PICTURE of the MONTH a High-Resolution Lightning Map Of

VOLUME 142 MONTHLY WEATHER REVIEW JULY 2014

PICTURE OF THE MONTH

A High-Resolution Lightning Map of the State of Colorado

BRANDON J. VOGT University of Colorado Colorado Springs, Colorado Springs, Colorado

STEPHEN J. HODANISH NOAA/National Weather Service, Pueblo, Colorado

(Manuscript received 18 October 2013, in ﬁnal form 4 February 2014)

ABSTRACT

For the state of Colorado, 10 years (2003–12) of 1 April–31 October cloud-to-ground (CG) lightning stroke data are mapped at 500-m spatial resolution over a 10-m spatial resolution U.S. Geological Survey (USGS) digital elevation model (DEM). Spatially, the 12.5 million strokes that are analyzed represent ground contacts, but translate to density values that are about twice the number of ground contacts. Visual interpretation of the mapped data reveals the general lightning climatology of the state, while geospatial analyses that quantify lightning activity by elevation identify certain topographic influences of Colorado’s physical landscape. Elevations lower than 1829 m (6000 ft) and above 3200 m (10 500 ft) show a positive relationship between lightning activity and elevation, while the variegated topography that lies between these two elevations is characterized by a fluc- tuating relationship. Though many topographic controls are elucidated through the mappings and analyses, the major finding of this paper is the sharp increase in stroke density observed above 3200 m (10 500 ft). Topography’s role in this rapid surge in stroke density, which peaks in the highest mountain summits, is not well known, and until now, was not well documented in the refereed literature at such high resolution from a long-duration dataset.

1. Introduction 2011; Barker Schaaf et al. 1988) and can support and re- fine empirically derived conclusions that result from In map form, lightning climatologies reveal variations shorter-duration datasets and/or from a more coarse in lightning activity across an area. This cartographic spatial analysis. information benefits a wide range of entities including A primary motivating factor for this work is that there those that maintain networks and infrastructures vul- is no comprehensive cloud-to-ground (CG) lightning nerable to lightning and those with interests in hydrology climatology for the state of Colorado in the formal lit- and wildland fire management. In Colorado and other erature. The most formal comprehensive work that an- mountainous regions with abundant summertime out- alyzes CG lightning in Colorado was published by Lopez door recreation opportunities, a high-resolution lightning and Holle (1986). The authors examine a single year climatology map is particularly relevant from a lightning (1983) of CG lightning data collected over the state’s hazards perspective. In meteorology and regional cli- northeast region where they found the lightning activity matological studies, the ability to document lighting to be mainly a warm season event, with a large majority attachment patterns at fine scales across a variegated of the flashes occurring June through September. Other landscape contributes a deeper understanding of thun- results of the work by Lopez and Holle (1986) show that derstorm behavior (Cummins 2012; Hodanish and Wolyn lightning activity ranges from extremely intense with some days experiencing 5-min flash rates in excess of 100, to thunderstorms producing less than 10 flashes in Corresponding author address: Brandon J. Vogt, Geography and Environmental Studies, University of Colorado Colorado Springs, a 5-min period. Geographically, Lopez and Holle (1986) 1420 Austin Bluffs Parkway, Colorado Springs, CO 80918. found that maximum flash concentrations were aligned E-mail: [email protected] north–south along the eastern slopes of the Colorado

DOI: 10.1175/MWR-D-13-00334.1

Ó 2014 American Meteorological Society 2353 Unauthenticated | Downloaded 09/29/21 08:27 AM UTC 2354 MONTHLY WEATHER REVIEW VOLUME 142

