i V LIBRARY Michigan State 1L University PLACE IN RETURN BOX to remove this checkout from your record. TO AVOID FINES return on or before data due. [DATE DUE DATE DUE DATE DUE I MSU Is An Affirmative Action/Equal Opportunity Institution cmmui i n T’:\F‘A -i‘J ‘ JRP TEMPORAL AND SPATIAL VARIATIONS OF CLOUD-TO-GROUND LIGHTNING IN THE GREAT LAKES REGION BY Claudia K. Gunreben A THESIS Submitted to Michigan State University In partial fulfillment of the requirements for the degree of MASTER OF ARTS Program in American Studies 1992 ABSTRACT TEMPORAL AND SPATIAL VARIATIONS OF CLOUD-TO-GROUND LIGHTNING IN THE GREAT LAKES REGION BY Claudia K. Gunreben An initial analysis of lightning data for the southern Great Lakes Area was conducted for the period of 1 January, 1989 through 25 October, 1990. Emphasis was placed on (1) the temporal variations in characteristics of cloud-to-ground flashes with respect to frequency, strength, polarity, and number of return-strokes and (2) the spatial variations in the timing of lightning activity across the study region. An inverse relationship between flash frequency and the percent- age of positive strikes was observed both on a nmmthly and diurnal scale. A higher flash total in 1990 coincided with a higher percentage of nighttime strikes. Considerable monthly and diurnal differences in lightning activity were observed for the eastern and western portions of the study area. A southwest-northeast axis of highest lightning activity could frequently be observed. Little correspondence between areas of high total flash receipts and number of days with light— ning activity was evident. Copyright by CLAUDIA KUNIGUNDA GUNREBEN 1992 TABLE OF CONTENTS LIST OF TABLES . . . . . . . . . . . . . . . . . . . . . vi LIST OF FIGURES . . . . . . . . . . . . . . . . . . . . viii Chapter 1. INTRODUCTION . . . . . . . . . . . . . . . . . . . 1 Background of the Study Lightning Detection Networks Diurnal and Seasonal Characteristics of Thunderstorms The Diurnal Variations of Precipitation Electrical Properties of Thunderstorms The Data Set Objectives of the Study Part I: TEMPORAL ANALYSIS 2. FREQUENCY OF LIGHTNING EVENTS . . . . . . . . . . 22 Annual Distribution Time of Day of Lightning Activity Summary 3. RETURN STROKES AND AMPLITUDE . . . . . . . . . . . 29 Methodology Return Strokes Associated with Positive and Negative Flashes iv V Monthly Variations in the Maximum Number of Return Strokes Diurnal Variations in the Maximum Number of Return Strokes Amplitude of Positive and Negative Strikes Monthly Variations in the Maximum Amplitudes for Positive and Negative Strikes Diurnal Variations in the Maximum Amplitudes for Positive and Negative Strikes Summary 4. POLARITY . . . . . . . . . . . . . . . . . . . . . 67 Monthly and Diurnal Variations Polarity in Relation to Return Strokes and Amplitude Summary Part II: SPATIAL ANALYSIS 5. SPATIAL ANALYSIS OF LIGHTNING ACTIVITY: PROCEDURES AND METHODOLOGY . . . . . . . . . . 75 Objectives The Analysis Grid Calculation of Grid Cell Frequencies and Mapping Temporal Component of the Spatial Analysis 6. SPATIAL ANALYSIS OF LIGHTNING ACTIVITY: RESULTS . . . . . . . . . . . . . . . . . . . 80 Monthly Variations in Daytime and Nighttime Lightning Activity April 1989 May 1989 June 1989 July 1989 August 1989 \‘J (7 LIST q I Arsenic vi September 1989 May 1990 June 1990 July 1990 August 1990 September 1990 Summary 7. CONCLUSION Methodology Major Findings of Temporal Analysis Major Findings of Spatial Analysis LIST OF REFERENCES Appendices A. FORTRAN Program Grid. B. FORTRAN Program Mmultmax C. FORTRAN Program Days 140 150 153 155 157 LL) -J 10. 12. Table 10. 11. 12. LIST OF TABLES Annual Distribution of Lightning Events and Lightning Days Nighttime Events as Percentages of the Monthly CG Totals . Percentages of -CGs and +CGs with only one Return Stroke Mean Number of Return Strokes for Positive and Negative CG Flashes Maximum Number of Return Strokes for +CGs and -CGs Five-minute Maxima of Return Strokes, 1989 and 1990 Mean Amplitude of +CGs and -CGs Five-minute Maxima of Amplitude for Positive and Negative Strikes, 1989 and 1990 Percentages of Five-minute Maxima of Amplitude for +CGs and -CGs in each category, 1989 and 1990 Percentages of Five-minute Periods with Maximum Amplitude ZlOOkA Relative to the Number of Five-minute Intervals with Recorded Lightning Activity for Positive and Negative Strikes Total Number of Events and Percentages of +CGs 1989 and 1990 Nighttime/Daytime Total Number of Events and Percentage of Positive Flashes vii Page 24 26 32 33 34 37 53 57 58 6O 67 68 13. viii Nighttime (0000-1200 GMT) and Daytime (1200-2400 GMT) Occurrence of Positive Flashes 71 Figure *4 o [\J 3a. 4a. 4b. 4C. 5a. 5c 6a 5b Figure 1. 2. 3a. 3b. 3c. 4a. 4b. 40. 5a. 5b. 5c. 6a. 6b. LIST OF FIGURES Area of Coverage of the SUNYA data set Total Number of Nighttime and Daytime Flashes Five-minute Maxima of Return Strokes, January 1989 and 1990 Five-minute Maxima of Amplitudes for Positive CG flashes, January 1989 and 1990 Five-minute Maxima of Amplitudes for Negative CG flashes, January 1989 and 1990 Five-minute Maxima of Return Strokes, February 1989 and 1990 Five-minute Maxima of Amplitudes for Positive CG flashes, February 1989 and 1990 Five-minute Maxima of Amplitudes for Negative CG flashes, February 1989 and 1990 Five-minute Maxima of Return Strokes, March 1989 and 1990 Five-minute Maxima of Amplitudes for Positive CG flashes, March 1989 and 1990 Five-minute Maxima of Amplitudes for Negative CG flashes, March 1989 and 1990 Five-minute Maxima of Return Strokes, April 1989 and 1990 Five-minute Maxima of Amplitudes for Positive CG flashes, April 1989 and 1990 ix Page 18 22 39 39 39 4O 4O 4O 41 41 41 42 42 60. 7b. 7c. 8a. .3 n6 9C 106 6c. 7a. 7b. 7c. 8a. 8b. 8c. 9a. 9b. 9c. 10a. 10b. 10c. 11a. 11b. 11c. X Five—minute Maxima of Amplitudes for Negative CG flashes, April 1989 and 1990 Five-minute Maxima of Return Strokes, May 1989 and 1990 Five-minute Maxima of Amplitudes for Positive CG flashes, May 1989 and 1990 Five-minute Maxima of Amplitudes for Negative CG flashes, May 1989 and 1990 Five-minute Maxima of Return Strokes, June 1989 and 1990 Five-minute Maxima CG flashes, June Five-minute Maxima CG flashes, June Five-minute Maxima 1989 and 1990 Five-minute Maxima of Amplitudes for Positive 1989 and 1990 of Amplitudes for Negative 1989 and 1990 . of Return Strokes, July of Amplitudes for Positive CG flashes, July 1989 and 1990 Five-minute Maxima of Amplitudes for Negative CG flashes, July 1989 and 1990 Five-minute Maxima of Return Strokes, August 1989 and 1990 Five-minute Maxima of Amplitudes for Positive CG flashes, August 1989 and 1990 Five-minute Maxima of Amplitudes for Negative CG flashes, August 1989 and 1990 . Five-minute Maxima of Return Strokes, September 1989 and 1990 Five-minute Maxima of Amplitudes for Positive CG flashes, September 1989 and 1990 Five-minute Maxima of Amplitudes for Negative CG flashes, September 1989 and 1990 42 44 44 44 45 45 45 46 46 46 48 48 48 50 50 50 12a. 12b. 12c. 13a. 13b. 13c. 14. 15. 16a 16b a 11" 17b 19a .hy P3 1 is 12a. 12b. 12c. 13a. 13b. 13c. 14. 15. 16a. 16b. 17a. 17b. 18a. 18b. 19a. 19b. xi Five-minute Maxima of Return Strokes, October 1989 and 1990 Five-minute Maxima of Amplitudes for Positive CG flashes, October 1989 and 1990 Five-minute Maxima of Amplitudes for Negative CG flashes, October 1989 and 1990 Five-minute Maxima of Return Strokes, November 1989 and 1990 Five-minute Maxima of Amplitudes for Positive CG flashes, November 1989 and 1990 Five-minute Maxima of Amplitudes for Negative CG flashes, November 1989 and 1990 Monthly Distribution of Strikes inside and outside the 32-by-19 MDR grid Total Number of Cloud—to-Ground Flashes for Study Period Nighttime (0000-1200 GMT) Lightning Strikes, April 1989 . . . . . . . . . . . Number of Days with Nighttime Lightning Activity, April 1989 . Daytime (1200-2400 GMT) Lightning Strikes, April 1989 . . . . . . . . . . . Number of Days with Daytime Lightning Activity, April 1989 . . . Hourly Flash Frequencies for the Western Portion of the Study Area, April 1989 Hourly Flash Frequencies for the Eastern Portion of the Study Area, April 1989 Nighttime (0000-1200 GMT) Lightning Strikes, May 1989 . . . . . . . . . . . . . Number of Days with Nighttime Lightning Activity, May 1989 . . . . . . 51 51 51 52 52 52 78 80 83 83 84 84 86 86 88 88 20a. 20b. 21a. 21b. 22b. 23a. 23b. 24a. 2% 27a 27h xii 20a. Daytime (1200-2400 GMT) Lightning Strikes, May 1989 . . . . . . . . . . . . . . . . . . . 90 20b. Number of Days with Daytime Lightning Activity, May 1989 . . . . . . . . . . . . . 90 21a. Hourly Flash Frequencies for the Western Portion of the Study Area, May 1989 . . . . . 92 21b. Hourly Flash Frequencies for the Eastern Portion of the Study Area, May 1989 . . . . . 92 22a. Nighttime (0000-1200 GMT) Lightning Strikes, June 1989 . . . . . . . . . . . . . . . . . . 94 22b. Number of Days with Nighttime Lightning Activity, June 1989 . . . . . . . . . . . . . 94 23a. Daytime (1200-2400 GMT) Lightning Strikes, June 1989 . . . . . . . . . . . . . . . . . . 96 23b. Number of Days with Daytime Lightning Activity, June 1989 . . . . . . . . . . . . . 96 24a. Hourly Flash Frequencies for the Western Portion of the Study Area, June 1989 . . . . 97 24b. Hourly Flash Frequencies for the Eastern Portion of the Study Area, June 1989 . . . . 97 25a. Nighttime (0000-1200 GMT) Lightning Strikes , July 1989 . . . . . . . . . . . . . . . . . . 99 25b. Number of Days with Nighttime Lightning Activity, July 1989 . . . . . . . . . . . . . 99 26a. Daytime (1200-2400 GMT) Lightning Strikes, July 1989 . . . . . . . . . . . . . . . . . . 100 26b. Number of Days with Daytime Lightning Activity, July 1989 . . . . . . . . . . . . . 100 27a. Hourly Flash Frequencies for the Western Portion of the Study Area, July 1989 . . . . 102 27b. Hourly Flash Frequencies for the Eastern Portion of the Study Area, July 1989 . . . . 102 28a. 28b. 29a. K) kg) CT 30a. 30b. 31b. 32a (A) (J I H) (.0 Ln (7" 28a. 28b. 29a. 29b. 30a. 30b. 31a. 31b. 32a. 32b. 33a. 33b. 34a. 34b. 35a. 35b. xiii Nighttime (0000-1200 GMT) Lightning Strikes, August 1989 . . . . . . . . . . . Number of Days with Nighttime Lightning Activity, August 1989 . . . Daytime (1200-2400 GMT) Lightning Strikes, August 1989 . . . . . . . . Number of Days with Daytime Lightning Activity, August 1989 . . Hourly Flash Frequencies for the Western Portion of the Study Area, August 1989 Hourly Flash Frequencies for the Eastern Portion of the Study Area, August 1989 Nighttime (0000-1200 GMT) Lightning Strikes, September 1989 . . . . . . . . . Number of Days with Nighttime Lightning Activity, September 1989 . . Daytime (1200-2400 GMT) Lightning Strikes, September 1989 . . . . . . . . . . Number of Days with Daytime Lightning Activity, September 1989 . Hourly Flash Frequencies for the Western Portion of the Study Area, September 1989 Hourly Flash Frequencies for the Eastern Portion of the Study Area, September 1989 Nighttime (0000-1200 GMT) Lightning Strikes, May 1990 . . . . . . . . . . . Number of Days with Nighttime Lightning Activity, May 1990 . . . . Daytime (1200-2400 GMT) Lightning Strikes, May 1990 . . . . . . . . . . . Number of Days with Daytime Lightning Activity, May 1990 . . . . 104 104 105 105 107 107 109 109 111 111 113 113 115 115 116 116 36a. 36b. 37a. 37b. 38a. 38b. 39a. 39b. 40a. 40b. 41a 41b. 42a. 42b. 43a, 43b, 36a. 36b. 37a. 37b. 38a. 38b. 39a. 39b. 40a. 40b. 41a 41b. 42a. 42b. 43a. 43b. xiv Hourly Flash Frequencies for the Western Portion of the Study Area, May 1990 Hourly Flash Frequencies for the Eastern Portion of the Study Area, May 1990 Nighttime (0000- -1200 GMT) Lightning Strikes, June 1990 Number of Days with Nighttime Lightning Activity, June 1990 Daytime (1200-2400 GMT) Lightning Strikes, June 1990 . . . . . . . . . . . Number of Days with Daytime Lightning Activity, June 1990 . . . . Hourly Flash Frequencies for the Western Portion of the Study Area, June 1990 Hourly Flash Frequencies for the Eastern Portion of the Study Area, June 1990 Nighttime (0000-1200 GMT) Lightning Strikes, July 1990 Number of Days with Nighttime Lightning Activity, July 1990 . . . . . Daytime (1200- -2400 GMT) Lightning Strikes, July 1990 . . . . . . . . . . Number of Days with Daytime Lightning Activity, July 1990 . . . . Hourly Flash Frequencies for the Western Portion of the Study Area, July 1990 Hourly Flash Frequencies for the Eastern Portion of the Study Area, July 1990 Nighttime (0000-1200 GMT) Lightning Strikes, August 1990 . . . . . . . . . . . Number of Days with Nighttime Lightning Activity August, 1990 . . . 