P348 THE INTRA-CLOUD LIGHTNING FRACTION IN THE CONTIGUOUS UNITED STATES Gina Medici*1, Kenneth L. Cummins1, William J. Koshak2, Scott D. Rudlosky3, Richard J. Blakeslee2, Steven J. Goodman4, Daniel J. Cecil2, David R. Bright5 1Department of Atmospheric Sciences, University of Arizona, Tucson, AZ 2NASA George C. Marshall Space Flight Center/NSSTC, Huntsville, AL 3NOAA/NESDIS/STAR, College Park, MD 4NOAA/NESDIS/GOES-R Program Office, Greenbelt, MD 5NOAA/NWS/NCEP/Aviation Support Branch, Kansas City, MO 1. INTRODUCTION Lightning is dangerous and destructive; cloud-to- an in-depth explanation of lightning and ground (CG) lightning flashes can start fires, interrupt thunderstorms, see books by Rakov and Uman power delivery, destroy property and cause fatalities. (2003), and MacGorman and Rust (1998). Its rate-of-occurrence reflects storm kinematics and microphysics. For decades lightning research has An important study of the relative occurrence of been an important focus, and advances in lightning IC and CG lightning was carried out by Boccippio et detection technology have been essential contributors al. (2001). Here they compiled a four year climatology to our increasing knowledge of lightning. A significant using the Optical Transient Detector (OTD) (Boccippio step in detection technology is the Geostationary et al. 2000) and the U.S. National Lightning Detection Lightning Mapper (GLM) to be onboard the Network (NLDN) (Cummins and Murphy 2009) to Geostationary Operational Environment Satellite R- compute an IC:CG ratio (Z) over the coterminous Series (GOES-R) to be launched in early 2016. GLM Continental United States (CONUS). They concluded will provide continuous “Total Lightning” observations that there is some correlation between terrain [CG and intra-cloud lightning (IC)] with near-uniform (Mountain ranges) and low Z values but the spatial resolution over the Americas by measuring relationship is non-unique. Other results from this radiance at the cloud tops from the different types of study show that high Z values are correlated with lightning (Goodman et al. 2003). These Total locations of severe weather events -- mostly areas Lightning observations are expected to significantly where positive CG lightning is a large fraction of the improve our ability to nowcast severe weather (Schulz CG flashes. This agrees with Carey and Rutledge’s et al. 2011). It may be important to understand the (1998, 2003) research indicating that thunderstorms long-term regional differences in the relative that are typically “dominated” by positive CG lightning occurrence of IC and CG lightning in order to are frequently associated with severe weather (large understand and properly use the short-term changes hail, tornadoes, etc.) and high Z ratios. Other studies in Total Lightning flash rate for evaluating individual also indicate that severe storms typically produce storms. unusually high rates of IC lightning and (to some degree) reduced CG lightning (MacGorman et al. A typical (simplified) electrified cloud is explained 1989; Williams et al. 1999; Wiens et al. 2005). as having vertical layers of electrical charge composed of an upper positive charge below the Given the potential value of understanding long- tropopause, a midlevel negative charge region just term regional variations in the Z ratio, we have above the freezing level, and a much smaller positive expanded upon the earlier analysis by Boccippio et al. charge layer below it (Williams 1989). IC flashes (2001) through the use of additional and longer-term generally “neutralize” charge between the upper datasets. Section 2 describes the datasets and positive and midlevel negative charge regions in a analysis methods. Results are presented in section 3 cloud while CG flashes typically transfer negative followed by the discussion (section 4) and charge to one or more locations on the ground conclusions (section 5). (Cummins and Murphy 2009). The relative