Front Range, and then southeast to east along the north (DEM) terrain surface roughness in a lightning study, facing slopes of the Palmer Divide. our approach is made possible through the use of a suite In another investigation, Reap (1986) analyzed two of geospatial tools that are managed in a geographic in- years (1983 and 1984) of June through September formation system (GIS). A GIS is a mapping technology lightning data and other remotely sensed meteorological that enables users to interact with and analyze maps and variables over the western United States, including other sources of data. In this study, we used GIS and its Colorado, using a grid size of 47 3 47 km2.Ahigh sophisticated analytical methods to visualize the relation- correlation was found between terrain elevation and ship between lightning activity and elevation and to the hour of maximum frequency of lightning: the higher quantify this relationship in 152 m (500 ft) elevation the elevation, the earlier the onset of lightning activity. classes. The paper first discusses the data and the data The author concluded that a fairly stable seasonal dis- processing. This is followed by a brief discussion of the tribution and a generally homogeneous geographical lightning climatology of Colorado. Next, the focus shifts distribution of lightning activity exists over the western to a demonstration of the utility of GIS in quantitative United States, generally west of 1038W longitude. The lightning climatology studies through the characteriza- consistent spatial and temporal distribution observed tion of lightning activity by elevation. was likely the result of the strong control exerted by the underlying topographical features characteristic to that 2. Data part of the United States. Building on the successes of the works of Lopez and Lightning detection in the state of Colorado has been Holle (1986), Reap (1986), and others who have exam- ongoing since the 1970s. The Bureau of Land Manage- ined components of lightning activity in some or all of ment (BLM) was first to use such a system in order to the Colorado landscape (e.g., Orville and Huffines 2001; provide early warning of fire starts and to direct re- Zajac and Rutledge 2001; Boccippio et al. 2001; Barker sources to potential fire initiation locations (Krider et al. Schaaf et al. 1988), a lightning climatology paper by 1980). Throughout the 1980s, the BLM system was up- Hodanish and Wolyn (2011) comprehensively analyzed graded (Reap 1986), and in 1989 it was combined with lightning patterns over Colorado. This informal paper other regional lightning detection systems across the (e.g., not peer reviewed) analyzed 17 yr of CG flash data United States (Orville et al.1983; Mach et al. 1986)to on a 1-km spatial resolution map. The authors found form the U.S. National Lightning Detection Network that mountainous regions that had access to a continued (NLDN; Orville 2008, 1991). The NLDN itself has gone source of low-level moisture had the largest concentra- through numerous upgrades and expansions through the tions of CG lightning. These areas included the mountain– 1990s and 2000s (Cummins et al. 1998, 2006), and is now plains interface and the southern exposure of the known as the North American Lightning Detection San Juan Mountains. Mountainous terrain that did not Network, operated by Vaisala (Orville et al. 2002). The have access to a continued source of available moisture NLDN detects CG lightning flashes and strokes over experienced significantly less CG activity. It was also North America with a stroke detection efficiency of hypothesized that a certain meteorological regime, 60%–80% (during 2003–12), a flash detection efficiency known as the Denver Cyclone vorticity zone (DCVZ; of 90%–95%, and a spatial accuracy of 250 m (Cummins Szoke et al. 1984), plays a key role in both enhancing and and Murphy 2009). limiting the lightning activity across the greater north- Most negative CG flashes consist of multiple strokes, and east Colorado region. The authors also noted that large about half of these flashes have multiple ground contact valley regions surrounded by mountains show notice- points (GCP; Stall et al. 2009; Cummins and Murphy able minimums in lightning activity. 2009). On average, a negative CG flash will have from 1.45 The purpose of the current paper is to demonstrate to 1.7 GCP per flash (Valine and Krider 2002). These in- that a set of geospatial analytical tools can be used to dividual GCPs can be separated over a range of tens to build upon the Hodanish and Wolyn (2011) findings hundreds of meters. In this paper, stroke data are ana- and ultimately create a high-resolution statewide light- lyzed. Spatially, strokes characterize the GCP distribution, ning climatology. Advantages of our approach lie in the but represent about twice the number of actual GCPs. opportunity to visually explore vast datasets in map Recognizing the significant 2002–03 NLDN improve- form, a research format that facilitates visual thinking ments in stroke detection efficiency (Cummins and (MacEachren et al. 2004), and in the ability to generate Murphy 2009), the dataset examined in our study ex- quantitative results over discrete geographical regions. cludes the years prior to 2003. The dataset examined in Similar in methodology to the work of Cummins (2012) this paper covers the 10-yr period 2003–12. The 2002–03 who incorporated measures of digital elevation model NLDN upgrades are evident when comparing stroke