118 118 120 120 121 121 123 123 125 125 126 126 128 128 130 130 44a. 44b. 45a. 46a. 46b. 47b. 48a. 48b. 44a. 44b. 45a. 45b. 46a. 46b. 47a. 47b. 48a. 48b. XV Daytime (1200-2400 GMT) Lightning Strikes, August 1990 . . . . . . . . . . . Number of Days with Daytime Lightning Activity, August 1990 Hourly Flash Frequencies for the Western Portion of the Study Area, August 1990 Hourly Flash Frequencies for the Eastern Portion of the Study Area, August 1990 Nighttime (0000—1200 GMT) Lightning Strikes, September 1990 . . Number of Days with Nighttime Lightning Activity, September 1990 Daytime (1200-2400 GMT) Lightning Strikes, September 1990 . . . . Number of Days with Daytime Lightning Activity, September 1990 Hourly Flash Frequencies for the Western Portion of the Study Area, September 1990 Hourly Flash Frequencies for the Eastern Portion of the Study Area, September 1990 131 131 133 133 135 135 136 138 138 OH ‘ 2818:! nd #02 CHAPTER I INTRODUCTION Wild); Lightning strikes have been termed "the nation's number one weather killer." According to Mogil et a1. (1977), lightning was responsible for the deaths of almost 7,500 Americans between 1939 and 1976. Over 20,000 lightning- related injuries occurred during the same time period. Also, lightning strikes are the cause of approximately 10,000 wild- land fires each year (Krider et al. 1980). The advent of networks of magnetic lightning direction finders in the early 19808 has triggered a series of research projects aimed at a better understanding of lightning-producing storms. Light- ning observations have been complemented by data obtained by radar, airborne balloons, field mills, and other field observations to acquire information about rainfall amounts, ground-flash rates, peak currents, numbers of return strokes, and orographic and synoptic controls of thunderstorms. The long-term goal of these studies is to improve the predictability of thunderstorms and their associated hazards. for c Coast Elect Maine al. Netwc Natic In 1983, a network of magnetic direction finders (DFs) for observing cloud-to-ground lightning (CG) along the East Coast was established by SUNY-Albany with support from the Electrical Power Institute. Ten DFs covered an area from Maine to North Carolina and as far west as Ohio (Orville et al. 1983). The area of coverage of the Lightning Detection Network was extended farther west in 1989 when the commercial National Lightning Detection Network was created from three regional networks. The eastern network run by the State University of New York at Albany (SUNYA) was joined by the western network managed by the Bureau of Land Management (BLM) and a midwestern branch operated by the National Severe Storms Laboratory (NSSL). The National Network now covers three million square miles and employs approximately 115 mag- netic direction finders (Orville et al. 1990)“ The DFs have a detection rate of 80 percent within a nominal range of 400 km. The time, angle, signal amplitude, polarity of the first return-stroke, and number of return strokes of each detected cloud-to-ground flash are stored in digital form. After two or more DFs detect the same strike, the location of the ground-flash is interpolated and the information about the lightning strike is archived at a central facility (Orville et al. 1990). The principle of wideband magnetic direction finders has been previously described in detail (e.g. Krider et al. 1980). ation tifie State of ma EIOUE In their fairly comprehensive study of the diurnal vari- ations of thunderstorms, Easterling and Robinson (1985) iden- tified nine thunderstorm regions for the conterminous United States. They based their classification upon both the "time of maximum storm occurrence and the concentration of activity around this time." They suggested that similar diurnal char- acteristics of thunderstorms are caused by "common causal mechanisms operating in each area." Throughout the nation they detected a clear trend for regions and seasons with an afternoon maximum in thunderstorm frequency to have a high amplitude diurnal frequency curve, whereas regions with noc- turnal maxima were seen as having a tendency for lower ampli- tude hourly frequency distributions. Of interest for this study are only four of the identified thunderstorm regions, the Central United States, the Great Lakes Region, the Northeast, and the Southeast. In the Central United States, Easterling and Robinson identified an area of mainly noctur- nal storms with a low normalized amplitude in all seasons ex— cept winter. The Great Lakes Region sets itself apart from the Central States through a significant number of winter storms. It is characterized by nocturnal low amplitude max— ima in all seasons. In the Northeast, winter storms are less frequent and the timing of thunderstorms shifts to the after- noon hours. The amplitude of the afternoon maximum increases to medium. The Southeast has afternoon storms in all seasons except winter, when nocturnal storms are dominant. In this region, the amplitude of the frequency distribution is lowest during winter; it reaches a maximum during summer. Analysis of satellite and lightning observations have further clarified the diurnal and seasonal variations of thunderstorm activity across the conterminous United States. In agreement with Easterling and Robinson's (1985) study, Orville's (1981) analysis of satellite imagery for September through November 1977 revealed a midnight maximum in thunder- storm frequency over the Midwest for autumn with a general decrease in lightning activity from September to November. Speheger et al. (1990) examined the climatology of se- vere thunderstorm events in Indiana during a 31-year period (1959-1989). Speheger’s findings for Indiana are in agree- ment with the findings of Easterling and Robinson (1985). Except for the extreme northern portion of the state, Indiana falls within the Southeastern region. as classified. by Easterling and Robinson. Typical of the Southeastern region, the onset of lightning activity in Indiana most commonly oc- curred between 2 and 7 p.m. Eastern Standard Time (1900-2400 GMT), and lightning activity was most frequent in June. Several studies have contrasted the duration, spatial patterns, and seasonality of thunderstorms for different parts of the US. with those for Florida, the state with the most frequent thunderstorm activity. Maier and Krider (1982), using lightning observations, noted considerable dif- ferences between the characteristics of three severe thunder- storms that occurred in north Texas and Oklahoma during April and N Flori Great locat trave eas." and t and May 1979 and 268 "nonsevere air-mass thunderstorms" in Florida for the summer of 1978. The severe storms of the Great Plains were found to possess "well-defined lightning location clusters which would exist for long periods of time, traveling at moderate to high rates, thus affecting large ar— eas." The average duration ranged from two to eight hours and the affected area for two of the storms was 1800 km2 and 5000 kmz, respectively. The mean rates of cloud-to-ground lightning ranged from about two to six strikes per minute with peak rates approaching twenty ground flashes per minute. The Florida air-mass storms, on the other hand, were de— scribed as going through their life cycles quickly, affecting smaller areas and possessing lower flash rates. Storm dura- tion rarely exceeded one to two hours with mean rates of cloud-to-ground (CG) lightning of about one strike per minute and peak rates of twelve strikes per minute. The total area affected was found to average about 450 kmz. On a typical summertime afternoon, however, ten to fifty airmass storms would form. Thus, the total storm duration, area, and the number of CG-flashes would be similar to that of severe storms in the Great Plains. Piepgrass et al.'s (1982) results from an analysis of lightning activity in a 625 km2 area of central Florida dur- ing the summers of 1976-1980 are in general agreement with the findings of Maier and Krider (1982). Seventy-nine storms with ten or more strikes produced a total of 27,494 dis- charges during the examined summers. Storm duration averaged tivi 2000 ligh‘ of M 107 minutes, and the average flash rate was 2.4 strikes per minute with maximum short-term discharge rates of 30.6 strikes per minute. Summer lightning in Florida was also the focus of Maier et al.'s (1984) study. They found the peak in lightning ac- tivity in coastal areas of south Florida to occur between 2000 and 2100 GMT. Inn addition, the diurnal variation in lightning activity was smaller over the Atlantic and the Gulf of Mexico than over land. Diurnal and spatial variability of lightning activity in Colorado and Florida was the focus of Lopez and Holle's study (1986). Topography and resulting diurnal circulations were found to exert strong influences on the time and place of lightning ground strike occurrences. For Colorado, they found the daily flash rate to be highly variable with a maxi- mum around the first part of August. The diurnal cycle re- vealed a pronounced maximum between 4 and 5 p.m. MST (2300 and 2400 GMT) and a minimum between 7 and 8 a.m. MST (1400 and 1500 GMT). Orographic and thermal influences resulted in the frequent development of a cyclonic convergence-vorticity zone from the Denver area north-northeastward in the summer. The resulting circulations and confluence zones were found to have a "decided impact on the preferred formation and propagation patterns of the convective systems forming in the area." For Florida, the annual maximum of lightning frequency also fell in the summer months of June, July, and August. relat the s in thl found Compared to Maier et al.'s (1984) previous study, Lopez and Holle identified a maximum in the diurnal cycle of lightning frequency between 2 and 3 p.m. EST. Three preferred times for the occurrence of first flashes were identified: around sunrise, around noon, and after midnight. Lopez and Holle related these relative maxima to the "varying influences of the sea breeze, the land breeze, or nocturnal cooling." As in the case of Colorado, Florida's lightning distribution was found to reflect topographical characteristics of the region. The close association between rainfall and cloud-to- ground lightning characteristics has been noted by Goodman and MacGorman (1986) and Nielsen et al. (1990). Of special interest for this study is the occurrence of summertime noc- turnal maxima in lightning and convective activity for the central part of the United States and an afternoon maximum for the eastern parts of the country. These maxima and min- ima are also evident in the diurnal frequency distribution of hourly precipitation. Previous publications on the diurnal variation in pre- cipitation and thunderstorm frequency include a study by Wallace (1975) conducted for the conterminous United States. Wallace examined differences in the timing of frequency max- ima for various precipitation levels and for thunderstorms. Harmonic analysis of hourly frequencies revealed an earlier maximum in summertime thunderstorm activity as compared to summertime heavy precipitation (>0.25 cm h”) for the central part of the country. A difference of over six hours was noted for parts of Illinois, Indiana, and Ohio. These areas lie in tflua transition zone between regions experiencing a nocturnal maximum of precipitation and those experiencing a daytime maximum. Trace precipitation also showed a tendency to be out of phase with thunderstorm activity in the central United States, particularly in a triangular shaped area ex- tending from Oklahoma to Michigan to North Dakota. In this region, thunderstorm activity was found to possess a midnight maximum and trace precipitation a morning maximum. During the winter months, weak nocturnal maxima in convective activ— ity were evident for the temperate latitudes of both the cen- tral and eastern United States. Balling (1985) also examined the summertime nocturnal maximum of hourly precipitation for different precipitation levels. He found that 60 percent of all warm season precipi- tation 20.25 m h”1 in southern Nebraska, central Kansas, western Oklahoma and northern Texas occurred at night. The area of nighttime maxima was centered on southeastern Nebraska. In western Nebraska, Kansas, and Oklahoma an east- west gradient of about 1h per 100 km in the timing of maximum frequency was evident. Balling detected a much stronger modulation of the diurnal cycle for the 22.54 mm hfl events compared to the lighter precipitation amounts. Also, larger precipitation levels displayed a frequency maximum that oc- curred one to two hours earlier than the maximum for the noon 1 ing f; 20118 .3 :r (0 r1 lighter categories. In agreement with Wallace (1975), Winkler (1987) also identified a nocturnal regime of summertime very heavy pre— cipitation (>6 mm h’l) for the Central Plains and an after- noon maximum for the eastern United States. An area extend- ing from central Texas to Indiana was seen as the transition zone between the two rainfall regimes. In this transition region, very heavy precipitation is likely to occur through— out the day, although two slight maxima at approximately 0600 LST (1200 GMT) and 1500 LST (2100 GMT) can be identified. The time of summertime very heavy precipitation was found to generally correspond with the time of thunderstorm activity. As previously identified by Wallace (1975), an exception was found in the CBntral Plains where the maximum thunderstorm activity occurred before the time of maximum frequency of very heavy rainfall. The two authors slightly diverged on the time lag between maximum thunderstorm and maximum rainfall activity. Winkler stated a difference of two to three hours for very heavy rainfall amounts. In a later study, Winkler et al. (1988) examined the di- urnal characteristics of heavy hourly precipitation (>2.5 mm Iri) for all four seasons. Harmonic analysis revealed a noc- turnal maximum across much of the eastern and central United States