1. Introduction
Regional cloud-to-ground lightning flash densities were first calculated approximately 15 years ago using lightning network data from northeastern Colorado (López and Holle 1985). They calculated the diurnal and spatial variability of lightning activity in northeastern Colorado during the summer. Subsequent regional studies have included the distribution of summertime lightning as a function of low-level wind flow in Florida (López and Holle 1987), cloud-to-ground lightning patterns in New Mexico (Fosdick and Watson 1995), diurnal cloud-to-ground lightning patterns in Arizona (Watson et al. 1994), and an 8-yr study of lightning in the southeastern United States in preparation for the 1996 Summer Olympics (Watson and Holle 1996).
In this paper, we continue our examination of the characteristics of cloud-to-ground lightning for the United States first published for the years 1989–91 (Orville 1991, 1994) and for the period 1992–95 (Orville and Silver 1997). We extend the analysis to cover the years 1995–97, repeating the year 1995 to show the set of three years following the upgrade to the National Lightning Detection Network (NLDN) in 1994 and discussed recently by Cummins et al. (1998). In addition, we extend our analyses to cover additional properties of the cloud-to-ground flash, including median peak currents for negative and positive lightning and the multiplicity for negative flashes.
The characteristics of cloud-to-ground lightning flashes are of fundamental interest. Lightning flash characteristics expressed as a function of time and location are of interest for their variation throughout the year. These characteristics include, 1) number of flashes per unit area, 2) number of positive flashes per unit area, 3) the percentage of positive lightning, 4) the median peak current for both positive and negative flashes, and 5) the flash multiplicity.
We will see that the measured number of cloud-to-ground flashes is approximately 25 million per year. Maximum flash densities occur over Florida, along the Louisiana coast, and in the Midwest. During the three years reported, the positive flash count increased from 2.1 to 2.7 million flashes with approximately 10% of ground flashes being positive, significantly higher than the 5% measured in previous years (Orville and Silver 1997). Median negative peak currents have maximum values along the coastlines, while median positive peak currents have the highest values in the upper Midwest, typically exceeding 40 kA. Multiplicity values or the number of strokes per flash are shown to vary from just over one in the West and the state of Maine to over 3.0 in the southeastern United States. These data, results, and discussion are presented in the following sections.
2. Data
All data analyzed were obtained by the NLDN operated by Global Atmospherics, Inc., Tucson, Arizona. The network consists of 106 sensors and was upgraded to a combination of Improved Accuracy from Combined Technology (IMPACT) and time-of-arrival (TOA) sensors in the fall of 1994. A description of the upgrade, the improved network characteristics, and the location of the sensors is given by Cummins et al. (1998). In previous years, we have assumed a detection efficiency of 70% and multiplied the measured values by 1.4. In the present analyses, we recognize that the detection efficiency is as high as 80%–90% (Idone et al. 1998) in some parts of the upgraded network and may be this high throughout the network. Consequently, we make no corrections for the detection efficiency and plot the measured flash density.
Since 1994 location errors are on the order of 500 m (Idone et al. 1998), a significant improvement over the preupgrade estimate of 10 km. All geographical plots in this paper are done with a spatial resolution of 0.2°, corresponding to an approximate spatial resolution of 20 km. Variations of flash parameters occurring on a smaller scale will of course not be resolved. A rectangular area with 0.2° sides corresponds to 425 km2 at 30°N and 350 km2 at 45°N. We note that the previous flash density analyses by Orville and Silver (1997) were calculated with a spatial resolution of 60 km.
3. Results
All results are summarized in categories of flash density, percent positive, peak currents, and multiplicity.
a. Flash density
The results of the ground flash count are presented in Table 1 for the years 1995–97. Total flashes refers to the combined number of negative and positive flashes (columns 2–4) and positive flashes refers to just those flashes lowering positive charge to ground (columns 5–7). A histogram of the monthly flash counts is shown in Fig. 1 and the July maximum for each year is apparent. If we plot the geographical distribution of the annual flash density for the years 1995–97, we obtain Fig. 2 where we have added a fourth panel to show the mean flash density for the three years. The highest values, exceeding 11 flashes km−2, occur in Florida, Kentucky (1996), and along the Louisiana coast. These flash densities are the measured flash densities and are not corrected for detection efficiency.