Unauthenticated | Downloaded 09/29/21 08:27 AM UTC JULY 2014 P I C T U R E O F T H E M O N T H 2355

FIG. 1. Stroke density by elevation histograms for two sets of years: (top) 1996–2002 and (bottom) 2003–12. By comparing the two histograms, the 2002–03 upgrades to the NLDN are clearly discernible. Most notably, the later years (2003–12) show a marked increase in stroke density in the higher elevations, above about 3200 m (10 500 ft). density by elevation class between two periods (1996– 2011). The resulting number of strokes examined over the 2002 and 2003–12) over the state of Colorado (Fig. 1). 10-yr period is 12.5 million. The later years (2003–12) show a marked increase in stroke density in the higher elevations above about 3. Data processing 3200 m (10 500 ft). The twenty-three 152-m (500 ft) elevation classes in Fig. 1 serve to outline the relationship A statewide 10-m2 spatial resolution U.S. Geological between stroke density and elevation, and are examined Survey (USGS) DEM serves as a base map for our ex- in greater detail in section 5. plorations. A DEM is a gridded array of elevation values November through March are not included in the that represent a sampled number of ground positions. dataset because these months capture a negligible portion The statewide DEM is utilized to visualize the topo- (less than one-half of 1%) of Colorado’s annual lightning graphic features of Colorado, such as mountains, valleys, activity (Hodanish and Wolyn 2011). Last, as with other plateaus, and plains, and serves as a platform from which CG lightning research, positive polarity strokes less than to quantify lightning activity by elevation in a GIS. 15 kA were excluded from the dataset (Cummins and The 2003–12 NLDN dataset and underlying DEM are Murphy 2009; Rudlosky and Fuelberg 2010; Orville et al. mapped in ESRI’s ArcMap 10.2 GIS software environment

Unauthenticated | Downloaded 09/29/21 08:27 AM UTC 2356 MONTHLY WEATHER REVIEW VOLUME 142

2 FIG. 2. Average warm season (1 Apr–31 Oct) strokes for the state of Colorado, 2003–12. Map resolution is 500 m , though each cell value represents stroke density at 1-km2 resolution. NLDN data are provided by Vaisala. The thick black lines represent major interstates.

(ESRI 2013). A 500-m2 vector fishnet consisting of Mount Elbert (U.S. Geological Survey 2013). Stroke 2 approximately 1.1 million cells was created within the densities in the state range from 30 km2 yr 1 to less than 2 political boundaries that define the state of Colorado. 1km2 yr 1. The greatest stroke densities occur in the A point-in-polygon count generates a 500-m2 raster higher elevations of the mountains, though these maxima image with each cell’s value representing the average vary from mountain range to mountain range. Areas of annual strokes received in each cell multiplied by 4 to high stroke densities occur along the gradual slope of the represent stroke density at 1 km2 across the state of mountain–plains interface, focusing over east–central Colorado (Fig. 2). The 152-m (500 ft) elevation ranges Colorado (Palmer Divide/Pikes Peak region) and along serve as templates through which terrain elevation is the Colorado–New Mexico border (Raton Mesa region) visualized and from which stroke count statistics are (Fig. 2). Minimum stroke densities occur over the San calculated (Fig. 3). Though represented here as two Luis Valley, along the major river valleys of the Colorado different maps (Figs. 2 and 3), in a GIS environment, River and Gunnison River, and throughout the state’s the information on both maps can be viewed as one steeply incised valleys. A well-defined minimum is noted scene and the transparency of both maps can be ad- on the plains north of Denver and immediately east of the justed to explore spatial relationships. Front Range. Similarly, northwest Colorado is characterized by a deficiency in stroke activity. 4. Summary of Colorado’s lightning climatology 5. The relationship between stroke density At 2070 m (6800 ft), Colorado has the highest mean and elevation, with an emphasis on elevations elevation of any state in the United States. Elevations above 3200 m (10 500 ft) range from 1010 m (3315 ft) in the northeast section of the state where the Arikaree River flows into Kansas to To identify interactions between the full range of 4399 m (14 433 ft) at the summit of centrally located terrain elevations and stroke density, the state’s 3389-m

Unauthenticated | Downloaded 09/29/21 08:27 AM UTC JULY 2014 P I C T U R E O F T H E M O N T H 2357

FIG. 3. Elevation at 152-m (500 ft) intervals for the state of Colorado. Several mountain ranges, cities, and physiographic regions are notated.