in winter and spring that gave way to an afternoon maximum across the southern and eastern states in the summer. Toward autumn the area of nighttime maxima increased again. It remained smaller than in winter and spring, however. With sons 1 intens 10 increasing intensity, rainfall was confined to increasingly shorter periods of the day. At the same time, more intense precipitation occurred considerably later in the day than light precipitation events. The only exception to this rule was found in the central United States "where during all sea- sons precipitation occurs earlier in the evening as rainfall intensity increases." We The electrical properties of lightning-producing thun- derstorms have been found to vary with season and with par- ticular phases in the life cycle of a storm. Orville et al. (1987) examined the characteristics of lightning strikes in the northeastern United States for twelve continuous months, from June 1984 through May 1985. The median amplitude for 720,284 first return strokes lowering negative charge to ground was 30 kA. Only a small percentage of negative flashes had first return strokes with "peak currents esti- mated to exceed 100 kA." The distribution of the peak cur- rent values for the recorded 17,694 positive cloud-to-ground flashes around the median value of 45 kA was significantly broader than for the negative strikes. In addition, a few positive strikes had peak currents exceeding 200 kA. The me- dian strength of both negative and positive ground flashes was found to increase in the winter season. The average peak current in February, for example, was approximately 65 kA for negative strikes and 90 RA for positive flashes. An inverse number I ma X l :11 11 relationship between the rate of negative cloud-to-ground lightning and mean peak current strength has frequently been identified (personal comment by Orville (1987) in Stolzenburg, 1990). Orville et al. (1987) also found that flash polarity varied with season. The percentage of posi- tive strikes increased from less than 5 percent of the total number of strikes in the summer (May through September) to a maximum in February of slightly over 80 percent. In spring, the percentage decreased again to less than 10 percent in April. The number of return strokes associated with ground flashes also described an annual curve with increasing values in the warm months, although only signals for up to fourteen return strokes were processed by the direction finder micro- computer. Throughout the year, approximately 90 percent of the positive flashes had only one return stroke. The per- centage of negative flashes with one return stroke increased from approximately 40 percent in June to over 80 percent in January and decreased again to less than 50 percent in April. The different seasonal characteristics of lightning—pro- ducing storms were also at the heart of studies by Fuquay (1982), Orville et al. (1983) and Brook et al. (1982). During the summer thunderstorm seasons of 1965-67, Fuquay documented 75 ground flashes, an average of 3 percent of the total number of strikes, that lowered positive charge to ground over an area of about 2830 km2 in the northern Rockies, . Each of these positive ground flashes was found to have one single return stroke. However, not all storms posse occur reco: occur km r2 tive 12 possessed positive lighting strikes. On storm days with the occurrence of positive ground flashes, positive strikes were recorded throughout the storms with the highest percentage occurring during the final stages of the storm. Within a 30- km radius centered on Missoula, Montana, storms that lowered positive charge to the ground ranged from 72 to 286 minutes The positive flashes averaged 6 percent of all flashes. The strike density per season was found to approximate 0.01 posi- tive and 0.3 negative ground flashes per square kilometer. Orville et al.'s (1983) case study of an autumn storm with 11,000 ground flashes over southeastern New York and New England revealed a symmetrical increase and decrease in lightning frequency over the duration of the storm. While all strikes combined showed a parabolic frequency curve, the percentage of positive strikes increased toward the end of the storm to 37 percent of the total number of flashes. Averaged over the duration of the storm, the positive flashes amounted to 4 percent of the total number of strikes. Brook et al. (1982) focused on the winter season of 1977-78 in their study of winter thunderstorms along the Hokuriku coast. They found that "positive currents peak about one order of magnitude greater than negative currents" and that the magnitude of the current for the positive return strokes occurring during the examined winter storms was com- parable to large negative currents in summer storms. Also, the positive and negative charge centers within thunderclouds were found. to ioccupy' different heights. The center' of posit: "while This verti« grounl nific "may Unite grour haivre '0 0 (I! (D O (f (D 13 positive charge was approximately 6.3 km above the surface, "while the highest negative charge was measured at 5.1 km." This "dipole" of different charge centers was attributed to vertical wind shear. Brook et al. speculated that positive ground flashes in summer storms might be accompanied by "sig- nificant shear in the cloud layer." They concluded that one "may indeed expect that the severe storms in the midwestern United States would exhibit a higher number of positive strokes than do the smaller orographic storms." The different characteristics of positive cloud-to- ground strikes (+CGs) as compared to negative strikes (-CGs) have been commented upon by several additional authors. Rust et al. (1981) reported in their study on severe storms that most of the recorded positive strikes only had one return stroke. The average duration of positive strikes was about twice as long as that of negative flashes. Rust et al. attributed the longer duration of +CG to the "time taken for discharge processes prior to the first return stroke and for the apparent continuing’ current afterwards." Positive strikes were "observed to emanate from several regions of severe storms: high on the back of the main storm tower, through the wall cloud, and from the downshear anvil." A spectacular lightning event was recorded by Idone et al. (1984). The authors were called to the scene where a positive cloud-to-ground flash had hit a residential home. Recordings from the lightning detection network revealed that the recorded +CG had been the only ground flash in a storm that posse ning (l98i being A tht 14 that produced only a few total flashes. The positive strike possessed an estimated peak current of 70 kA. The processes generating positive cloud-to—ground light- ning have not yet been completely clarified. Takagi et al. (1986) supported the hypothesis of a tilted vertical dipole being a prerequisite for positive cloud-to-ground lightning. A thundercloud is generally assumed to be composed of a ver— tical dipole with net positive charge located above net nega- tive charge. Initiating positive streamers progress downward into the negatively charged regions of the cloud. In the case of a tilted dipole with horizontal displacement between the positive and negative charge centers, a number of posi- tive flashes can reach the ground instead of the negatively charged regions within the cloud. Takagi et al. used an elec- tric dipole model to calculate the necessary horizontal dis- placement between the charge centers for the occurrence of +CG flashes. Winter thunderstorms in Norway and spring thun- derstorms in Japan were examined to determine the height of the —10°C isotherm which is believed to be near the center of charge separation. The temperature difference between the positive and negative charge centers was assumed to be 203:, which can be translated into a vertical distance of approxi- mately 3 km. Different heights for the negative charge cen— ters were tested with respect to the necessary horizontal distance between the positive and negative charge centers. Summer thunderclouds with a 5 km height of the negative charge center were found to generate positive ground flashes only was 9 negat posit place Tagak prod: 15 only if the horizontal distance between the charge centers was greater than 2.3 knn In Norwegian winter storms with negative charge centers at 1.]un height, the fraction of positive CGs was high even in the absence of horizontal dis- placement between the positive and negative charge centers. Tagaki et al. concluded from their study that it was "easy to produce positive ground flashes when the height of the charge is low and the wind shear is strong." The discussion about the origin and location of positive flashes is still continuing. Orville et al. (1988), Stolzen- burg (1990), and Engholm et al. (1990) examined more closely the phenomenon of a bipolar pattern in the distribution of +CGs and -CGs generated by mesoscale storm systems. Orville et a1. (1988) and Stolzenburg's (1990) findings differed with respect to the seasonal occurrence of bipolar patterns. Orville suggested a higher frequency for autumn and winter; Stolzenburg, on the other hand, maintained that summer is the season of higher frequency of this charge structure. She, however, shared Orville's opinion that in the winter months the percentage of storms with this pattern is higher and the bipolar pattern is better defined. In a February 1987 storm system along the Gulf coast, Orville found the ratio of posi- tive to negative flash density to be 0.1 and the approximate length of the bipolar pattern to be 100 km, clearly longer than the cloud extend of the mesoscale system. Engholm et al. (1990) found support for the "tilted dipole hypothesis" in case studies of winter and summer stor: (l) 1 and deepe catic Enghc Orvii ing cont< expa: vvvarié 16 storms. Common features of the lightning bipoles included (1) the alignment of the bipole with the vertical wind shear, and (2) predominance of negative strikes in proximity to the deepest convection and a mixture of positive and negative lo- cations displaced downshear from the deepest convection. Engholm et al. (1990) and Stolzenburg (1990) did not affirm Orville's hypothesis of a surface boundary typically separat- ing the positive contoured flash center from the negative contoured flash center. Engholm et al. (1990) in addition expanded on Orville et al.'s findings suggesting seasonal variations in the alignment of bipoles. They suggested that "lightning 'bipole' orientations are aligned with the geostrophic wind in winter storms and with the vertical wind shear (which may be highly ageostrophic) in summer storms, with positive locations downwind (downshear) from negative locations." Engholm et al. (1990) and Stolzenburg (1990) also hold different views on the generation of positive flashes. According to Stolzenburg the two main theories for positive charge build-up are: (l) the transfer of negative charge to the cloud-base by large precipitation particles and transfer of positive charge to the higher cloud regions by tempera- ture-related cloud processes of small particles, and (2) con- vection of positive space charge into a growing cloud, whereby positive charges are carried to the upper portions of the cloud, and negative charges are carried down by convec- tive motions at the outside of the cloud. While Stolzenburg 17 adheres to the theory of a large-scale advection of positive charge downwind, Engholm et al. (1990) favor the idea of an interplay of several mechanisms in the occurrence of positive lightning. Lightning bipoles are seen as the result of "ac- tive charge separation in convective clouds distributed over the entire area of the observed bipoles and not just in the deepest convection." In agreement with other studies (e.g. Rutledge and MacGorman 1988), the authors proposed that stratiform precipitation regions "may produce lightning inde- pendently of the deeper convection with which they are asso- ciated." Engholm et al. (1990) found lightning rates to in- crease dramatically with cloud depth. While negative strikes were prevalent in deeper clouds, positive lightning was more prevalent in shallow clouds located downwind from the deeper convection. Return stroke characteristics have also been found to vary with different stages in convective systems. Goodman and MacGorman (1986) pointed out the relationship between different stages of Mesoscale Convective Complexes and the number of return strokes. They found that the most active electrical period (i2 h of the peak ground discharge rate) was characterized by "the greatest average number of discrete strokes (3-4 component strokes to ground) per flash and largest fraction of multiple stroke discharges." The first hour in the development of a mesoscale complex, on the other hand, contained a greater fraction of single stroke dis- charges. and : [er-”’7 ~ -11-, 1 18 W The lightning-strike data set utilized in this research project was obtained from the eastern branch of the national lightning detection network operated by the State University of New York at Albany (SUNYA). The data set is comprised of observations for 22 months, January 1, 1989 to October 25, 1990. The short length of the data set is due to the fact that no data were available