The positive flash counts are reproduced in Fig. 3 for each year as a function of month. The geographical variation of the positive flash density is shown in Fig. 4 for the years 1995, 1996, and 1997 with a fourth panel showing the 3-yr mean. High values are apparent in Florida, Louisiana, and Tennessee in all three years. The reasons for this are not obvious.
b. Percent positive polarity
The monthly and annual percentage of positive lightning is summarized in Table 2 and graphed as a histogram in Fig. 5. Annual values are approximately 10% but the monthly values range from a low of 6.5% in July 1995 to a high of 24.5% in January 1996. Clearly, the lowest percentages dominate in the summer and the highest values in the winter.
The percentage of positive lightning has a large geographical variability. Figure 6 shows the distribution of positive lightning throughout the continental United States for the years 1995, 1996, and 1997, and finally the mean of the three years in the fourth panel. The years 1996 and 1997 show values exceeding 25% in the upper Midwest. Similarly, a high percentage positive is apparent along the West Coast, in parts of Tennessee, and in Maine. Areas exceeding 25% appear in Louisiana in 1997.
c. Median peak currents
We have examined the median peak currents for each month and for each year using the Cummins relation for converting the peak signal strength to peak current (Cummins et al. 1998). All peak currents are included in this analysis. We are aware of the recommendation by Cummins et al. that positive peak currents below 10 kA may be regarded as cloud discharges, but we do not adopt this recommendation in our analyses. The extent of the problem is poorly understood at the moment. Our result for all median peak currents is presented in Fig. 7. Because the monthly median peak current for each year is nearly the same, we have averaged the values by month and plotted the result for negative and positive flashes. Note that the median negative peak current for each month varies only slightly from 20 kA, except for December when it reaches 24 kA. The median positive peak current, however, oscillates from 25 kA in February to a low of approximately 14 kA in July, to 24 kA in December. Inherent in this calculation is the assumption that the velocity of the return stroke is a constant for positive flashes and does not vary with season.
The geographic distribution of peak currents is shown in Figs. 8 and 9. Figure 8 indicates that the highest negative peak currents are found along the coastal areas and in a sector common to Pennsylvania, Ohio, and West Virginia. This is readily apparent in the fourth panel showing the average for the three years. Low median peak currents, on the order of 10–14 kA, are measured in the western area of the Carolinas and this will be explained further in the discussion as a result of several sensors closer than 75 km, giving the NLDN an enhanced capability to record weak flashes.
Figure 9 shows the geographic distribution of median positive peak currents for the years 1995, 1996, and 1997. The fourth panel shows the median for the three years, 1995–97. Note the high positive values in the upper Midwest that exceed 40 kA. In 1997 the values in yellow and red, indicating peak currents greater than 30 and 40 kA, respectively, extend in a more or less continuous pattern from the Canadian border to Nebraska and Kansas. This upper Midwest area is where sprites were first observed (Winckler et al. 1993) and continue to be reported (Lyons 1994, 1996).
d. Multiplicity
The geographical distribution of multiplicity, or the number of strokes per flash, is shown in Fig. 10 for three individual years and then averaged in the fourth panel for the period 1995–97. Note that the multiplicity values systematically increase from the northwest to the southeastern United States. The highest multiplicity values, exceeding three strokes per flash, are found in South Carolina and Georgia (1997) and in Florida for all three years (1995–97). Multiplicity values for positive flashes have been plotted but omitted from this paper because they show no consistent pattern and range only from 1.0 to slightly greater than 1.2 strokes per flash.