occurrence 2. DATA AND METHODS of IC and CG lightning is thought to be determined in- part by the relative locations of these charge centers, This study uses the U.S. National Lightning and their spatial relationship to the terrain below. For Detection Network (NLDN), the Optical Transient Detector (OTD), the Lightning Imaging Sensor (LIS), *Gina Medici, University of Arizona, Department of and a combination of OTD and LIS (OTD/LIS). The Atmospheric Science, Tucson, AZ, 85719; email: [email protected] following sub-sections briefly discuss these datasets IC flashes has steadily increased over the last 8 and the methods used to evaluate and compare them. years. In the present study, only NLDN-reported CG flashes are employed. A history and implications of 2.1. NLDN data NLDN upgrades is provided in Koshak et al. 2014. Given the high CG flash DE for the NLDN, no DE The NLDN has been providing lightning data corrections were applied to the NLDN flash density since the early 1980s and in 1989 began to be used values. for continental-scale lightning research in the United States (Cummins and Murphy 2009). The NLDN was 2.2. OTD and LIS originally made up of gated wide-band magnetic direction finders that employed magnetic field OTD orbited for 5 years (1995 -2000) on the waveforms to determine the direction to the channel Orbital Sciences Corporation Microlab-1 Satellite (OV- bases of lightning discharge to the ground (Krider et 1) (Mach et al. 2007 and Christian et al. 2003). It al. 1976). The first major improvement to the NLDN detected and located lightning during both day and occurred in 1995 when the direction finders were night due to its sensitivity and dynamic range upgraded to include GPS timing data, resulting in the (Christian et al. 2003). OTD recorded lightning so-called IMPACT (Improved Accuracy through between 75 degrees North and South due to its 70 Combined Technology) sensor (Cummins et al. degree inclination orbit (Christian et al. 2003). The 1998a). The IMPACT geo-location algorithm flash DE has been reported to be between 49% and computes the latitude, longitude and discharge time 65% (Boccippio et al. 2000 and Boccippio et al. using as few as two sensors (Cummins and Murphy 2001). OTD’s field-of-view was about 1300×1300 km2 2009). The CG flash detection efficiency (DE) with a spatial resolution of 10 km and about 14 orbits following this upgrade ranged between 80 and 90%, each day (Cecil et al. 2014). depending on location (Cummins et al., 1998a). In 2002–03, the NLDN improved as a result of replacing LIS is part of the Tropical Rainfall Measuring all NLDN sensors with better IMPACT-ESP sensors Mission that was launched in 1997 and remains in and also adding eight additional sensors to the orbit but it is on its last years. Like OTD, LIS detects network. This further improved the flash DE to the total lightning (IC and CG flashes), but is limited to between 90 and 95% (Cummins and Murphy 2009). 38° N to 38° S with a flash DE of about 69% during The spatial boundaries of the NLDN are 250 km into local noon and 88% at night (Cecil et al. 2014). LIS’s Canada, 600 km into Mexico, 600 km into the Pacific field-of-view changed from about 600×600 km2 to and Atlantic Ocean (Holle 2014). The flash DE about 700×700 km2 following a boost in the TRMM decreases in all directions outside CONUS except satellite average altitude from ~350 km before August Canada since there is the Canadian Lightning 2001 to ~400 km after August 2001. The respective Detection Network (CLDN) that operates in spatial resolution changed from ~5 km to ~6 km. LIS conjunction with the NLDN (Holle 2014). Figure 1 has about 16 orbits each day (Cecil et al. 2014). provides a contour map of the estimated flash DE for This analysis uses the global Total Lighting the combined NLDN and CLDN since late 1998. gridded OTD and LIS datasets produced by NASA Finally, as a result of the 2013 NLDN upgrade, the (see Cecil et al. 2014 for a