for the time-period before January 1989, when the network was established in its en- tirety. The data were purchased in November 1990, which ex- plains the abrupt ending in October 1990. The area of observations extends from 78.9° to 96.7° W. and from 38.3° to 47.1° N. (Figure 1). our/M” Fig. 1. Area of Coverage of the SUNYA data set. abox thu: ice: Char a1. Stud. tiVi 19 It thus includes portions of both areas of interest for this study; namely, the region of nocturnal maxima of precipitation and thunderstorm activity in the central US. and the region of afternoon maxima in the eastern US. The recorded components for each lightning event were time, location, number of return strokes, polarity, and amplitude. :1. l' E I] SI 1 This study is a preliminary assessment of the climatol- ogy of cloud-to-ground lightning activity in the southern Great Lakes region. It is being conducted in connection with a larger, ongoing research project at Michigan State University's Department of Geography concerned with the cli- matology of convection in this region. The southern Great Lakes region was chosen as the study area because, as noted above, it lies astride the boundary between the nocturnal thunderstorm and precipitation regime in the central United States and the afternoon regime found in the eastern and southern United States. Consequently, this region is ideal for a systematic comparison of the characteristics of noctur- nal and daytime convection. As discussed above, a number of previous authors have identified temporal variations in the frequency and characteristics of lightning activity (i.e. Fuquay, Brook et al., Orville, and Orville et al.). In the first part of the study, the diurnal and seasonal variations of lightning ac- tivity with respect to the frequency of ground strikes, 20 number of return strokes, peak amplitude, and polarity, are investigated for the entire southern Great Lakes region. This analysis expands on that of the previous authors in that it provides valuable climatological information for an addi- tional part of the United States and more systematically com— pares, for each month of the study period with recorded lightning activity, lightning characteristics for daytime (1200-0000 GMT) and nighttime (0000-1200 GMT) periods. In the second part of the study, the focus is on the spatial variations across the study region in the charac- teristics of lightning activity. In particular, spatial variations in the timing of lightning activity are empha- sized. As noted above, previous authors have identified the southern Great Lakes region as a transition zone between noc- turnal and afternoon convective regimes based on the fie- quency of thunderstorms and precipitation events. An objec- tive was to determine whether lightning activity also dis- plays similar spatial and diurnal variations. One final objective of this thesis is the development and/or assessment of computer software suitable for spatial and temporal analysis of lightning observations. This software will be used for future studies encompassing broader regions and utilizing longer periods of observations. PART I TEMPORAL ANALYSIS 21 spanr 1990. moat? were 1990 than Thc I1] CHAPTER 2 FREQUENCY OF LIGHTNING EVENTS E J D' | .1 . As noted previously, the data set used in this analysis spanned almost 22 months, January 1, 1989 until October 25, 1990. Lightning activity was reported during only twenty months. In both February 1989 and December 1989 no strikes were recorded for the study area. Despite the fact that in 1990 data were only available until October, the total number of strikes was considerably higher in 1990 (25,549 strikes) than in 1989 (22,168 strikes) (Figure 2). Thousands EJDay Eflnmhr O -1. E ;' J FIAAIM Fig. 2. Total Number of Nighttime and Daytime Flashes. 22 23 The annual distribution of lightning strikes is in general agreement with other studies (e.g. Orville et al. 1987). As expected, a summer maximum in lightning frequency is evident, followed by a rapid decrease during autumn and a frequency minimum in the winter months (Figure 2). Bear in mind, that no information was available for November and December 1990. Significant variations in lightning frequency are evi- dent between 1989 and 1990. While the period of enhanced lightning activity ends abruptly in September in both years, the start of the enhanced lightning period varies consider- ably. In 1989, the period of increased lightning activity lasts from April through September; a steady increase in lightning activity is evident from April to August followed by a marked decrease toward September. Compared to 1989, 1990 has a: shorter period of increased lightning activity starting in May and continuing until September. The fre— quency of lightning activity in 1990 does not increase steadily at the beginning of the period; rather, lightning frequency decreases in April after higher March values. In addition, the monthly occurrences of lightning events in 1990 do not display as smooth a frequency curve as compared to 1989. Note the increase in the number of light- ning events1 in May 1990 after considerably lower April val- ues. In fact, the number of occurrences in May 1990 exceeds 1Lightning event is used synonymously with cloud-to-ground flash or strike throughout the thesis. 24 those reported in June 1989. Two pronounced peaks in the 1990 frequency distribution, evident in June and August, are separated by July values that were lower than those recorded the previous year. The higher number of strikes and the uneven distribution of occurrences in 1990 cannot be explained by differences in the number of days with lightning occurrences between the two years (Table 1). Table 1.—-Annual Distribution of Lightning Events and Lightning Days 1989 1990 Month lightning Lightning Events Lightning lightning Events Events Davs per Events Davs per Lightning lightning lDay D2¥___. Jan 37 6 6.17 62 8 7.75 Feb -- -- -- 4 3 1.34 Mar 211 12 17.58 513 10 51.30 Apr 1869 17 109.94 143 14 10.21 May 2539 24 105.79 3764 29 129.79 Jun 3688 29 127.17 7148 30 238.26 Jul 5192 31 167.48 4500 30 150.00 Aug 5303 30 176.77 5888‘ 28 210.29 Sep 2575 18 143.06 2882 28 102.93 Oct 445 17 26.18 645 12 53.75 Nov 309 9 34.34 Dec -- -- -- The total number of lightning days is almost identical in both years with 193 lightning days in 1989 and 192 days with lightning occurrences in 1990. However, the comparable months of both years show remarkable differences in the number of lightning events per lightning day. The ratio of 25 lightning events per lightning day is considerably higher in 1990. For example, April 1990, has almost the same number of lightning days as April 1989, but the lightning events of April 1989 amount to less than one tenth of the number of lightning strikes of April 1989. Also, June 1990 recorded only one more day of lightning activity than June 1989 but 3500 more occurrences of lightning strikes. In August 1990 a larger number of strikes was recorded on fewer days than in August 1989. The higher number of strikes per lightning day in 1990 is likely a result of different characteristics of the lightning-producing storm systems. The difference in the to- tal number of events may result from differences in the areal extent of storm systems in 1989 and 1990. Storms that cover larger areas and are long—lived most likely have considerably higher flash totals than small, short-lived storms. Maier and Krider (1982), for example, reported that severe thunder- storm systems in the Great Plains were slow moving, well— organized structures with fairly high flash-density rates. Therefore, a possible cause of the discrepancy between the two years is a higher flash—density rate2 due to major storm events in 1990. A few major storm systems can produce a high percentage of the annual cloud-to-ground lightning receipts in an area. A future research goal in this context is to de- termine if time recorded ground flashes during seasons and 2Flash-density rate shall here be defined as the receipt of ground flashes in a unit area per unit time. 26 years of greater activity result from a few major storm events or if they are caused by an increased number of smaller individual thunderstorm cells during those seasons. Ti E D E Ii 1| . E I' 'I To assist in the initial investigation of the diurnal variability of lightning activity, CG strikes were first as- signed to one of two broad periods, 0000-1200 GMT (nighttime events) and 1200-2400 GMT (daytime events). Both 1989 and 1990 display a higher total number of nighttime events than daytime events for the entire study area (Figure 2). While there is a general tendency for a larger ground flash occurrence at night, for some months the maximum lightning activity occurred during the day. The likelihood of a daytime maximum is greater during warm season months. July, for example, experienced a daytime maximum in lightning activity during both years (Table 2). In addition, more ground flashes during the day than at night were reported during June, 1989; October, 1989 and April, 1990. Table 2.--Nighttime Events as Percentages of the Monthly CG Totals Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov 1989 51 -- 61 70 60 47 43 52 62 49 53 1990 77 50 62 38 64 64 46 63 57 65 -- SOD plaj sma. 27 The larger number of lightning events in 1990 may be re— lated to the time of day of lightning occurrence. Note that in 1989 53 percent of the total number of C63 occurred at night, whereas in 1990 60 percent of the recorded lightning strikes were nighttime events. Note also that the warm sea- son months of May through August, 1990 (except for July) dis- play pronounced nighttime frequency maxima. In contrast, the small number of events of April 1990 compared to April 1989, were associated with a daytime frequency maximum. One can speculate that nighttime events either have higher flash densities or larger areal extent or are longer duration events compared to daytime storms. Summary. The annual and diurnal variation of lightning frequency during January, 1989 through October, 1990 was examined. Lightning activity was most common in the summer months with the monthly frequencies varying considerably between the two years. The number of lightning days did not show much variation between years although the number of days varied significantly between months. Lightning is more likely to occur on a particular day during May through September when lightning activity was reported somewhere in the study area on almost every day of the month. A relationship between the time of day of lightning events and the frequency of strikes was proposed. The larger frequency during 1990 is suspected to be a consequence of the different areal extent, 28 persistence, and flash density of nocturnal versus daytime convective systems. CHAPTER 3 RETURN STROKES AND AMPLITUDE Methodology The previous chapter examined the frequency of lightning activity separately for each month of the study period in or- der to better illustrate the large monthly and annual varia- tion in lightning activity. In this chapter, the return stroke and amplitude characteristics of Great Lakes lightning months are discussed by individual month and by each month of the two year period combined. This was done in the hope of better identifying variations in these two lightning proper- ties. Due to the lack of data for November and December 1990, only those months for 1989 were used. Return stroke and amplitude characteristics have previ— ously been recognized to vary with different stages of storm systems. Orville (1987, personal comment in Stolzenburg, 1990), for example, had reported an inverse relationship be- tween the rate of negative cloud-to-ground lightning and mean peak currents. For mesoscale systems, on the other hand, Goodman and MacGorman (1986) had noted the correspondence of the most active lightning period with the period of the high- est number of return strokes per flash. 29 30 For the analysis of any temporal trends in return stroke and amplitude characteristics, five-minute maxima were em- ployed. Five-minute maximum values have previously been used by Lopez and Holle (1986) to examine flash density rates. The analysis of five-minute intervals has the advantage that a fairly high degree of diurnal variability is preserved. For this thesis, maximum values of return strokes and ampli- tude were of special interest since strong cloud-to-ground lightning flashes produced by major thunderstorms pose the greatest danger to both property and human life. Each successive five minute period of a nmmth (00:05- 00:10, 00:10-00:15, etc.) was examined for (1) the strike with the maximum number of return strokes, (2) the positive and the negative strike with the strongest amplitude. In the analysis of five-minute maxima for return strokes no distinction was made with respect to polarity. An initial check of the data set had shown that it was in agreement with the findings by Orville et al. (1987) and Fuquay (1982) who proposed that most positive flashes only contained one return stroke. The frequency distributions of five-minute maxima of return strokes therefore describe the characteristics of negative flashes. Positive and negative strikes possess different charac- teristics with respect to amplitude, however. Positive strikes had been reported to possess higher median amplitudes than negative strikes (Orville et al. 1987). In addition, Brook et al. (1982) found that in winter storms positive 31 flashes can have peak currents about one order of magnitude greater than negative currents, thus being comparable to neg- ative peak currents in summer storms. The different ampli- tude characteristics of positive and negative flashes re— quired a distinction by polarity. Therefore, five-minute maxima amplitudes for negative and positive flashes were cal- culated and plotted separately. The resulting maximum values of all three variables were grouped into seven categories each. The plots for all three variables, return strokes, positive and negative amplitude, are placed alongside each other in the following figures to facilitate better compari- son. The temporal characteristics of the number of return strokes will be discussed first, the discussion of monthly variation in strike amplitude is delayed until the end of this chapter. A comparison of the frequency distributions of five- minute maxima of return strokes and negative and positive amplitude did not reveal any corresponding temporal patterns. While diurnal patterns were not very pronounced, an annual curve was evident in the five-minute maxima values for all three parameters. 