4. Discussion
Understanding the results in this paper requires that we separate the natural variations in the lightning flash characteristics from the effects produced by the NLDN instrumentation and network configuration. In Fig. 11 we have plotted the locations of the NLDN sensors, which are a combination of IMPACT and of TOA sensors (Cummins et al. 1998). Note the cluster of IMPACT sensors in the western Carolinas that we show in the inset. A lightning flash recording by the NLDN requires that at least one IMPACT sensor detect the flash. If only TOA sensors detect the flash, it is rejected. If we take the distance between the IMPACT sensors in Fig. 11 and contour this separation throughout the United States, the result is Fig. 12. This figure shows that the sensor separation ranges from 525 km in a narrow region from Montana to Kansas, to a “bull’s-eye” of 75 km in South Carolina. Consequently, we expect lightning with low peak currents to be detected with a higher efficiency in the Carolinas.
a. Flash density
Flash density maxima in Fig. 2 occur in the upper Midwest, along the Louisiana coast, and in Florida. The flash density plot in Fig. 4 shows a maximum in the western Carolinas. This can be ignored as a network artifact resulting from the enhanced detection efficiency in this region. In Fig. 4, the positive flash density maxima occur in Kentucky–Tennessee, Texas–Louisiana, and in Florida. In both Figs. 2 and 4, there appear to be two east–west maxima bands in Florida, one in the north and one in the south. A lower lightning flash density occurs over the Appalachian Mountains in Figs. 2 and 4.
b. Percent positive polarity
The percentage of positive lightning plotted in Fig. 5 (Table 2) undergoes a significant variation with month, reaching a minimum in July and August, increasing to a maximum in the winter months of January and February. Orville and Silver (1997) reported similar results. The measured annual percentage of positive lightning has increased significantly from approximately 4%–5% in 1992–94 to 10% in 1995–97. This increase is documented by Wacker and Orville (1996) and is apparently caused by a change in the detection criteria in the NLDN (Cummins et al. 1998). It remains to be determined whether the increase in the percentage of positive lightning is the result of an increase in the detection of positive flashes that were previously missed or an increase in the contamination of intracloud flashes, or a combination of both. Note that the areas of high percent positive in the southeast correspond to the lowest median positive peak currents (less than 10 kA in Fig. 9), consistent with the hypothesis of contamination by intracloud flashes with small peak currents.
c. Median peak currents
The variation of median peak current in the NLDN as function of month (Fig. 7) shows a significant difference between negative and positive flashes. Negative flashes have a value of approximately 20 kA from January through November, jumping to 24 kA in December. The variation through the year, including the jump in December, is repeated in each of the three years. It is unknown why the negative peak current would increase by 20% in December.
The geographical changes in the median peak currents in Figs. 8 and 9 show important variations. The bull’s-eye in the western Carolinas is easily explained by the proximity of sensors in the western Carolinas (Fig. 11). Other geographic variations appear to be real. Median negative peak currents appear to be lower over the mountains in the West and over the Appalachian Mountains in western Virginia, confirming the first reports of this measurement for the Appalachians (Orville 1990). Median positive peak currents (Fig. 9), however, do not appear to show this relationship to topography. If the charge center altitude were lower in negative lightning than positive lightning, and if the charge were stored uniformly along the leader channel, then the charge lowered to ground and the corresponding peak current would be less. Positive lightning, however, originating from higher altitudes, would be expected to show less of an effect.
We note that Fig. 9 shows positive median peak currents that are less than 10 kA in the western Carolinas, in Kentucky and Tennessee, and in Louisiana and central Florida. These are suspect. Cummins et al. (1998) caution that “the subset of small positive discharges with peak currents less than 10 kA be regarded as cloud discharges unless they are verified to be cloud-to-ground.” We agree with this caution, but take no action to eliminate these flashes in our analyses. Further research is needed on this issue.