detailed description). More estimated CG flash DE throughout CONUS is in specifically, we employ the gridded flash rate (flash excess of 95% (see Nag et al., 2013 for details). Prior density in units of flashes/km2/yr) product which is to 1996, the NLDN did not report any discharges part of the High Resolution Flash Climatology (HRFC) classified as cloud lightning. The number of reported dataset with a spatial resolution of 0.5×0.5 degrees. The measured flash counts were scaled to correct for observation times and flash detection efficiency, as described in Cecil et al. 2014. This study examines the OTD and LIS data for the period of May of 1995 to December 2012. The combined OTD/LIS data provide the Total Lightning portion while NLDN is used for CG flashes, making it possible to determine the Z ratio in the same manner as Boccippio et al. (2001) who utilized four years of OTD and NLDN data. The result is a 17.5 year flash density climatology below 38 N and a 5-year climatology above 38 N. For most of the analyses the data were smoothed based on Gaussian smoothing with a standard Figure 1: A contour map of the estimated flash Detection deviation of one grid (0.5 degrees) using 5×5 grid Efficiency for the combined NLDN and CLDN since late points. All smoothing was preformed prior to any 1998. Reprinted with permission. arithmetic manipulations (ratios, differences, etc.). This improves numerical stability since the data sets 𝑇𝑜𝑡𝑎𝑙−𝐶𝐺 𝑍= (2) (OTD in particular) are small due to orbital sampling. 𝐶𝐺 Smoothing also helps reduce the impact of inter- annual variability. where Total is the composite OTD and LIS density, and CG is the composite NLDN cloud-to-ground In this work, we also compare the individual OTD density. Spatial maps of Z were computed and and LIS datasets below 38 N to evaluate represented in two different ways. In the Results instrumentation or calibration biases in the datasets section, Z is plotted as a smoothed grid map with the and set expectations for the variability in the 5-year “native” 0.5×0.5 degree resolution. The Discussion OTD climatology in the northern latitudes. The section includes Z plotted as a highly smoothed comparisons are carried out in two different ways -- contour map for direct comparison with Fig. 2 in one is a signed (+/-) spatial bias percent for each grid Boccippio et al (2001). When the CG flash density is point and the other is a magnitude error percent. The small, the calculated value of Z will be sensitive to spatial bias percent shows locations where there is a small random variations in that value. The “cloud strong bias towards one data set, using the equation: fraction” (CF) does not suffer from this instability as much as Z, and is given by the following equation: 100∗(𝑂𝑇𝐷−𝐿𝐼𝑆) (𝑂𝑇𝐷+𝐿𝐼𝑆)/2 . (1) 𝑇𝑜𝑡𝑎𝑙−𝐶𝐺 𝐶𝐹= (3) 𝑇𝑜𝑡𝑎𝑙 The magnitude error percent is simply the absolute value of the signed error, and is used to Figure 2 shows the fractional sensitivity of Z and CF show the locations where the two data sets are very to a 1% change in CG fraction (Sz and SCF, different. respectively), determined from the derivative of these functions with respect to CG fraction. The sensitivity equations used are presented below: 𝐶𝐺 𝑆 = . (4) 𝐶𝐹 𝐶𝐹 𝐶𝐺𝐹−1 1 𝑆 =𝐶𝐺[ 𝐶𝐺𝐹2 −𝐶𝐺]= 𝑆𝐶𝐹 = 𝐶𝐺 . (5) 𝑧 𝑍 𝐶𝐺𝐹 𝐶𝐹∗𝐶𝐺𝐹 where the absolute value is taken and the CG Fraction (CGF) = 1-CF. It is evident that CF is more stable than Z for small values of CGF. Sz is inversely related to the smaller of CF and CGF and is therefore insensitive to the CG Fractions between 0.2 to 0.8 (see Fig. 2), but has high sensitivity elsewhere. Scf is inversely related to Cloud Fraction, resulting in fairly unstable behavior for CG Fraction > 0.8. Since real storms do not produce 4 times more CG flashes than Figure 2: The fractional sensitivity of Z and CF to a 1% IC flashes, this sensitivity is not a practical problem. change in CG fraction, determined from the derivative of these functions with respect to CG Fraction. 