8' SI 1 a I I i 0'12 I]. i W The different characteristics of positive and negative strikes with respect to the number of return strokes have been noted by several authors. Rust et. al. (1981) and 32 Fuquay (1982) noted that most positive CGs only contained one return stroke. Orville et al. (1987) found that the percent- age of negative strikes with only one return stroke increased from 40 percent in the summer to 80 percent in January and then decreased again to less than 50 percent in the spring. For positive strikes they reported approximately 90 percent possessing only one return stroke throughout the year. As expected, the number' of return strokes varies markedly for positive and negative flashes in the Great Lakes region during 1989-1990 (Table 3). February 1990 is rmm taken into account due to the scarcity of recorded strikes. Table 3.--Percentages of -CGs and +CGs with only one Return Stroke 1989 1990 Month Percent +CG Percent -CG Percent +CG Percent -CG with one with one with one with one Return Stroke Return Stroke Return Stroke Return Stroke Jan 84 47 82 31 Feb -- -- -- -- Mar 70 42 77 35 Apr 75 35 70 48 May 77 34 71 33 Jun 75 35 75 34 Jul 81 34 83 35 Aug 82 32 76 29 Sep 71 29 75 35 Oct 79 38 80 34 Nov 72 39 The percentages of positive strikes with only one return stroke, which range from 70 percent to 84 percent of the total number of flashes throughout the year, are somewhat lower than the percentages reported by Orville et a1 (1987). 33 Also, unlike Orville et al.'s findings, the percentages of negative CGs with one return stroke do not display an annual curve in this study. The percentages ranged from 29 to 47 throughout the year, with the percentages for most months falling between 32 and 35. These values approximate the lower percentages reported by Orville et al. for the summer months. The higher percentage of +CGs with only one return stroke is also reflected in the mean number of return strokes (Table 4). Again, per flash February 1990 was excluded due to the scarcity of recorded flashes. The average number of return strokes for +CG flashes ranged between 1.14 and 1.48 whereas corresponding numbers for —CG flashes were 2.1 and 3.04. Table 4.--Mean Number of Return Strokes for Positive and Negative CG Flashes ' 1989 1990 Month Mean Number Mean Number Mean Number Mean Number +CG Return -CG Return +CG Return -CG Return Strokes Strokes Strokes Strokes Jan 1.14 2.47 1.18 2.49 Feb -- -- -- -- Mar 1.35 2.32 1.35 2.39 Apr 1.41 2.59 1.39 2.10 May 1.37 2.75 1.43 2.73 Jun 1.35 2.65 1.37 2.71 Jul 1.24 2.71 1.25 2.67 Aug 1.27 2.83 1.37 2.97 Sep 1.48 3.04 1.33 2.67 Oct 1.24 2.51 1.33 2.68 Nov 1.28 2.28 Orville et al. (1987) reported the highest number of re— turn strokes to be fourteen. In this study, the highest rec fla per inc max 1 fla Tab. t m wwrmemflmfipt ax . » vi. JFMJLAMJJA%W%U “V O... h J 34 recorded number of return strokes associated with one ground flash is twenty-six. While the mean number of return strokes per flash is similar throughout the year, the maximum values increase during the warm months (Table 5). As expected, the maximum number of return strokes associated with negative CG flashes greatly exceeds that for positive strikes. Table 5.--Maximum Number of Return Strokes for +CGs and -CGs 1989 1990 Month Max. Number of Max. Number of Max. Number of Max. Number of Return Strokes Return Strokes Return Strokes Return Strokes +CG -CG +CG —CG Jan 2 9 2 6 Feb -- -- 2 2 Mar 3 12 4 11 Apr 4 15 3 9 May 8 13 7 13 Jun 6 16 5 16 Jul 5 15 4 15 Aug 5 15 7 26 Sep 11 15 4 16 Oct 3 10 5 14 Nov 2 10 900 o ._ '. or '9 0‘ vo..u_u \_II_.‘ o 4‘ . o 0.‘ The return strokes for both +CG flashes and -CG flashes for both years were combined for the analysis of the monthly variations in the number of return strokes per flash. Since the number of return strokes associated with negative flashes is considerably higher than that for positive flashes, the frequency distributions of five-minute maximum return stroke values, presented below, can be considered to reflect the characteristics of negative rather than positive strikes. revi the Gem sum (26] We re 35 Each five-minute interval per nmnth (00:00-00:05 of April 1989 and 1990, etc.) was searched for the maximum num- ber of return strokes within that interval, and the maximum value was assigned to one of seven equally spaced categories. The monthly distributions of the five-minute maxima reveal some seasonal variations (Table 6). As shown before, the winter months had comparatively fewer return strokes. Generally, the number of return strokes was largest in the summer months. The absolute maximum number of return strokes (26) occurred in August 1990. Flashes with 26 return strokes were recorded both during the day and at night. November through March had the largest portion of night- time (0000-1200 GMT) five-minute periods with maxima falling in category I (0-3 return strokes). A shift toward a higher percentage of five-minute periods with at least four return strokes is evident in April and May. From May through September, the maximum number of return strokes for most of the five-minute periods fell into category III (8-11 return strokes). Note that during these months at least four return strokes were reported as neximum values for all the five- minute periods from 0000 to 1200 GMT. June and August were the only months when nocturnal lightning events with 16 to 20 return strokes were recorded. The largest number of return strokes per lightning event was reported in August. Note that August was the only month when the maximum number of re- turn strokes for a five—minute period exceeded 24. Also, the for what mont five 36 largest number of five-minute periods with 8-12 (category II) and 12-16 (category IV) return strokes occurred in August. The distribution of the maximum number of return strokes for daytime (1200-2400 GMT) five-minute periods varies some- what from that for the nighttime period. Note that in each month, the maximum number of return strokes for at least one five-minute period was less than four, although a smaller number of return strokes per event was still most common in the cool months, as was the case for the nighttime period. As for the nighttime periods, the maximum number of return strokes per daytime five—minute interval is likely to fall between four and eight strokes in April and May. The period with the highest percentage of five-minute maximum values of return strokes falling within category III was shortened con- siderably (June-July) compared to nighttime activity (June- September). Only in June, August, and September did the max- imum number of return strokes fall within category V (16-20 return strokes) for daytime five-minute periods, and August is the only month where more than 24 return strokes were recorded. However, the majority of five-minute nighttime pe- riods during August reported maximum rates in categories III, whereas the majority of the August daytime periods recorded only 4-8 return strokes (category II). Ta Cate \«J me no: 37 Table 6.--Five-minute Maxima of Return Strokes, 1989 and 1990.3 Nighttime (0000-1200 GMT) Category Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov (N of R.S.) I (<4) 42 2 68 14 4 -- -- -- -- 50 85 II (4-8) 12 -- 62 106 64 34 56 27 59 79 15 III (8-12 2 -- 8 20 63 96 80 100 75 14 3 IV (12-16) -- -- 1 2 13 13 8 14 10 1 -- V (16-20) -- -- -- -- -- 1 -- l -- -- -- VI (20-24) -- -- -- -- -- -- -- -— -- -- -- VII (>24) -- -- -- -- -- -- -- 2 -- -- -- Daytime (1200-2400 GMT) Category Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov (N of R.S) I (<4) 22 2 99 44 1 3 1 1 3 73 71 II (4-8) 3 -- 35 86 87 57 47 89 78 62 20 III (8-12) -- -- 7 13 49 75 82 42 55 9 3 IV (12-16) -- -- -- 1 7 7 14 8 7 -- -- V (16-20) -- -- -- -- -- 2 -- 2 1 -- -- VI (20-24) -- -- -- -- -- -- -- -- -- -— -- VII >24) -- -- -- -- -- -- -- g -- -- -- 3 The frequency values indicate the number of five-minute periods with maximum reported number of return strokes in the respective category. Data for November are for 1989 only. 38 W The diurnal distribution of the maximum number of return strokes recorded during five-minute periods for January shows an increase in the frequency of return strokes during the evening hours (Figure 3a). More specifically, within the time period of greater lightning activity from approximately 2200 to 0800 GMT (1600-2000 CST), the highest values of return strokes per flash were observed in the late afternoon/early evening hours (2300-0200 GMT). The four recorded strikes in February4 do not reveal any diurnal pattern in the number of return strokes (Figure 4a). In Figure 5a, note the general increase in the maximum number of return strokes per flash during March. The period with the largest value (12) of return strokes per flash was 0550 to 0555 GMT, although this value is the result of one singular, extreme strike. In general, the maximum number of return strokes is larger in the nighttime period. The number of return strokes per flash declines sharply during the early morning hours (approximately 0900 GMT) but increase once again in the early afternoon. A secondary maximum is evident at 2110 to 2115 GMT. In April, large (214) return stroke values were recorded around 0700 and 0900 GMT (Figure 6a). Again, these lightning events are singular, unique occurrences, as the maximum num- ber of return strokes for all other nighttime five-minute 4 February 1990 only; no flashes were recorded in February 1989. 39 BMW. 20 20 ‘5..’ ................................. ‘5 ................................... 1° ................................... 5 ......... . ....................... o I l 1 I ll [1 1 ill 1234567891011 1314151617181920212223 TIIMIGM‘D Tim-(GMT) Fig. 3a. Five-minute Maxima of Return Strokes, January 1989 and 1990. Strength Strength 300 300 2” ........................ a“ “..H a” ..................... an .............. ,w .................................. ,w .................................. ,w .............................. ,m ................................. w ...... w ............................ o ‘ ‘ 1 l 1 I I 1314151617181920212228 Fig. 3b. Five-minute Maxima of Amplitudes for Negative CG flashes, January 1989 and 1990. sou-19111 Sir-oath soo 300 250 ................................. 250 ....................... , . . ...... 20° ................................. 20° ...................... 15° . ................... 150 ...................... 1oo ............................... .00 ................................ 5° ............................. 5° ............................. II I I o 1 2 a 4 s o 7 a 9 1o 11 o 1: 141510 1713 197720 21 22 2: Fig. 30. Five-minute Maxima of Amplitudes for Positive CG flashes, January 1989 and 1990. 40 Ream Stone Return Sirokee 20- 20 ‘ s ................................... 1'5 ................................... ‘0 .................................... 1o ................................... 5 .................................. .1 5 ................................... o l' i o 1 l 1 2 3 4 5 8 7 8 9 1O 11 13 14 15 18 17 18 19 20 21 22 23 Tune (mm Time (mm Fig. 4a. Five-minute Maxima of Return Strokes, February 1989 and 1990. Strength Strength 300 300 250 .................................. 25° .................................. 20° .................................. 20° .................................. ‘ so .................................. 1 so .................................. ‘m .................................. I” .................................. so .................................. 5° ..... I: ............................ O 1 2 3 4 8 8 7 8 8 10 11 13 14 15 18 17 18 19 20 21 22 23 Fig. 4b. Five-minute Maxima of Amplitudes for Negative CG flashes, February 1989 and 1990. Strength 300 Strength 250 ............................. 150 ........... ...................... ................................ 50 Fig. 4c. flashes, February 1989 and 1990. 13 14 15 18 17 18 12 20 21 22 23 Five-minute Maxima of Amplitudes for Positive CG 41 Return Strokes Return Strokes ‘5 .................... 