(11 119 ft) elevation range is divided into 152-m (500 ft) just west of Pueblo, the northern section of the Denver sections creating 23 elevation classes from which stroke metropolitan area, a strip of land west of Interstate 25 densities are calculated (Fig. 4). The elevation range from just east of Boulder to the Fort Collins area, and classes in map form presented in Fig. 3 correspond to the vast relatively featureless landscape that slopes into those presented in histogram form in Fig. 4. Because Kansas along the eastern edge of Colorado. The other 2% some elevation classes cross state boundaries and most of the state’s elevations that lie below 1524 m (5000 ft) in cross a series of topographically and latitudinally con- the state’s western portion include the lowlands adjacent trolled climate regimes, it is not possible to make explicit to Colorado River near the Utah border, and the ex- statements about relationship between stroke density treme tip of the state’s southwestern corner. The major and elevation for many of the elevation classes across controls of lighting activity in the eastern lower eleva- the state of Colorado. However, some of the stronger tions are associated with storms moving off the Front signals in Fig. 4, such as the positive relationships found Range mountains and moving east across the plains. in segment B–D and the anomalous dips at label E and Lightning activity gradually increases from east to west in label F, are traceable to specific landscape controls. The this region as low-level southeasterly flow interacts with remainder of this section elaborates on the topographic topographic forcing. The small fraction (one-tenth of controls that create the shape of the histogram in Fig. 4. 1%) of the state’s elevation class below 1067 m (3500 ft) A total of 98% of Colorado’s elevation below 1524 m experiences slightly higher stroke densities that the ad- (5000 ft) lies in the eastern portion of the state, generally jacent elevation class because the majority of this eleva- covering the area east of the higher elevations that form tion class lies in the far southeast section of the state, east the Cheyenne Ridge, Palmer Divide, and Raton Mesa of the Raton Mesa region (label A in Fig. 4). (segment A–C in Fig. 4). Specifically, these lower ele- Moving up in elevation, 78% of the state’s area be- vations capture the Arkansas River valley as far west as tween 1524 m (5000 ft) and 1829 m (6000 ft) is found on

Unauthenticated | Downloaded 09/29/21 08:27 AM UTC 2358 MONTHLY WEATHER REVIEW VOLUME 142

FIG. 4. Average annual strokes by elevation range for the state of Colorado. Labels A–G correspond to various land surfaces that are depicted in Fig. 3.

2 the east side of the state, east of the Rocky Mountains. average annual stroke density of 1.8 km2 yr 1, the San This elevation range includes the gently sloping plains Luis Valley occupies 35% of the state’s elevation range that outline the two notable lightning maxima in the between 2286 m (7500 ft) to 2438 m (8000 ft). state: the Palmer Divide region and the Raton Mesa The relative minimum at label F in Fig. 4 [3048 m region. This landscape of strong topographic forcing is (10 000 ft) to 3200 m (10 500 ft)] corresponds to the ex- indicated in segment C–D. In contrast, the same eleva- tensive network of valleys that incise the higher terrain tion range on the western side of the state generally regions of Colorado. Equilibrium line altitudes (ELAs) follows lowlands near major river drainages. If the stroke characterize the elevation of formerly glaciated valleys. densities captured in this western portion of the elevation Average late-Pleistocene glaciation ELAs in the San range were not included in the data that forms Fig. 4,the Juan Mountains (Leonard 1984) and in the Front Range slope in segment C–D would be steeper (e.g., higher (Meierding 1982) fall within this elevation class, and in stroke densities). To summarize, the lower elevations as the Sawatch Range (Brugger and Goldstein 1999) are characterized by segment A–D are dominated by a vast, only 175 m above the high end of the class. Examples of contiguous, gently sloping landscape whose structure fa- these incised valleys are visible in Fig. 3 in and around vors upslope ﬂow. the set of three aforementioned mountain ranges. With the exception of the dips at label E and label F in Segment F–G reveals a rapid increase in lightning Fig. 4, the information provided in segment D–F [1829 m density with elevation in the highest 11% of the state. (6000 ft) to 3200 m (10 500 ft)] is limited in its usefulness Most notably, stroke densities rapidly increase above for gaining insight into Colorado’s lightning climatology 3658 m (12 000 ft), the elevations in Colorado generally because, as mentioned earlier, these elevation classes characterized by steep, glacially sculpted, frost-shattered, capture a host of climate types, landforms, and land- near-or-above treeline rocky alpine landscapes (National scapes across the western and eastern slopes of the Col- Park Service 2013). Within segment F–G, stroke densities 2 orado Rocky Mountains. Label E, however, captures the increase from 4.6 km2 yr 1 between 3048 m (10 000 ft) 2 vast (7200 km2), relatively featureless, mountain-bound and 3200 m (10 500 ft; label F in Fig. 4) to 12.4 km2 yr 1 at depositional surface of the San Luis Valley. With an elevations above 4267 m (14 000 ft; label G in Fig. 4).