3. RESULTS 3.1. LIS and OTD differences An NLDN “composite” CG flash density dataset was constructed to match the time periods of the Figures 3 and 4 show the OTD and LIS Flash satellite-derived climatologies. For the composite densities (respectively) taken from the HRFC. LIS OTD/LIS dataset, the NLDN data above 38N was was limited to 37 N to discard the bias error from the limited to the early 5 years period, and the NLDN data edges due to smoothing. The Total Lightning flash below 38N included all 17.5 years of data. densities range from 35-40 fl/km2/yr in Florida and western Mexico, to less than 1 along the U.S. west A “composite” Z was also calculated using the coast and north-eastern Canada. It is clear that both combined OTD/LIS data set that is from May 1995 – OTD and LIS generally agree about the maxima over December 2012 and the NLDN data as described Florida, Cuba, and western Mexico, but there are above. Z is calculated for each grid point using: some differences between the datasets over the rest of the domain. Figure 4: The average flash density over CONUS from the LIS Figure 3: The average flash density over CONUS from OTD data from 1998–2012. data from 1995–99. Figure 5: The signed difference between OTD and LIS flash Figure 6: The absolute difference between OTD and LIS densities, illustrating the spatial bias of the data being used. flash densities. Figure 5 shows the signed spatial bias percent flash datasets only have two years of overlap. Given that densities,between the two sets. Negative (green/blue) the LIS data is a 14-year climatology, it would be values represent a bias towards LIS and positive reasonable to ascribe most of the variability to the (orange/red) values represent a bias towards OTD. OTD dataset. White regions are either “no data” or indicate biases less than 10%. The values mostly vary between ± There are moderate magnitude differences over 40% with the highest locations of spatial bias in parts CONUS as shown in Fig. 6, with the larger variations of the Gulf of Mexico, along the west coast of the Gulf of roughly 50% in south-central and eastern Texas. of California, and east of the Gulf Stream off the east The same “heterogeneous” pattern seen in the spatial coast. For most of the United States (below 37° N) bias plot is also seen in Fig. 6. A histogram of the and over the water, visual inspection suggests that magnitude differences for all grids is shown in Fig. 7 there may be some bias towards higher OTD density (bar graph), along with the associated cumulative since there are more positive values. However, the distribution (line graph). About 90% of the grids have average over the whole domain shows that the LIS a variation of 50% or less, with steadily decreasing reported 4% more lightning than OTD. If the bias were likelihood of larger variations. due to instrumental differences one would expect a more uniform bias towards either LIS or OTD. However, the largest variations between the two are spatially very close to each other, going from a negative extreme to a positive extreme. This finding is more likely due to the orbital sampling and year-to- year variations in storm location since these two Figure 7: Cumulative sum (line plot) and the histogram (bars) Figure 8: The 17.5 year climatology of the average flash for the magnitude error over CONUS between OTD and LIS. density over CONUS from the combination of LIS and OTD data from HRFC. 3.2. Combined OTD/LIS and NLDN smaller values over the Ocean, and enhancing again over the Gulf Stream. This is discussed further in An underlying limitation of this analysis is the section 4. short (5-year) observation period for the combined satellite-derived climatology above 38 N, shown in 3.3. Z and Cloud Fraction Fig. 8. It might be possible to gain some insight into Z and Cloud Fraction are plotted in Figs. 11 and the implication of this limitation by comparing the NLDN climatologies in Figs. 9 and 10. Figure 