1s . . i l 1. | ‘ 2| ‘ ii 1! i‘zli I t i , 9 J 1 ilk-hi: I! ,ifl‘ i , o 999199: 1.1191 ‘ 19,119 MUM." 1 2 a 4 s e 7 e 9 1011 13141518171819 20 21 22 23 Ti m TimeIm Fig. 5a. Five-minute Maxima of Return Strokes, March 1989 and 1990. 3003114119111 300311-1911. ,5, ................................ m m ............................. m ................ m . . ....................... ,5, .............................. m . ....... 5° 9‘19 l | 9‘ “mil 9“’*,. ‘ 9 .,, gnu onww.*n . - H ; 1.‘ L ‘ 5 M“ " 1 2 3 ‘ 5 C 7 U 9 10 11 13 14 15 16 17 18 19 20 21 22 23 nun-(own TImo(GMT) Fig. 9a. Five-minute Maxima of Return Strokes, July 1989 and 1990. 1234507891011 1314151017101920212223 Fig. 9b. Five-minute Maxima of Amplitudes for Negative CG flashes, July 1989 and 1990. Slmgth Stronglh 0 12 3 4 5 G 7 l .1011 1314151617181920212223 Fig. 9c. Five-minute Maxima of Amplitudes for Positive CG flashes, July 1989 and 1990. 47 declines between 1400 and 1530 GMT and then increases from 1530 to 1930 GMT. At this time, the maximum number of return strokes is greater or equal 8 for most five-minute periods. The maximum number (15) of return strokes for the month of July was recorded three times between 1655 and 2000 GMT, and again was recorded in five-minute intervals with no equally high values preceding or following it. August is the month with the highest number of return strokes recorded (Figure 10a; note the change in scale). Four five-minute periods during August 1989 and 1990 con- tained flashes with 26 return strokes. In general, the num— ber of return strokes is higher for the nighttime compared to daytime flashes. From the early morning hours until the early afternoon (0900-2000 GMT), the maximum number of return strokes is lower than at other times of the day except for the five-minute maxima of 26 return strokes at around 2000 GMT. An interesting feature of the flashes with 26 return strokes is that the maximum number of return strokes for the preceding five-minute intervals is larger than the number of return strokes for other five-minute periods in that hour. This suggests that the storms or cells associated with these unusual flashes produced a number of flashes with exception- ally large numbers of return strokes. It remains to be tested if the flashes with 26 return strokes are singular features or if several flashes with extraordinarily high re- turn stroke rates occurred in the same five—minute period. 48 Mum 3m” Rotum Strokes E"“1..thun a H“ 1314151017101920212223 mueum Fig. 10a. Five-minute Maxima of Return Strokes, August 1989 and 1990. Stung". Stungm 2501 w l I l 200 0 150! ‘ M ‘ "HM \ U‘HIH I m N111“ 1W 0 1 2 3 4 5 IL 7| 0 0 10 11 13 14 15 10 17 10H19 20 21 22 23 Fig. 10b. Five-minute Maxima of Amplitudes for Negative CG flashes, August 1989 and 1990. Strength Slnnglh -.._ ‘ :33 Ll; ix ‘1‘} 12 3 4 5 0 7 0 01011 1314151017181920212223 Fig. 10c. Five-minute Maxima of Amplitudes for Positive CG flashes, August 1989 and 1990. 49 In September, the maximum number of return strokes for nighttime flashes is generally higher than that for daytime flashes, despite the fact that the flash with the highest number of return strokes was recorded during the day (Figure 11a). In October, a general drop in the number of return strokes as compared to the preceding months is apparent (Figure 12a). .As for September, the nighttime five-minute maxima of return strokes are somewhat larger than those for the daytime periods. From slightly larger values between 0300 and 0900 GMT, the maximum number of return strokes de- creases beginning at 0900 GMT and extending throughout the day until 2400 (flflh Note also that the number of return strokes varies considerably from period to period, particu- larly during the daytime hours. In November, the maximum number of return strokes per flash generally is largest between 1100 and 1600 GMT, al- though the largest number of strokes was recorded for an in- dividual strike during the night between 0215 and 0220 GMT (Figure 13a). 50 Pig. mas Rm sm- Fig. 11a. 1989 and 1990. Suenglh H W My]: 11b. M Fig. 50 Return Strokes 1314151017131920212235 M0161“) Five-minute Maxima of Return Strokes, September Strength flashes, September 1989 and 1990. Sirength sac . 1O 11 Fig. 11c. Five—minute Maxima of Amplitudes for Negative CG Strength flashes, September 1989 and 1990. 13 14 15 13 17 1B 19 2O 21 22 23 Five-minute Maxima of Amplitudes for Positive CG w 254 2’3: ‘5! 51 Return Strokes Return Stroke: 20 20 ‘5 . . . . .............................. ‘5 ................................... 1° ................................. II I ‘.I 'I II I IEII @351 OIIIIIII II Ii‘»""1?“=wi;w"‘ 13 14 15 16 17 18 19 20 21 22 23 Timemu‘n Fig. 12a. Five—minute Maxima of Return Strokes, October 1989 and 1990. Strength mSttenglh ‘50 ............................ . . 150 .............. . . . . ..... 1oo . .. ...... I ”I L 5° EIII “I“IIWIIIIIII ‘ ‘ IIIMII ”II IIIWIIII Fig. 12b. Five—minute Maxima of Amplitudes for Negative CG flashes, October 1989 and 1990. 3W smogm zoo 300 250 .................................. 25° ............... 20° ................................. 20° ............... 15° ......................... ,50 10°“... ...... ..... ‘00 II I o 1 7 0,. 1o 11 . 13 14 15 16 17 13 19 20 21 22 23 Fig. 12c. Five-minute Maxima of Amplitudes for Positive CG flashes, October 1989 and 1990. ‘5 ................................... 123433739 human 10 11 Fig. 13a. 1989 and 1990. 8W ‘50 .................................. 1m .............................. I 1 2 3 4 3 3 7 3 91011 Fig. 13b. Five-minute Maxima of 52 Retum Strokes 2O ‘5 ................................... ‘0 ................................... 5 .. . .......................... .II II ‘II ‘III :I,‘ ‘ I O IIHIIWHIII IIII I I 1314131317131920212223 "11.16“” Return Strokes, November Sttength ‘50 ................. ‘m ............. m. I". o I» II; IIIIIIIIII II : I 1314151317 1319 20 2122 23 Five-minute Maxima of Amplitudes for Negative CG flashes, November 1989 and 1990. Strength 300 30° 25° ................................. 25° ........................... 2.. .................................. 2.. ........................ m ................................. m, ................................. .0. ................................ ,0 .................. I.. .....I o I II 1 2 3 4 3 3 7 3 9 Fig. 13c. flashes, November 1989 and 1990. 1314131317131920212223 Five-minute Maxims of Amplitudes for Positive CG 53 E JII 1 E H I. 1 E .I . SI .1 The amplitude of lightning strikes has previously been examined with respect to polarity. Orville et al. (1987), for example, reported that the median peak current for posi- tive strikes exceeded the median peak current for negative strikes. Amplitudes of >100 kA were observed with negative strikes, whereas amplitudes associated with positive strikes exceeded 200 kA. Also, the median peak current of both +CG and -CG increased in the winter season. The results of this study are in general agreement with those of Orville et al. (1987). While the median values were not calculated, the mean values of amplitude for positive strikes are considerably higher than the mean values for neg- ative strikes (Table 7). Table 7.--Mean Amplitude of +CGs and -CGs 1989 1990 Month Mean Amplitude Mean Amplitude Mean Amplitude Mean Amplitude +CG (in kA) -CG (in kA) +CG (in kA) -CG (in kA) Jan 100.4 45.5 106.2 47.4 Feb -- -- 93.0 37.0 Mar 79.9 45.4 67.6 39.9 Apr 83.1 35.1 76.7 41.6 May 77.8 36.1 84.7 40.5 Jun 68.4 35.2 64.9 33.4 Jul 63.3 35.7 56.5 33.0 Aug 60.4 35.8 65.6 34.1 Sep 57.5 36.4 66.7 34.3 Oct 81.1 38.1 73.3 36.0 Nov 75.7 39.4 54 The largest mean amplitudes of both +CGs and —CGs occurred in January of both years. A slight decrease in amplitude toward summer was found for both positive and negative events fol- lowed by a slight increase toward autumn. The previously described procedure for cmmaining the five-minute maximum value of a variable was also employed in the analysis of the diurnal patterns of the amplitudes of ground flashes. A distinction, however, was made with re- spect to polarity. The occurrence of fewer positive strikes on one hand and the expected higher amplitude of positive strikes on the other hand required that the five-minute maxi- mum values for amplitude be calculated separately for posi- tive and negative strikes. In order to identify and compare the temporal characteristics of maximum amplitude, separate plots were made for negative and positive strikes. The same five-minute intervals employed for the analysis of the number of return strokes were used in the analysis of strike ampli- tude to allow for comparison of the diurnal properties of all three variables. Also, the distributions of maximum ampli- tude were summarized by grouping the five-minute periods into seven categories based on the maximum amplitude observed in that period. In the following discussion, February 1990 will not be included due to the scarcity of recorded flashes. The distributions of the five-minute maxima of return strokes and negative and positive amplitude are not in agree- ment. Diurnal and monthly variations are much less pro- nounced for maximum amplitude compared to the number of 55 return strokes. Nevertheless, some variations between nega- tive and positive amplitude maxima are evident. The distribution of five-minute maxima for negative and positive strikes reveals considerable 'variation Ibetween strikes of either polarity. While the largest amplitudes are associated with negative flashes, the likelihood of am- plitudes exceeding 100 Ink is greater for positive strikes than for negative strikes. For the following discussion, the percentages of five- minute maxima for +CGs and -CGs (Table 9) were calculated for each category in addition to the number of five-minute inter— vals with maximum values in respective categories (Table 8) to facilitate comparison. Due to the limited length of the data set, run: all five-minute periods contained lightning flashes. Also, comparatively fewer five-minute periods re- corded positive than negative flashes. Emua conversion to percentages of five-minute periods helps detect annual trends in the maximum values of return strokes. The maximum strength per five-minute period is generally higher during the warm months than during the cold months (Table 8). Amplitudes 2250 kA during the night were only recorded for negative strikes. During the daytime hours one five-minute interval in May contained one or more positive flashes with amplitudes 2250 RA in addition to six 56 five-minute intervals in June and July with negative strike amplitudes 2250 kA. Amplitudes exceeding 300 kA were observed only for negative flashes during both nighttime (0000-1200 GMT) and daytime hours (1200-2400 GMT) in June and July. These findings are somewhat surprising as positive flashes have generally been attributed larger amplitudes. Orville et al. (1987) for example only mention amplitudes in excess of 200 kA in relation to positive flashes. During the nighttime hours, throughout the year, the ma- jority of five-minute maxima reporting positive strikes fall into category II (SO-100 kA) (Tables 8 and 9). In January, the maximum amplitudes for positive strikes are as likely to fall between 50-100 kA as between 100—150 kA (category III). In November, categories I, II, and III contain an equal num- ber of five-minute intervals. The distribution of the maxi- mum amplitude of negative strikes is somewhat different dur- ing the nighttime hours. From March to September, the high- est percentage of five-minute intervals falls in category II. However, in the cold months (October through January), a higher percentage of five-minute periods are found in cate- gory I (<50 kA). Thus, for negative strikes peak amplitudes tend to be smaller during the cool season compared to the warm season . Ta? III (10 IV (15 57 Table 8.--Five-minute Maxima of Amplitude for Positive and Negative Strikes, 1989 and 1990.5 Nighttime (0000-1200 GMT) Category Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov (in kA) I + 1 4 8 18 17 18 35 25 6 3 (<50) - 32 60 39 1 1 4 -- 7 73 71 II +- 3 15 28 44 47 43 45 44 19 3 (50-100) - 15 62 95 89 99 110 104 109 59 24 III + 3 3 12 32 18 8 23 12 7 3 (100-150) - 3 13 9 39 35 23 30 23 7 5 IV + 1 -- 4 16 4 1 3 4 4 2 (150-200) - -- 4 1 13 6 5 10 3 4 -- V +, -- 1 -- 1 1 -- 4 1 -- -- (zoo-250) - 1 1 -- 1 -- -- -- 1 -- -- v1 +_ —- -- -— -- -- -- -- -- -- -- (250-300) - -- -- -- 1 -- -- -- 1 -- -- VII +. -- -- -- -- -- -- -- -- -- -- (>300) - -- -- -- -- 3 2 -- -- -- -- Daytime (1200-2400 GMT) Category Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov (in kA) I +’ 2 12 8 13 32 30 32 27 13 3 (<50) -' 12 70 57 2 4 3 4 7 75 64 II + 4 14 13 44 39 27 24 16 10 6 (SO-100) - 5 39 74 81 102 100 91 112 57 21 III + 1 3 6 30 18 10 19 5 4 3 (100-150) - 1 8 11 49 26 33 41 22 7 3 IV +- 2 2 3 8 3 4 3 2 5 -- (150-200) - —- 2 -- 12 7 5 4 2 0 -- V + l -- -- 1 1 -- -- -- -- —- (200-250) - -- -- 1 -- 1 1 4 1 2 -— v1 +, -- -- -- 1 -- -- -- -- -- -- (250-300) - -- -- -- -- 2 -- -- -- -- -- VII +- -- -— -- -- -- -- -- -- -- -- (>300) - -- -- -- -- 2 2 —- -- -- -- 5 positive/negative maxima in the respective categories. Figures indicate the number of five-minute periods (out of 144) with 58 Table 9.