Negative flashes appear to have higher median peak currents along all coasts of the continental United States (Fig. 8) and in parts of the interior, centered on the western Pennsylvania–West Virginia border.
d. Multiplicity
The systematic geographic variation in the mean negative multiplicity in Fig. 10, from the northwest to the southeast and from Maine to Florida, may be related to the average size of the cumulonimbus cloud. The increase in the multiplicity is particularly noticeable along the East Coast from Maine to Florida. Note that multiplicity values of 1–1.5 in Maine increase almost monotonically along the coast to over 3.0 in Florida. A comparison of expected cumulonimbus depth in Maine versus Florida by Orville (1990) showed that the cloud depth might vary from 15.2 km in Florida to 11.5 km in Maine, or a ratio of 1.3. An increase of 30% is far short of the observed factor of two increase in multiplicity, but an increase in the horizontal extent of the typical cumulonimbus in Florida may, in addition, account for the increased multiplicity. Krehbiel et al. (1979) found that subsequent strokes extend from new regions, predominately in the horizontal direction. In short, larger cloud volume may account for the observed increase in multiplicity.
5. Conclusions
Cloud-to-ground lightning data for the years 1995–97 have been analyzed for the geographical distribution of total flashes, positive flashes, percentage of positive flashes, negative and positive first stroke peak currents, and negative flash multiplicity. In summary, the results are as follows.
-
The total measured flash counts were 22.7 million (1995), 26.2 million (1996), and 26.9 million (1997).
-
The areas of maximum measured flash density in 1995 were in Florida, Louisiana, and the Illinois–Indiana border (9–11 flashes per square kilometer). In 1996, the maximum flash densities exceeded 11 flashes per square kilometer and were found in Florida, Louisiana, eastern Texas, and northern Kentucky. In 1997, the maximum flash densities exceeding 11 flashes per square kilometer were in Louisiana and most of Florida.
-
Positive flash densities exceed 1.1 flashes per square kilometer in Florida, Louisiana, and a large area overlapping the states of Tennessee, Mississippi, and Kentucky.
-
The annual mean percentages of lightning flashes that lowered positive charge were 9.3% (1995), 10.2% (1996), and 10.1% (1997). All values were obtained after the NLDN upgrade in late 1994 (Cummins et al. 1998).
-
The monthly percentage of positive lightning flashes ranged from a low of 6.5% in July 1995 to a high of 24.5% in January 1996.
-
The percent positive flashes is >25% from the Canadian border as far south as Kansas and in Maine and along the entire West Coast.
-
The median negative peak current remains approximately constant from January through November at 20 kA, jumping to 24 kA in December. The median positive peak current ranges from 25 kA (February) to 15 kA (July) to 24 kA (December).
-
Median negative peak currents are uniformly high along all continental coastal areas, typically exceeding 26 kA. Mountainous regions appear to have lower median negative peak currents, on the order of 16 kA.
-
Median positive peak currents exceed 40 kA in the upper Midwest, but are less than 10 kA in Louisiana and Florida.
-
The mean negative multiplicity appears to increase with decreasing latitude in the eastern half of the United States.
We are completing the 10th year (1998) of the NLDN operation. Analyses underway will produce a composite year of lightning data consisting of 10-yr summaries of each month for flash density, percentage of positive lightning, peak currents, and multiplicity. We suspect, however, that an understanding of the variations of the cloud-to-ground lightning characteristics will come from field programs followed by detailed studies. An example of this is the Mesoscale Electrification and Polarization Radar Study (Schuur 1998) and detailed studies similar to the ones published by Stolzenburg (1994) and MacGorman and Morgenstern (1998).
Acknowledgments
The lightning data were obtained from the Global Atmospherics, Incorporated, Tucson, Arizona. The interest and assistance of Ken Cummins in this project are greatly appreciated. We thank Barbara Orville for her editorial assistance. Data handling at Texas A&M University is under the direction of Jerry Guynes and we thank him for his assistance. This research is part of a lightning program supported by the National Science Foundation (ATM-9806189) and the National Oceanic and Atmospheric Administration (Cooperative Agreement NA87WA0063).
REFERENCES
Cummins, K. L., M. J. Murphy, E. A. Bardo, W. L. Hiscox, R. B. Pyle, and A. E. Pifer, 1998: A combined TOA/MDF technology upgrade of the U.S. National Lightning Detection Network. J. Geophys. Res.,103 (D8), 9035–9044.