Unauthenticated | Downloaded 09/29/21 08:27 AM UTC JULY 2014 P I C T U R E O F T H E M O N T H 2359

Likely topographic influences that shape this rapid in- into the analyses. We will focus our future research on crease in lightning activity with elevation are discussed select regions of Colorado, including the Denver Cyclone in a few research papers, but the specific controls are vorticity zone (DCVZ; Szoke et al. 1984) and the San not well understood. Cummins (2012) foundthatinthe Juan Mountains. Specifically, the authors will explore the mountains of Colorado and northern New Mexico the role of the DCVZ in enhancing and limiting lightning highest ground stroke densities occur in regions with activity, and the influence of the southwestern monsoon the highest terrain gradient, with the highest densities on the lightning climatology of the San Juan Mountains. typically occurring more than halfway up mountain slopes. The author offered several possible explanations Acknowledgments. The authors thank Vaisala for for the variability in stroke density in the mountainous providing all stroke data; Jennifer Stark, MIC NWS areas, including surface-driven turbulence, terrain Pueblo, for her support and encouragement; NWS Cen- roughness, along with height and slope variations of the tral Region Headquarters for covering page charges; Ron surface electrical boundary conditions. Bourscheidt et al. Holle and Ken Cummins for their input; and the anony- (2009), in an analysis of lightning flash density over sec- mous reviewers for their suggestions. tions of the Rio Grande do Sul state of Brazil, also observed the greatest flash densities to occur where the REFERENCES terrain slope was greatest. Barker Schaaf, C., J. Wurman, and R. M. Banta, 1988: Thunderstorm- producing terrain features. Bull. Amer. Meteor. Soc., 69, 272– 6. Conclusions 277, doi:10.1175/1520-0477(1988)069,0272:TPTF.2.0.CO;2. Boccippio, D. J., K. L. Cummins, H. J. Christian, and S. J. Goodman, We identified relationships between stroke densities 2001: Combined satellite- and surface-based estimation of the and specific landscape features, such as gently sloping intracloud–cloud-to-ground lightning ratio over the continental surfaces and broad valleys. The increase in lightning United States. Mon. Wea. Rev., 129, 108–122, doi:10.1175/ , . activity by elevation in the lower elevations corresponds 1520-0493(2001)129 0108:CSASBE 2.0.CO;2. Bourscheidt, V., O. Pinto, K. P. Naccarato, and I. R. C. A. Pinto, to the broad and gently sloping surfaces in the eastern 2009: The influence of topography on the cloud-to-ground sections of the state that ascend the Palmer Divide and lightning density in South Brazil. Atmos. Res., 91, 508–513, Raton Mesa, regions of strong topographic forcing and doi:10.1016/j.atmosres.2008.06.010. available low-level atmospheric moisture. With the ex- Brugger, K. A., and B. S. Goldstein, 1999: Paleoglacier re- ception of the vast flat-floored and mountain-bound