--Percentages of Five-minute Maxima of Amplitude for +CGs and -CGs in each category, 1989 and 1990.6 negative maxima in the respective categories. Nighttime (0000-1200 GMT) Category Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov (in kA) I + 12 18 15 16 19 26 32 29 17 27 (<50) - 63 43 27 1 1 3 -- 5 51 71 II + 38 65 54 40 54 61 41 51 53 27 (SO-100) - 29 44 66 61 69 76 72 75 41 24 III + 38 13 23 29 21 11 21 14 19 27 (100-150) - 6 9 6 27 24 16 21 16 5 5 IV + 12 -- 8 14 5 2 2 5 11 19 (150-200) - -- 3 1 9 4 3 7 2 3 -- V +' -- 4 -- 1 1 -- 3 1 -- -- (200-250) - 2 1 -- 1 -- -- -- 1 -- -- vI + -- -- -- -- -- -- -- -- -- -- (250-300) - -- -- -- l -- -- -- 1 -- -- VII +_ -— -- -- -- -- -- -- -- -- -- (>300) - -- -- -- -- 2 2 -- -- -- -- Daytime (1200-2400 GMT) Category Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov ‘ (in kA) I +' 20 39 27 14 35 42 41 54 41 25 (<50) - 67 58 40 2 3 2 3 5 53 73 II + 40 45 43 45 42 38 31 32 31 50 (SO-100) - 28 33 51 56 71 70 63 78 41 24 III + 10 10 20 31 19 14 24 10 12 25 (100-150) - 5 7 8 34 18 23 28 15 5 3 IV + 20 6 10 8 3 6 4 4 16 -- (150-200) - -- 2 -- 8 5 3 3 1 0 -- v + 10 -- -- 1 1 -- -- -- -- -- (200-250) - -- -- 1 -- 1 1 3 1 1 -- VI + -- -- -- 1 -- -- -- -- -- -- (250-300) - -- -- -- -- 1 -- -- -- -- -- VII + -- -- -- -- -— -- -- -- -- -- (>300) - -- -- -- -- 1 1 -- -- -- -- 6Figures indicate the percentage of five-minute periods with positive/ 59 During the daytime hours, a higher percentage of five- minute periods fall in category II than in any other category for lightning activity of either polarity during April-June. However, the maximum amplitude for five-minute intervals with positive strikes during July-October are more likely to be <50 kA. For negative strikes, March is the only month where the distribution of five-minute maximum amplitude values dif- fers between the nighttime and daytime period. While during the night, the highest percentage of five-minute periods fall in category II, during the day the highest percentage of five-minute intervals fall in category I. The distribution of the percentage of five-minute peri- ods with maximum values 2100 kA relative to the number of five-minute periods with recorded lightning activity reveals considerable differences of amplitude characteristics between positive and negative strikes (Table 10). During the night, between 13 and 50 percent of the five-minute periods with recorded lightning activity reported maximum amplitudes 2100 kA for positive strikes whereas the comparable percentages for negative strikes lay between 5 and 38 percent. During the daytime period, between 14 and 41 percent of positive five-minute maximum amplitudes and between 3 and 42 percent of the negative five-minute maximum amplitudes were 2100 kA. The maximum amplitude values of negative strikes display an annual curve during both the nighttime and daytime hours. The percentage of five-minute intervals with maximum values exceeding 100 kA strongly increases in the warm months. ex tt tu of 60 These findings are contrary to Orville et al.'s (1987) who found the median amplitude to increase in the winter season. During the daytime, May through September is the season with a greater’ percentage of five-minute periods containing flashes 2100 kA. There is no clear annual curve in the dis- tribution of percentages of maximum amplitudes 2100 kA for positive strikes. However, there seems to be a tendency for a greater likelihood for maximum values of positive strikes to exceed 100 kA in periods with positive lightning activity in the cold months (January and November). The maximum ampli- tude values for May are intriguing: Note that the likelihood of a five-minute period reporting maximum amplitudes 2100 kA is considerably higher in May than in the surrounding months for both positive and negative strikes. The reasons for the seemingly stronger maximum amplitudes of lightning strikes in May remain to be clarified. Table 10.--Percentages of Five-minute Periods with Maximum Amplitude 2100 kA Relative to the Number of Five-minute In- tervals with Recorded Lightning Activity for Positive and Negative Strikes. Nighttime Nighttime - Daytime Daytime - +CGs CGS +CGs CGs January 50% 8% 40% 5% February -- -- -- -- March 17% 13% 16% 9% April 31% 7% 30% 9% May 44% 38% 41% 42% June 27% 30% 23% 26% July 13% 21% 20% 28% August 26% 28% 28% 34% September 20% 20% 14% 17% October 30% 8% 28% 6% November 46% 5% 25% 3% The diurnal variations of five-minute maxima of ampli- tude for negative and positive strikes are not as pronounced as for the number of return strokes. However, some differ- ences between the distributions of negative and positive strikes can be detected. The diurnal distribution of the maximum values of ampli- tude for negative (Figure 3b) and positive strikes (Figure 3c)7 during January reveals diurnal characteristics for the strikes of different polarity. While negative cloud-to- ground lightning displays a strong temporal concentration be- tween 2200 and 0800 GMT with peak activity at approximately 0200 GMT, positive CG flashes are spread out more evenly throughout the day and are considerably less frequent. However, the maximum amplitude of positive lightning strikes per five-minute period is considerably higher than for nega- tive strikes. The difference in amplitude between positive and negative lightning strikes is especially pronounced dur- ing the day. The February plot does not allow for extensive interpre- tation due to the scarcity of recorded strikes. Note the difference in amplitude between negative (Figure 4b) and pos- itive strikes (Figure 4c), however. ‘7Figures for maximum values of amplitude for negative and positive strikes are included previously in the section on maximum values of return strokes (pp. 39-52). 62 Temporal preferences of negative and positive strikes are also evident in March. While negative lightning was recorded during almost all five-minute intervals during the night (0000-1200 GMT), some five-minute periods did not re- port negative CG strikes (Figure 5b). In general, the ampli- tude of negative lightning strikes is stronger during the night, especially between 0000 and 0600 GMT. Interestingly, the period with.ru> positive lightning activity (0130-0330 GMT) occurs during the period of high amplitude negative strikes. Positive lightning (Figure 5c) is less frequent during the night than during the day. Maximum amplitudes of positive and negative flashes do not differ greatly between the daytime and nighttime period. Generally, the five-minute maximum values of amplitude for negative CGs in April (Figure 6b) do not show as much diurnal variation as negative CGs in March. While the smallest maximum values in almost all five-minute periods are 250 kA, very few five-minute periods contain strikes that exceed 100 kA. Also, the highest values of maximum amplitude do not reach values for March. The number of five-minute periods with. positive CGs during' the nighttime period increases considerably from March to April (Figure 6c). Concentrations in activity are evident between 0000 and 0100 GMT, 0200 and 0400 GMT, and 0500 and 0900 GMT. During the day, fewer five-minute periods recorded positive lightning strikes in April than in March. The maximum amplitude per five-minute period for positive lightning is considerably ti 11C De li 5t 63 larger in April than in March. Noteworthy is the absence of positive lightning between 1500 and 1600 GMT and 1930 and 2100 GMT. Compared to April, the maximum amplitudes of both posi- tive and negative strikes show a marked increase in May. The maximum amplitudes for negative strikes were somewhat larger during the nighttime period, especially between 0000 and 0200 GMT and between 1000 and 1100 GMT (Figure 7b). During both the nighttime (0000-1200 GMT) and daytime (1200-2400 GMT) pe- riods, positive lightning flashes were recorded in almost all five-minute periods (Figure 7c). Positive lightning was most infrequent between 1500 and 1900 GMT. At the same time, the maximum amplitudes were generally lowest in the afternoon hours. Increased maximum amplitudes are evident between 2100 and 2400 GMT. The June plots reveal surprisingly little diurnal varia- tion in the maximum amplitude of negative strikes (Figure 8b; note the scale difference between May and June). Except for a few extraordinarily high values of amplitude, the five-minute maximum values are less than the May values. Slight increases in amplitude can be observed between 0900 and 1000 GMT, 1300 and 1500 GMT and around 1900 GMT. The diurnal variation is only slightly greater for positive strikes (Figure 8c). The distribution of maximum amplitude per five-minute period for July is similar to that in June. There is very little diurnal variation in the maximum amplitude of negative strikes with only a few exceptionally large amplitudes (Figure 1 plitudes Positive smaller : in maxim can be d1 In . of posit. maximum ' and 1400 crease if ues Of a] the 1830- tive stri for p081 those for rePorted Strong pc 0430 GMT strikeS w lightning 2100 GMT. The qUent duz distinct strikeS 0 64 (Figure 9b). During the night, periods of larger maximum am- plitudes are evident between 0100-0200 GMT and 1000-1100 GMT. Positive strikes are more infrequent than in June and have smaller maximum amplitudes (Figure 9c). No diurnal variation in maximum amplitude of positive cloud-to-ground lightning can be detected. In August some diurnal trends with respect to amplitudes of positive and negative CGs are evident. A period of higher maximum values of negative strikes is apparent between 0400 and 1400 GMT (Figure 10b). After 1400 GMT, there is an in- crease in the number of five-minute periods with maximum val- ues of approximately 50 kA until 2100 GMT. An exception is the 1830-1835 GMT period when the highest amplitude for nega- tive strikes in August was recorded. The maximum amplitudes for positive nighttime strikes are generally larger than those for negative strikes although positive strikes were not reported in all five-minute periods. Three time periods with strong positive amplitudes are evident: 0200-0230 GMT, 0400- 0430 GMT, and 0900-0930 GMT. During the day, positive strikes were recorded in fewer five-minute periods. Positive lightning activity was especially infrequent between 1500 and 2100 GMT. The tendency for positive lightning to be more infre- quent during the day is continued in September. While no distinct diurnal distribution can be detected for negative strikes (Figure 11b), positive strikes are considerably more infrequent and of lower maximum amplitude during the day 65 (Figure 11c). Daytime positive lightning strikes occurred most frequently between 1200 and 1900 GMT with highest ampli- tudes being recorded between 1500 and 1700 GMT. The nighttime distribution reveals an absolute maximum in amplitude shortly after 0000 GMT. The maximum is succeeded by decreasing maxi- mum amplitudes until 0500 GMT, after which five-minute maxi- mum values of amplitude increase again until 0730 GMT. A drop in amplitude after 0730 is followed by higher values starting at approximately 1100 GMT. October displays very little diurnal characteristics of amplitude. For negative strikes, the strongest amplitudes were observed during ‘the daytime period (Figure 12b). Positive strikes were most frequent between 0100 and 0500 GMT (Figure 12c). The negative flashes in November (Figure 13b) show a general increase in both the number of five-minute intervals with recorded lightning activity and the maximum values of amplitude from 0000-1200 GMT, after which the amplitude de- creases again and lightning becomes increasingly less fre- quent. Contrary to that general picture, a few five-minute periods between 2000 and 2400 GMT stand out with relatively high amplitudes. The diurnal curve of negative amplitude values is not duplicated in the frequency distribution of five-minute maxima of positive CGs (Figure 13c). Apart from a slight concentration of positive lightning between 0600 and 1100 GMT, no diurnal trend is apparent. The amplitude of pos- itive lightning during the day is generally lower than during the niq lightni Th terns c strokes flashes Systema mer, fc fied. evident negativ night. flashes tiVe fl Cial ro lihood ' in the 1 66 the night, although the number of occurrences of positive lightning strikes is small in November. Summarx The data set was examined for annual and diurnal pat- terns of five-minute maximum values of amplitude and return strokes. No direct relationship between the peak current of flashes and the number of return strokes was detected. A systematic increase in the number of return strokes in sum- mer, followed by a decrease in the cool season was identi- fied. For maximum values of amplitude, an annual curve was evident as well. The strongest amplitudes were reported for negative flashes in June and July both during the day and at night. However, maximum amplitudes values of positive flashes were more likely to exceed 100 kA than those of nega- tive flashes. With respect to amplitude, May takes on a spe- cial role. For both negative and positive strikes, the like- lihood of maximum amplitudes to exceed 100 kA was higher than in the surrounding months. CHAPTER 4 POLARITY 11 ll] 1 E' 11! . I' The total number of recorded positive lightning strikes during the study period amounted to 4 percent of the number of negative strikes. In general agreement with other studies (e.g. Orville et al. 1987), the percentage of positive strikes was higher in the winter months than in the summer months. NOte in Table 11 the inverse relationship between the total number of events and the percentage of positive strikes. The smaller the total number of strikes per month, the higher the percentage of +CGs. Table 11.