Fosdick, E. K., and A. I. Watson, 1995, Cloud-to-ground lightning patterns in New Mexico during the summer. Natl. Wea. Dig.,19 (4), 17–24.
Idone, V. P., D. A. Davis, P. K. Moore, Y. Wang, R. W. Henderson, M. Ries, and P. F. Jamason, 1998a: Performance evaluation of the U.S. National Lightning Detection Network in eastern New York. Part I: Detection efficiency. J. Geophys. Res.,103 (D8), 9045–9055.
——, ——, ——, ——, ——, ——, and ——, 1998b: Performance evaluation of the U.S. National Lightning Detection Network in eastern New York. Part II: Location accuracy. J. Geophys. Res.,103 (D8), 9057–9069.
Kriebiel, P. R., M. Brook, and R. A. McCrory, 1979: An analysis of the charge structure of lightning discharges to ground. J. Geophys. Res.,84, 2432–2456.
López, R. E., and R. L. Holle, 1985: Diurnal and spatial variability of lightning activity in northeastern Colorado during the summer. NOAA Tech. Memo. ERL ESG-14, National Oceanic and Atmospheric Administration, Boulder, CO, 35 pp.
——, and ——, 1987: Distribution of summertime lightning as a function of low-level wind flow in central Florida. NOAA Tech. Memo. ERL ESG-28, National Oceanic and Atmospheric Administration, Boulder, CO, 43 pp.
Lyons, W. A., 1994: Low-light video observations of frequent luminous structures in the stratosphere above thunderstorms. Mon. Wea. Rev.,122, 1940–1946.
——, 1996: Sprite observations above the U.S. High Plains in relation to their parent thunderstorm systems. J. Geophys. Res.,101 (D23), 29 641–29 652.
MacGorman, D. R., and C. D. Morgenstern, 1998: Some characteristics of cloud-to-ground lightning in mesoscale convective systems. J. Geophys. Res.,103 (D12), 14 011–14 023.
Orville, R. E., 1990: Lightning return stroke peak current variation as a function of latitude. Nature,342, 149–151.
——, 1991: Lightning ground flash density in the contiguous United States—1989. Mon. Wea. Rev.,119, 573–577.
——, 1994: Cloud-to-ground lightning flash characteristics in the contiguous United States: 1989–1991. J. Geophys. Res.,99 (D5), 10 833–10 841.
——, and A. C. Silver: 1997: Lightning ground flash density in the contiguous United States: 1992–95. Mon. Wea. Rev.,125, 631–638.
Schuur, T. J., 1998: MCS Electrification and Polarimetric Radar Study. NSSL, 42 pp. [Available from National Severe Storms Laboratory, 1313 Halley Circle, Norman, OK 73069.].
Stolzenburg, M., 1994: Observations of high ground flash densities of positive lightning in summertime thunderstorms. Mon. Wea. Rev.,122, 1740–1750.
Wacker, R. S., and R. E. Orville, 1996: Changes in measured lightning flash count and return stroke peak current since the 1994 National Lightning Detection Network upgrade. Eos, Trans. Amer. Geophys. Union, 77, F76.
Watson, A. I., and R. L. Holle, 1996: An eight-year lightning climatology of the southeast United States prepared for the 1996 Summer Olympics. Bull. Amer. Meteor. Soc.,77, 883–890.
——, R. E. López, and R. L. Holle, 1994: Diurnal cloud-to-ground lightning patterns in Arizona during the southwest monsoon. Mon. Wea. Rev,122, 1716–1725.
Winckler, J. R., R. C. Franz, and R. J. Nemzek, 1993: Fast low-level light pulses from the night sky observed with the SKYFLASH program, J. Geophys. Res.,98 (D5), 8775–8783.
The total number of cloud-to-ground lightning flashes (negative and positive) is plotted as a function of month for the years 1995–97. The maximum is in July for each of the three years.