9 is the 12 respectively based on the modified NLDN sample NLDN CG flash density that is time-associated with which matches the OTD and LIS periods. There is no the combined OTD/LIS flash density, as described in clear transition from land to water, with the possible exception off the coast from New Jersey, and there is the Methods section. Figure 10 shows the NLDN CG flash density for the whole 17.5 year period over no clear anomaly over the Gulf Stream. The high Z CONUS. Noteworthy contrasts between Figs. 9 and boundaries over the ocean and northern Mexico are 10 are lower-density “dips” seen in Missouri and Ohio due to the fall-off of NLDN DE with increasing distance from CONUS, as shown in Fig. 1. Most of with differences in the northern plains from northwest Kansas northward, Pennsylvania and parts of New the United States exhibits values of Z between 1 and York. There were a lot of storms over the 17.5 year 4 (CF between 0.5 and 0.8), with some notable period that are not included in the smaller 5 year exceptions. Distinctly high Z values occur in parts of Northwest Texas, Kansas, Nebraska and South climatology, resulting in a less-representative flash density in the north. Also, as latitude increases the Dakota, more-clearly illustrated in Fig. 11. This region flash density decreases so this area may be more is known from previous studies to be associated with sensitive to changes of sample size than the high percentages of positive CG flashes and severe southeastern U.S. Among all the density maps it is weather (Carey and Buffalo 2007; Carey et al. 2003; Orville et al. 2011).) Large Z values (>8) are also seen clear that off the coast of the Carolina’s the density transitions from high values over land, decreasing to in the Northwest U.S., Vancouver Canada, and off the Figure 9: The National Lightning Detection Network’s Figure 10: The National Lightning Detection Network’s Ground Flash Density over CONUS where below 38° N the Ground Flash Density (flashes km-2 yr-1) over CONUS climatology is 17.5 years and above 38° N the climatology based on a 17.5 year climatology. is 5 years following OTD. Figure 12: The Cloud Fraction using LIS/OTD and NLDN for Figure11: The Z ratio using LIS/OTD and NLDN for a 17.5 a 17.5 year period. year period. northern California coast. This finding may not be lightning over this area compared to near-coastal significant due to the low flash density in these areas. waters has been associated with almost stationary Small regions of high Z that occur over the Great Salt convective clouds and precipitation associated with Lake, Lake Huron, and Vermont could be significant. large fluxes of heat and water vapor from the warm They are all within large regions of low ground flash waters of the Gulf to the colder air above (Biswas and density (0.5 – 2 fl/km2/yr), but their feature size is Hobbs 1990). A more in-depth discussion on the much smaller than the surrounding regions of low meteorological mechanisms for lightning in the U.S. ground flash density. can be found in Holle et al. (2010) and Holle (2014). Other notable references are Smith et al. (2005), and 4. DISCUSSION Lopez and Holle (1986). 4.1 Overall Lightning Climatology 4.2 LIS/OTD Both the Satellite and NLDN observations The largest differences between the OTD and reveal similar general patterns of flash density over LIS climatologies are seen in the Gulf of Mexico, parts CONUS. The greatest CONUS flash densities are in of Mexico, and the Caribbean. The largest source of Florida and the Gulf Coast, with a nearly steady fall- variability seems to be OTD and LIS covering different off to the west, northwest, and north. The 17.5-year time periods, interacting with inter-annual variability satellite climatology shows flash density maxima and the limited sampling period for orbital satellites. along the west coast of Mexico, as indicated by Additional work is required to demonstrate this Murphy and Holle (2005). Other maxima regions quantitatively. There does not appear to be significant include the Gulf Stream and south-central U.S. instrumental biases. An important fact is that OTD is Furthermore, the Front Range of the Rocky only for 5 years while LIS is for 14 years, suggesting Mountains has relatively greater density values when that most of the variability is in the OTD climatology. compared to the central Rockies. Thus, the composite satellite-derived climatology will have more variability and uncertainty north of 38° An important contributor for lightning along the latitude, with an expected percent variability similar to Gulf Coast is deep low-level moisture driven from the those depicted in Figs. 5-7. very warm ocean waters (Holle 2014). When supplemented by coastal land-mass heating, 4.3: Z and Cloud Fraction conditions are ideal for strong convection (Stroupe et al. 2004). There is also an area of high flash density Very high Z values occur over the Northwest over Florida. The driving factor for lightning here is the U.S., near Vancouver Canada, and off the northern differential heating resulting from the thermal California coast. This finding may not be significant contrasts between land and water, helping to create due to the low flash density in this area, but it is convergent boundaries that help trigger convection interesting that this general area also exhibits high (Hodanish et al. 1996). percentages of positive CG lightning (see Orville et al. 2011, Fig. 4j). The low flash densities in this area The well-defined lightning increase across the may result from the cold water and large-scale sinking Gulf Stream off the East Coast is because the Gulf which inhibits deep convection, with local variations Stream consists of warm waters that favor deep produced by the terrain-driven convection typically convection (Christian et al. 2003). The increased seen in the West (Reap 1986). (b) (a) Figure 13:(a) Boccippio et al. 2001 IC:CG ratio and (b) new IC:CG ratio. Both are smoothed and contoured the same. Both Z and Cloud Fraction exhibit high values in strong correlation between high Z and positive CG parts of northwest Texas, Kansas, Nebraska, and the lightning in the Central United States observed by Dakotas, which is a stable and dominant feature. Boccippio et al. (2001) remains as a key observation. Carey and Rutledge (2003) observed that in the High Z ratio values along the U.S. west coast may be Upper Great Plains, severe storms have a large an observational problem associated with the low impact on the mean annual Z and the associated high flash densities that occurs in those areas. percentage of positive CG lightning. Typically what is seen throughout CONUS is that around 80% of warm- This work provides the first assessment of Z over season severe storms produce mostly negative CG coastal waters. Both the NLDN and satellite dataset flashes, but about 20% have a large (>25%) fraction reported enhanced lightning over the Gulf Stream, but of positive polarity CG flashes (Carey and Buffalo there was no clear variation in the Z ratio in this 2007). They also note that a large fraction of severe region. storms are “positive dominant” storms in central and Future work will address the sources of variability Northern plains from the Texas Panhandle (inter-annual variability vs. satellite sampling northwestward to Minnesota. The meteorological limitations) in order to place quantitative bounds on reasons for these anomalies are discussed in Carey uncertainty in the Z ratio and Cloud Fraction and Buffalo (2007) and Bruning et al. (2012). climatologies. Our long-term objective is to extend this Figure 13 compares the Boccippio et al. (2001) 4- analysis to a global IC:CG climatology employing the year Z ratio climatology and our new 17.5 year complete OTD/LIS dataset and the Global Lighting climatology. Both studies show a clear maximum over Dataset (GLD360), in order improve our Kansas, Nebraska, and