San construction and late-Pleistocene equilibrium-line altitudes, southern Sawatch Range, Colorado. Glacial Processes: Past Luis Valley, the complex topography captured in the and Present, D. M. Mickelson and J. W. Attig, Eds., Geological intermediate elevation classes blurs the identification of Society of America, 103–112. specific landscape controls on lightning activity. The Cummins, K. L., 2012: On the relationship between terrain varia- relative stroke density minimum found between 3048 m tion and LLS-derived lightning parameters. 2012 Int. Conf. (10 000 ft) and 3200 m (10 500 ft) corresponds to the ex- on Lightning Protection (ICLP), Vienna, Austria, Vienna University of Technology, 9 pp. [Available online at http:// tensive network of steep-walled glacial valleys that dis- ieeexplore.ieee.org/stamp/stamp.jsp?arnumber506344336.] sect the higher elevations of the major mountain ranges ——, and M. J. Murphy, 2009: An overview of lightning location across the state. Our analysis corroborates extant re- systems: History, techniques, and data uses, with an in depth search that suggests a steep terrain gradient and an in- look at the U.S. NLDN. IEEE Trans. Electromagn. Compat., crease in surface-generated turbulence enhance lightning 51, 499–518, doi:10.1109/TEMC.2009.2023450. ——, ——, E. A. Bardo, W. L. Hiscox, R. B. Pyle, and A. E. Pifer, activity in elevations above about 3200 m (10 500 ft). 1998: A combined TOA/MDF technology upgrade of the U.S. Plotting elevation against stroke density year by year National Lightning Detection Network. J. Geophys. Res., 103, from 1996 to 2012 demonstrates that the 2002–03 NLDN 9035–9044, doi:10.1029/98JD00153. upgrades improved stroke detection efficiency in the ——, J. A. Cramer, C. J. Biagi, E. P. Krider, J. Jerauld, M. A. higher elevations (Fig. 1). In Colorado, these enhance- Uman, and V. A. Rakov, 2006: The U.S. National Lightning ments are integral for detecting the lighting–landscape Detection Network: Post-upgrade status. Second Conf. on Meteorological Applications of Lightning Data, Atlanta, GA, relationships above about 3200 m (10 500 ft). Amer. Meteor. Soc., 6.1. [Available online at https://ams. This paper demonstrated that the visual and analytical confex.com/ams/Annual2006/techprogram/paper_105142.htm.] products of GIS, such as maps, histograms, and statistical ESRI, 2013: ArcGIS Desktop: Release 10.2. Environmental Sys- measures, can help create highly detailed lightning cli- tems Research Institute. matologies for user-defined regions. The next steps for Hodanish, S. J., and P. Wolyn, 2011: Lightning climatology for the state of Colorado. Fourth International Lightning Meteorology developing a complete lightning climatology of Colorado Conf., Broomfield, CO, Amer. Meteor. Soc., P8.5. [Available are to explore discrete regions within the state across online at https://ams.confex.com/ams/91Annual/webprogram/ elevation ranges and incorporate a temporal component Paper185671.html.]