Citation: Monthly Weather Review 127, 11; 10.1175/1520-0493(1999)127<2693:LGFMOT>2.0.CO;2
Lightning ground flash density for the years 1995–97 is shown in the first three panels followed by the 3-yr mean.
Citation: Monthly Weather Review 127, 11; 10.1175/1520-0493(1999)127<2693:LGFMOT>2.0.CO;2
The number of positive cloud-to-ground lightning flashes is shown as a function of month in each of the three years, 1995–97. Note that the positive flashes peak in July in 1996 and 1997, but in May in 1995.
Citation: Monthly Weather Review 127, 11; 10.1175/1520-0493(1999)127<2693:LGFMOT>2.0.CO;2
The annual positive flash density geographical distribution is shown in the first three panels followed by the mean of the three years, 1995–97, in the fourth panel.
Citation: Monthly Weather Review 127, 11; 10.1175/1520-0493(1999)127<2693:LGFMOT>2.0.CO;2
The percentage of positive lightning is plotted as a function of month for each of the years, 1995–97. The minimum occurs in the Jul–Aug period and the maximum occurs in Jan–Feb.
Citation: Monthly Weather Review 127, 11; 10.1175/1520-0493(1999)127<2693:LGFMOT>2.0.CO;2
The percentage of flashes lowering positive charge to earth is plotted for the years 1995, 1996, and 1997 in the first three panels. The fourth panel shows the geographical mean percent positive for the three years.
Citation: Monthly Weather Review 127, 11; 10.1175/1520-0493(1999)127<2693:LGFMOT>2.0.CO;2
The median monthly peak current is plotted for the period 1995–97. The monthly values for the individual years are approximately the same, so one curve for each polarity suffices to show the variation. Note that the Nov and Dec values for both negative and positive flashes are approximately the same.
Citation: Monthly Weather Review 127, 11; 10.1175/1520-0493(1999)127<2693:LGFMOT>2.0.CO;2
The median negative peak current is shown for three individual years followed by a plot of the median peak current for the period 1995–97. White areas over land occur, for example in California, where there are less than two flashes in a 0.2° grid box.
Citation: Monthly Weather Review 127, 11; 10.1175/1520-0493(1999)127<2693:LGFMOT>2.0.CO;2
Median positive peak currents are plotted geographically for the individual years 1995, 1996, and 1997. The median value for the three years is plotted in the fourth panel. White areas over land occur, e.g., along the West Coast, where there are less than two flashes in a 0.2° grid box.
Citation: Monthly Weather Review 127, 11; 10.1175/1520-0493(1999)127<2693:LGFMOT>2.0.CO;2
The mean negative multiplicity, i.e., the number of strokes per flash, is plotted for each for the years, 1995, 1996, and 1997. The mean multiplicity for the three years, 1995–97, is plotted in the fourth panel.
Citation: Monthly Weather Review 127, 11; 10.1175/1520-0493(1999)127<2693:LGFMOT>2.0.CO;2
Lightning sensor locations are shown in the National Lightning Detection Network. The locations of the IMPACT direction-finding and time-of-arrival sensors are plotted with a filled circle. The location of the Lightning Position and Tracking System (LPATS) time-of-arrival sensors are plotted with a filled triangle. The inset shows a cluster of IMPACT sensors in the western Carolinas.
Citation: Monthly Weather Review 127, 11; 10.1175/1520-0493(1999)127<2693:LGFMOT>2.0.CO;2
The distance between the nearest pair of sensors with a direction finder capability (IMPACT sensor) is calculated and contoured. The closest IMPACT sensors are located in the Carolinas and the farthest apart are in the western states. A lightning flash must be detected by at least one IMPACT sensor to be recorded.
Citation: Monthly Weather Review 127, 11; 10.1175/1520-0493(1999)127<2693:LGFMOT>2.0.CO;2
Cloud-to-ground flashes for 1995–97.
Monthly and annual percentage positive lightning for 1995–97.