the Dakotas, as well as understanding of lightning behavior throughout the maxima over Washington, Idaho, Oregon, and world. Northern California. This is expected given that our REFERENCES climatology above 38N is only 25% larger (5 years vs. 4 years). Similar patterns are seen over Eastern Biswas, K. R., and P. V. Hobbs, 1990: Lightning over United States and over the Rocky Mountains. the Gulf Stream. Geophys. Res. Lett., 17, 941–943. However, there are significant differences over Texas with lower Z values for the updated climatology, Carey, L. D and K. M. Buffalo, 2007: Environmental resulting in nearly-constant Z values between 1.5 and Control of Cloud-to-Ground Lightning Polarity in 2.0 from the southern inter-mountain west, through Severe Storms. Mon. Wea. Rev., 135, 1327–1353. the gulf coast, and throughout the eastern U.S. We ascribe these differences to a much longer Boccippio, D. J., and Coauthors, 2000: The Optical observation period south of 38°N. Transient Detector (OTD): Instrument characteristics and cross-sensor validation. J. 5. CONCLUSIONS AND FUTURE WORK Atmos. Oceanic Technol., 17, 441–458. This study extends the IC:CG ratio (Z) climatology study carried out by Boccippio et al. (2001) by employing larger datasets and including a Cloud Fraction analysis. It used OTD, LIS, and NLDN data for a 17.5 year period from May 1995 through December 2012 (OTD for the first 5 years). The Boccippio, D. J, K. L. Cummins, H. J. Christian, and Hodanish S., D. Sharp, W. Collins, C. Paxton, and S. J. Goodman, 2001: Combined Satellite- and R.E. Orville, 1997: A 10-yr Monthly Lightning Surface- Based Estimation of the Intracloud-Cloud- Climatology of Florida: 1986–95. Wea. to-Ground Lightning Ratio over the Continental Forecasting, 12, 439–448. United States. Monthly Weather Rev., 129, 108 - 122. Hodanish, S., and P. Wolyn, 2012: Lightning climatology for the state of Colorado. Preprints, Bruning, E. C., S. A. Weiss, and K. M. Calhoun, 2012: Fourth Int. Lightning Meteorology Conf., Broomfield, Continuous variability in thunderstorm primary CO, Vaisala, 12 pp. [Available online at electrification and an evaluation of inverted-polarity http://www.vaisala.com/en/events/ildcilmc/Documen terminology. Atmos. Res., 135–136, 274–284. ts/Climatology/Lightning%20Climatolog%20for%20t he%20State%20of%20Colorado.pdf.] Carey, L.D. and S. A. Rutledge, 1998: Electrical and multiparameter radar observations of a sever Holle R.L, 2014: Diurnal Variations of NLDN-Reported hailstorm. Journal of Geophysical Research, 103, Cloud-to-Ground Lightning in the United 13,979 – 14,000. States. Mon. Wea. Rev., 142, 1037–1052. Carey, L. D., and S. A. Rutledge, Holle, R. L., and K. L. Cummins, 2010: Monthly 2003: Characteristics of cloud-to-ground lightning in distributions of U.S. NLDN cloud-to-ground severe and nonsevere storms over the central lightning. Preprints,Int. Lightning Meteorology United States from 1989–1998, J. Geophys. Conf., Orlando, FL, Vaisala, 13 pp. Available online Res., 108, 4483, doi:10.1029/2002JD002951, D15. at www.vaisala.com/Vaisala%20Documents/Scientific Cecil, D. J., D. E. Buechler, and R. J. Blakeslee, %20papers/6.Holle,%20Cummins.pdf 2014: Gridded lightning climatology from TRMM- LIS and OTD: Dataset description. Atmos. Res., Koshak, W.J., K.L. Cummins, D.E. Buechler, B. Vant- 135-136, 404-414. Hull, R.J. Blakeslee, E.R. Williams, and H.S. Christian, H. J., Hugh J. Christian, R.J. Blakeslee, D. Peterson, 2014: Variability of CONUS Lightning in J. Boccippio,W. L. Boeck, D.E. Buechler, K.T. 2003-12 and Associated Impacts. J. Appl. Met. & Driscoll, S. J. Goodman, J. M. Hall, W. J. Koshak, Clim., 54, 15-41, dio: 10.1175/JAMC-D-14-0072.1. D. M. Mach, and M. F. Stewart, 2003: Global frequency and distribution of lightning as observed Krider E.P., R.C.Noggle, and M.A.Uman, 1976: A from space by the Optical Transient Detector, J. gated wideband magnetic direction-finder for Geophys. Res., 108(D1), 4005. lightning