ground lightning characteristics in the contiguous United States. Mon. Wea. Rev., 138, 3623–3633, doi:10.1175/2010MWR3283.1.
Ryzhkov, A. V., , and Zrnić D. S. , 2007: Depolarization in ice crystals and its effect on radar polarimetric measurements. J. Atmos. Oceanic Technol., 24, 1256–1267, doi:10.1175/JTECH2034.1.
Ryzhkov, A. V., , Zrnić D. S. , , and Gordon B. A. , 1998: Polarimetric method for ice water content determination. J. Appl. Meteor., 37, 125–134, doi:10.1175/1520-0450(1998)037<0125:PMFIWC>2.0.CO;2.
Ryzhkov, A. V., , Schuur T. J. , , Burgess D. W. , , Heinselman P. L. , , Giangrande S. E. , , and Zrnić D. S. , 2005: The Joint Polarization Experiment: Polarimetric rainfall measurements and hydrometeor classification. Bull. Amer. Meteor. Soc., 86, 809–824, doi:10.1175/BAMS-86-6-809.
Saunders, C. P. R., 1993: A review of thunderstorm electrification processes. J. Appl. Meteor., 32, 642–655, doi:10.1175/1520-0450(1993)032<0642:AROTEP>2.0.CO;2.
Saunders, C. P. R., , and Peck S. L. , 1998: Laboratory studies of the influence of the rime accretion rate on charge transfer during crystal/graupel collisions. J. Geophys. Res., 103, 13 949–13 956, doi:10.1029/97JD02644.
Schrom, R. S., , Kumjian M. R. , , and Lu Y. , 2015: Polarimetric radar signatures of dendritic growth zones within Colorado winter storms. J. Appl. Meteor. Climatol., doi:10.1175/JAMC-D-15-0004.1, in press.
Schultz, D. M., 1999: Lake-effect snowstorms in northern Utah and western New York with and without lightning. Wea. Forecasting, 14, 1023–1031, doi:10.1175/1520-0434(1999)014<1023:LESINU>2.0.CO;2.
Schultz, D. M., , and Vavrek J. , 2009: An overview of thundersnow. Weather, 64, 274–277, doi:10.1002/wea.376.
Scott, R. D., , Krehbiel P. R. , , and Rison W. , 2001: The use of simultaneous horizontal and vertical transmissions for dual-polarization radar meteorological observations. J. Atmos. Oceanic Technol., 18, 629–648, doi:10.1175/1520-0426(2001)018<0629:TUOSHA>2.0.CO;2.
Straka, J. M., , Zrnić D. S. , , and Ryzhkov A. V. , 2000: Bulk hydrometeor classification and quantification using polarimetric radar data: Synthesis of relations. J. Appl. Meteor., 39, 1341–1372, doi:10.1175/1520-0450(2000)039<1341:BHCAQU>2.0.CO;2.
Stuart, N., 2001: Multi-dimensional analysis of an extreme thundersnow event during the 30 December 2000 snowstorm. Preprints, 18th Conf. on Weather Analysis and Forecasting, Fort Lauderdale, FL, Amer. Meteor. Soc., P3.10. [Available online at https://ams.confex.com/ams/pdfpapers/22524.pdf.]
Takahashi, T., 1978: Riming electrification as a charge generation mechanism in thunderstorms. J. Atmos. Sci., 35, 1536–1548, doi:10.1175/1520-0469(1978)035<1536:REAACG>2.0.CO;2.
Takahashi, T., , and Miyawaki K. , 2002: Reexamination of riming electrification in a wind tunnel. J. Atmos. Sci., 59, 1018–1025, doi:10.1175/1520-0469(2002)059<1018:ROREIA>2.0.CO;2.
Thomas, R., , Krehbiel P. R. , , Rison W. , , Harlin J. , , Hamlin T. , , and Campbell N. , 2003: The LMA flash algorithm. Proc. 12th Int. Conf. on Atmospheric Electricity, Versailles, France, International Commission on Atmospheric Electricity, 655–656.
Thomas, R., , Krehbiel P. R. , , Rison W. , , Hunyady J. , , Winn W. P. , , Hamlin T. , , and Harlin J. , 2004: Accuracy of the lightning mapping array. J. Geophys. Res., 109, D14207, doi:10.1029/2004JD004549.
Tsenova, B., , Mitzeva R. , , and Saunders C. , 2014: The effect of charge separation in upper cloud layers on thunderstorm electrification—Numerical simulations. Extended Abstracts, XV Int. Conf. on Atmospheric Electricity, Norman, OK, International Commission on Atmospheric Electricity, P0812. [Available online at http://www.nssl.noaa.gov/users/mansell/icae2014/preprints/Tsenova_284.pdf.]
Warner, T. A., , Lang T. J. , , and Lyons W. A. , 2014: Synoptic scale outbreak of self-initiated upward lightning (SIUL) from tall structures during the central U.S. blizzard of 1–2 February 2011. J. Geophys. Res. Atmos., 119, 9530–9548, doi:10.1002/2014JD021691.
Wiens, K. C., , Rutledge S. A. , , and Tessendorf S. A. , 2005: The 29 June 2000 supercell observed during STEPS. Part II: Lightning and charge structure. J. Atmos. Sci., 62, 4151–4177, doi:10.1175/JAS3615.1.
Wilkinson, J. M., , Wells H. , , Field P. R. , , and Agnew P. , 2013: Investigation and prediction of helicopter-triggered lightning over the North Sea. Meteor. Appl., 20, 94–106, doi:10.1002/met.1314.
Williams, E. R., , and Lhermitte R. M. , 1983: Radar tests of the precipitation hypothesis for thunderstorm electrification. J. Geophys. Res., 88, 10 984–10 992, doi:10.1029/JC088iC15p10984.
Williams, E. R., , Mushtak V. , , Rosenfeld D. , , Goodman S. , , and Boccippio D. , 2005: Thermodynamic conditions favorable to superlative thunderstorm updraft, mixed phase microphysics, and lightning flash rate. Atmos. Res., 76, 288–306, doi:10.1016/j.atmosres.2004.11.009.
Zipser, E. J., , and Lutz K. R. , 1994: The vertical profile of radar reflectivity of convective cells: A strong indicator of storm intensity and lightning probability? Mon. Wea. Rev., 122, 1751–1759, doi:10.1175/1520-0493(1994)122<1751:TVPORR>2.0.CO;2.
Zook, D., 2014: When lightning strikes. Ski Area Manage., 53 (4), 36. [Available online at http://www.saminfo.com/article/when-lightning-strikes.]
Zrnić, D. S., , and Ryzhkov A. V. , 1999: Polarimetry for weather surveillance radars. Bull. Amer. Meteor. Soc., 80, 389–406, doi:10.1175/1520-0477(1999)080<0389:PFWSR>2.0.CO;2.
The farther distance from the LMA network also explains why cells 1 and 8 on 25 Oct 2012 had strokes detected by the CONUS networks but not the LMA.