Unauthenticated | Downloaded 09/29/21 08:27 AM UTC 2360 MONTHLY WEATHER REVIEW VOLUME 142

Krider, E. P., R. C. Noggle, A. E. Pifer, and D. L. Vance, 1980: 64, 1029–1037, doi:10.1175/1520-0477(1983)064,1029: Lightning direction finding systems for forest fire detec- AECLDN.2.0.CO;2. tion. Bull. Amer. Meteor. Soc., 61, 980–986, doi:10.1175/ ——, G. R. Huffines, W. R. Burrows, R. L. Holle, and K. L. 1520-0477(1980)061,0980:LDFSFF.2.0.CO;2. Cummins, 2002: The North American lightning detection Leonard, E. M., 1984: Late Pleistocene equilibrium-line altitudes and network (NALDN)—First results: 1998–2000. Mon. Wea. modern snow accumulation patterns, San Juan Mountains, Rev., 130, 2098–2109, doi:10.1175/1520-0493(2002)130,2098: Colorado, U.S.A. Arct.Alp.Res.,16, 65–76, doi:10.2307/1551173. TNALDN.2.0.CO;2. Lopez, R. E., and R. L. Holle, 1986: Diurnal and spatial variability ——, ——, ——, and K. L. Cummins, 2011: The North American of lightning activity in northeastern Colorado and central Lightning Detection Network (NALDN)—Analysis of flash Florida during the summer. Mon. Wea. Rev., 114, 1288–1312, data: 2001–09. Mon. Wea. Rev., 139, 1305–1322, doi:10.1175/ doi:10.1175/1520-0493(1986)114,1288:DASVOL.2.0.CO;2. 2010MWR3452.1. MacEachren, A. M., M. Gahegan, W. Pike, I. Brewer, G. Cai, Reap, R. M., 1986: Evaluation of cloud-to-ground lightning data E. Lengerich, and F. Hardisty, 2004: Geovisualization for from the western United States for the 1983–84 summer knowledge construction and decision-support. IEEE Comput. seasons. J. Climate Appl. Meteor., 25, 785–799, doi:10.1175/ Graph. Appl., 24, 13–17, doi:10.1109/MCG.2004.1255801. 1520-0450(1986)025,0785:EOCTGL.2.0.CO;2. Mach, D. M., D. R. MacGorman, W. D. Rust, and R. T. Arnold, Rudlosky, S. D., and H. E. Fuelberg, 2010: Pre- and postupgrade 1986: Site errors and detection efficiency in a magnetic distributions of NLDN reported cloud-to-ground lightning direction-finder network for locating lightning strikes to characteristics in the contiguous United States. Mon. Wea. ground. J. Atmos. Oceanic Technol., 3, 67–74, doi:10.1175/ Rev., 138, 3623–3633, doi:10.1175/2010MWR3283.1. 1520-0426(1986)003,0067:SEADEI.2.0.CO;2. Stall, C. A., K. L. Cummins, E. P. Krider, and J. A. Cramer, 2009: Meierding, T. C., 1982: Late-pleistocene glacial equilibrium-line al- Detecting multiple ground contacts in cloud-to-ground light- titudes in the Colorado Front Range: A comparison of methods. ning flashes. J. Atmos. Oceanic Technol., 26, 2392–2402, Quat. Res., 18, 289–310, doi:10.1016/0033-5894(82)90076-X. doi:10.1175/2009JTECHA1278.1. National Park Service, cited 2013: Natural features in ecosystems. Szoke, E. J., M. L. Weisman, J. M. Brown, F. Caracena, and T. W. [Available online at http://www.nps.gov/romo/naturescience/ Schlatter, 1984: A subsynoptic analysis of the Denver tornadoes naturalfeaturesandecosystems.htm.] of 3 June 1981. Mon. Wea. Rev., 112, 790–808, doi:10.1175/ Orville, R. E., 1991: Lightning ground flash density in the contig- 1520-0493(1984)112,0790:ASAOTD.2.0.CO;2. uous United States—1989. Mon. Wea. Rev., 119, 573–577, U.S. Geological Survey, cited 2013: Science in your state: Colo- doi:10.1175/1520-0493(1991)119,0573:LGFDIT.2.0.CO;2. rado. Denver, Colorado. [Available online at http://www.usgs. ——, 2008: Development of the National Lightning Detection gov/state/state.asp?State5CO.] Network. Bull. Amer. Meteor. Soc., 89, 180–190, doi:10.1175/ Valine, W. C., and E. P. Krider, 2002: Statistics and char- BAMS-89-2-180. acteristics of cloud-to-ground lightning with multiple ——, and G. R. Huffines, 2001: Cloud-to-ground lightning in the ground contacts. J. Geophys. Res., 107, 4441, doi:10.1029/ United States: NLDN results in the first decade, 1989–98. Mon. 2001JD001360. Wea. Rev., 129, 1179–1193, doi:10.1175/1520-0493(2001)129,1179: Zajac, B. A., and S. A. Rutledge, 2001: Cloud-to-ground lightning CTGLIT.2.0.CO;2. activity in the contiguous United States from 1995 to 1999. Mon. ——, R. W. Henderson, and L. F. Bosart, 1983: An east coast Wea. Rev., 129, 999–1019, doi:10.1175/1520-0493(2001)129,0999: lightning detection network. Bull. Amer. Meteor. Soc., CTGLAI.2.0.CO;2.

Unauthenticated | Downloaded 09/29/21 08:27 AM UTC