return strokes. J. Appl. Meteorol., 15, 301– 306. Cummins, K. L., M. J. Murphy, E. A. Bardo, W. L. Hiscox, R. B. Pyle, and A. E. Pifer, 1998: A Lopez R.E. and R. L. Holle, 1986: Diurnal and Spatial combined TOA/MDF technology upgrade of the Variability of Lightning Activity in Northeastern U.S. National Lightning Detection Network. J. Colorado and Central Florida during the Geophys. Res., 103, 9035–9044. Summer. Mon. Wea. Rev., 114, 1288–1312. Cummins, K. L., and M. J. Murphy, 2009: An overview MacGorman D.R., D. W. Burgess, V. Mazur, W. D. of lightning locating systems: History, techniques, Rust, W. L. Taylor, and B. C. Johnson, 1989: and data uses, with an in-depth look at the U.S. Lightning rates relative to tornadic storm evolution NLDN. IEEE Trans. Electromagnetic Compatibility, on 22 May 1981. J. Atmos. Sci., 46, 221–250. 51, 3, 499-518. MacGorman and Rust, 1998: The Electrical Nature of Cummins, K. L., 2012: Analysis of multiple ground Storms, Oxford University Press. Mach, D. M., H. J. contacts in cloud-to-ground flashes using LLS data: Christian, R. J. Blakeslee, D. J. Boccipio, S. J. The impact of complex terrain. Preprints, 22nd Int. Goodman, and W. L. Boeck, 2007:Performance Lightning Detection Conf., Broomfield, CO, Vaisala, assessment of the Optical Transient Detector and 7pp. Lightning Imaging Sensor. J. Geophys. Res., 112, D09210. Goodman, S. J, R. J. Blakeslee, W. J. Koshak, D. Mach, J. Bailey, D. Buechler, L. Carey, C. Mach, D. M., H. J. Christian, R. J. Blakeslee, D. J. Schultz, M. Bateman, E. McCaul Jr. and G. Stano, Boccipio, S. J. Goodman, and W. L. Boeck, 2007: 2013: The GOES-R Geostationary Lightning Performance assessment of the Optical Transient Mapper (GLM). Atmos. Res., 125–126, 34–49. Detector and Lightning Imaging Sensor. J. Geophys. Res., 112, D09210. Hobbs, P.V., 1987: The Gulf Stream rainband, Geophys. Res. Letts, 14, 1142-1145. Martin J. Murphy and R. L. Holle, 2005: Where Is the Williams E., B. Boldi, A. Matlin, M. Weber, S. Real Cloud-to-Ground Lightning Maximum in North Hodanish, D. Sharp, S. Goodman, R. Raghavan, America?. Wea. Forecasting, 20, 125–133. and D. Buechler,:1999. The behavior of total lightning activity in severe Florida thunderstorms. Atmos. Res., 51,245–265. Nag, A., M. J. Murphy, K. L. Cummins, A. E. Pifer, and J. A. Cramer, 2013: Upgrade of the U.S. National Lightning Detection Network in 2013, paper presented at International Symposium on Lightning Protection (XII SIPDA), Belo Horizonte, Brazil, 5 pp Orville 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 E.R., G.R.Huffines, W.R. Burrows, K.L. Cummins, 2011: The North American Lightning Detection Network (NALDN): Analysis of Flash Data – 2001-2009. Mon. Wea. Rev., 139. Rakov, V.A. and M. A. Uman, 2003: Lightning: Physics and Effects. New York: Cambridge Univ. Press. Reap, R. M., 1986: Evaluation of cloud-to-ground lightning data from the western United States for the 1983–84 summer seasons. J. Climate Appl. Meteor., 25, 785–799 Schultz,C.J., W. A. Petersen, and L. D. Carey, 2011: Lightning and Severe Weather: A Comparison between Total and Cloud-to-Ground Lightning Trends. Wea. Forecasting, 26, 744–755. Smith J.R., H.E. Fuelberg, and A. I. Watson, 2005: Warm Season Lightning Distributions over the Northern Gulf of Mexico Coast and Their Relation to Synoptic-Scale and Mesoscale Environments. Wea. Forecasting, 20, 415–438. Stroupe, J.R., A.I. Watson, H.E. Fuelberg, K.G. Kuyper, S.K. Rinard, and M.C. Koziara, 2004: Incorporating mesoscale lightning climatologies into the NWS IFPS/GFE forecast routine along the Gulf Coast. Preprints, 20th International Conference on Interactive Information and Processing Systems (IIPS) for Meteorology, Oceanography, and Hydrology, January 11-15, Seattle, WA, American Meteorological Society, 11 pp. 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,. Williams, E. R., 1989: The tripole structure of thunderstorms. J. Geophys. Res., 94(D11),13151– 13167.