--Total Number of Events and Percentages of +CGs 1989 and 1990 L 1989 1990 lfkonth Total Number Positive Total Number Positive of C63 Strikes (%) of C63 Strikes (%) .k-nuary 37 18.92 62 17.74 February -- -- 4 25.00 March 211 8.06 513 8.38 April 1869 4.17 143 16.08 May 2539 4.41 3764 6.43 June 3688 2.14 7148 3.15 July 5192 1.75 4500 2.24 August 5303 3.43 5888 2.50 September 2575 4.16 2882 2.95 October 445 8.54 645 5.58 November 309 8.09 67 68 Note especially the correspondence of a drop in light- ning frequency from March to April 1990 with a considerable increase in the percentage of positive flashes (from 8.38 to 16.08 percent). The inverse relationship between flash frequency and percent positive strikes holds up to a considerable degree on (Table 12). the diurnal scale as well With the exception of January 1989, August 1989, September 1989, and September 1990, the time of day of less lightning activity corresponds with higher percentages of positive strikes. Table 12.--Nighttime/Daytime Total Number of Events and Percentage of Positive Flashes 1989 1990 Nighttime Daytime Nighttime Daytime Month Total +CGs Total +CGs Total +CGs Total +CGs Events (%) Events (%) Events (%) Events (%) Jan 19 21% 18 17% 48 8% 14 17% Feb -- -- -- -- 2 50% 2 50% Mar 146 6% 65 12% 319 5% 194 14% Apr 1304 4% 565 4% 54 20% 89 13% May 1511 4% 1028 5% 2395 6% 1369 8% Jun 1725 2% 1963 2% 4555 2% 2593 5% Jul 2229 2% 2963 2% 2075 2% 2425 2% Aug 2768 4% 2535 3% 3717 2% 2171 3% Sep 1603 5% 972 3% 1633 3% 1249 2% Oct 217 8% 228 9% 418 5% 227 6% Nov 163 7% 146 9% The percentage of positive strikes in 1989 decreased from 21 percent at night and 17 percent during the day in January to 2 percent in the summer months for both nighttime and daytime period and then decreased again toward winter. 1990 also witnessed a general decrease in the percentage of positive strikes in the first half of the year with a 69 subsequent increase in autumn. As mentioned previously, how- ever, the percentage of positive strikes at night shows a considerable increase in April after lower values in March. The percentage decreased again in May and remained 55 percent into October. The percentages for January and February are unreliable due to the small number of events in those two months. A more systematic annual trend is evident for day- time positive strikes in 1990 with the lowest percentages ev- ident in June through August. September shows a slight drop in the percentage of positive strikes. The inverse relation- ship between lightning frequency and positive strikes, espe- cially in the summer months, still needs further clarifica- tion. A possible reason for the relative increase in positive strikes in months with fewer events may be found in cloud characteristics. In the case of bipoles, negative strikes are believed to predominate in proximity to the deepest con- vection whereas positive strikes are likely found in strati- form precipitation regions (Engholm et al. 1990). Stolzenburg (1990) and Orville et al. (1988) found a higher occurrence of storms with bipolar patterns relative to the total number of storms in the winter months. In the winter months, the wind shear of storms is often sufficient to produce the necessary horizontal displacement between the center of positive charge and the negative charge center to generate a significant amount of positive ground flashes. A small scale investiga- tion of the charge patterns during periods of increased 70 numbers of positive strikes might provide some answers to the observed phenomenon. A closer look at the temporal distribution of positive strikes in Table 13 reveals high monthly and annual variabil- ity in terms of the time of day of occurrence. The percent- ages of nighttime events among the positive strikes range from 45 to 72 percent in 1989 and from 35 to 65 percent in 1990. Thus the percentages of positive strikes occurring at night were generally higher in 1989 than in 1990. In 1989, January through April had a relative nighttime maximum of positive ground flashes. After May, a transition month with an equally high occurrence of positive strikes during the day and at night, most positive strikes were recorded during the daytime period in June and July. The relative maximum then shifted to the night again where it remained from August through September. Inn October and November, more positive strikes were recorded during the day than at night. Unlike 1989, early 1990 witnessed a relative daytime maximum of pos- itive strikes from January through April. It was followed by a nighttime maximum in May. Relative daytime maxima in June and July were followed by relative nighttime maxima in August and September. The daytime maximum evident during the cool months of 1989 was not present during 1990. October, for ex- ample, recorded a nighttime maximum. Due to lack of data, November 1990 could not be included in this comparison. 71 Table 13.--Nighttime (0000-1200 GMT) and Daytime (1200-2400 GMT) Occurrence of Positive Strikes. 1989 1990 Month Nighttime Daytime Nighttime Daytime Jan 4 (57%) 3 (43%) 4 (36%) 7 (64%) Feb - -- -- -- 1 (100%) 0 (0%) Mar 9 (53%) 8 (47%) 15 (35%) 28 (65%) Apr 54 (69%) 24 (31%) 11 (48%) 12 (52%) May 56 (50%) 56 (50%) 139 (57%) 103 (43%) Jun 39 (49%) 41 (51%) 105 (47%) 120 (53%) Jul 44 (48%) 48 (52%) 46 (46%) 55 (54%) Aug 112 (62%) 70 (38%) 82 (56%) 65 (44%) Sep 77 (72%) 30 (28%) 55 (65%) 30 (35%) Oct 17 (45%) 21 (55%) 22 (61%) 14 (39%) Nov 12 (48%) 13 (52%) E J 'I . E J I' | E I SI 1 i E ].| i The average number of return strokes is higher for nega- tiVe strikes than for positive strikes (Tables 3-4). Note also that the maximum number of return strokes of -CGs is considerably higher than that of +CGs. (Table 5). The maxi- mum number of return strokes for both positive and negative flashes was larger for summertime lightning flashes compared to winter flashes. The percentage of events with only one return stroke is largest in winter for both -CG and +CG flashes. The interrelationships reflected here between polarity and mean flash amplitude support Orville et al's (1987) find- ings. Orville et al. reported a median amplitude of 30 kA for negative strikes and 45kA for positive flashes. As shown in Table 7, the mean amplitude of +CG strikes is considerably higher than that of -CG flashes. However, while the mean am- plitude values are highest in the winter months for both pos- itive and negative flashes, the maximum values of amplitude 72 per five-minute period occurred in June and July for negative flashes. Amplitudes >300kA were recorded for negative flashes during both the daytime and nighttime period in June and July (Tables 8 and 9). The maximum values for positive strikes are considerably lower. The highest values of ampli- tude (200-250kA) were recorded during the nighttime period in March, May, June, August and September and during the day in May and June. However, except during warm months (nighttime period in June and July and daytime period in July and August), the percentage of five-minute periods with maximum amplitude values 2100kA was greater for positive strikes than for negative strikes (Table 10). Summarx The data set was examined for temporal variations in the distribution of the frequency and characteristics of positive and negative strikes. Contrary to other studies, the largest amplitudes were reported in association with negative strikes in June and July. Also, positive strikes did not display a general increase in maximum values of amplitude in winter as expected. The percentage of five-minute periods with maximum amplitude values 2100kA was greater for positive strikes than for negative flashes. An additional interesting finding was the inverse rela- tionship between the number of positive strikes and the total number of ground flashes on both a diurnal and annual scale. The relationship between cloud characteristics and flash 73 polarity was suggested as a possible cause for the relative increase in positive strikes during periods with less light- ning activity. PART TWO SPATIAL ANALYSIS 74 CHAPTER 5 SPATIAL ANALYSIS OF LIGHTNING ACTIVITY: PROCEDURES AND METHODOLOGY 3 . I' The major objective of the second part of the study was the analysis of the spatial distribution of lightning strikes with respect to time of day. A combination of spatial and temporal analyses was chosen to achieve this objective. Since this study was designed as a preliminary examination of a SUNYA lightning data set with the overall goal of interre- lating the timing and locations of lightning activity to pre- cipitation activity in the study region, the spatial compo- nent of the analysis was structured in a way to allow for the comparison of precipitation and lightning data. A grid sys- tem commonly used for radar data of precipitation was thus chosen for the analysis (MDR grid). The special location of the study area in the transition zone between nocturnal and afternoon regimes of precipitation and thunderstorm activity has previously been noted. To account for the different di- urnal characteristics, the study area was divided into two equal halves and frequency curves of lightning activity were calculated separately for the eastern and western portions of the study area. 75 76 A second objective was incorporated in the spatial- temporal analysis: the distinction of lightning events by polarity. While the spatial discrimination of positive and negative cloud-to-ground lightning strikes were not under- taken in this thesis, a distinction with respect to polarity was made in the hourly frequency distributions. A final objective was the development and testing of computer software for spatial analysis and graphic output of lightning observations . Was—Grid The data set for the previously-discussed temporal anal- ysis of lightning activity in the first part of the study in- cluded all observations located in an area bounded by 38.3°N and 47.1°N and by 78.9°W and 96.7°W. However, for the spa- tial analysis of lightning activity, observations from only a subset of this area were utilized. The primary motivation for reducing the analysis area was to better coordinate this PrOject with a second project concerning the spatial varia- tions of convection in the Great Lakes region currently un- derWay in the Department of Geography at Michigan State university. This second study is employing manually digitized radar (MDR) intensities to characterize convection. MDR reports are for grid cells within a regular rectangular matrix super- ix“FOSed on a polar stereographic projection. For the conter- manUS United States the matrix extends 113 cells in an 77 east-west direction and runs 89 cells in a north-south. Grid spacing is 47.625 km at 60°N latitude. The principal inves- tigators chose a 32x19 cell subgrid for the analysis of radar observations in the Great Lakes region. This subgrid is com- posed of east-west cells 57 through 88 and south cells 48 through 66 from the larger MDR grid. This same subgrid is used for the present study of lightning activity in order to share analysis and plotting software. A future goal, al- though not undertaken in this thesis, is the study of the re- lationship between lightning activity and radar echo strength in the Great Lakes region. Lightning locations in latitude and longitude were translated to the MDR grid cells using software programs pro- vided by the National Meteorological Center. All strikes that fell into a particular grid cell were assigned the X and Y coordinates for that cell. The center of the grid cell was taken as the reference point when assigning lightning obser- Vat ions rather than the grid intersection points. The relative size of the data set used for the spatial analysis of lightning activity compared to the data set used for: the first part of the study is shown for each month in Figure 14. 78 Thousands a ,. i ii ,, 5 .z ..................... J ’ 7 Doutsid. Grid 4 ’ """" ‘ inside Grid .Q § 1. . ‘7" .u.-.. n... 1‘ .N! Lwn‘b‘buu r4 ' ~‘I * r I ‘ 9 .-’l J v F‘. - x — V‘-. :: Fig. 14. Monthly Distribution of Strikes inside and outside the 32-by-19 MDR grid. 3 J 1 l° E E 'I C 1] E . i H I The calculation of lightning frequency and density per grid cell for subsequent display was performed using FORTRAN programs (see Appendix for example programs). Atlas Mapmaker was employed in the analysis of grid frequencies and to map grid cell values. The grid basemap and the state outlines for the following figures were entered as boundary files in Atlas Mapmaker, into which the frequency values for each cell were read” Ln order to make the following figures easily comparable, the numerical ranges for categories on each of the monthly maps were standardized. Due to the low frequency of lightning strikes in the Winter and several of the transitional months, only months with "increased lightning activity", as previously defined 79 (p. 23), were plotted. These months include April through September of 1989 and May through September 1990. I J 3 | E I] S I' J E J . To examine the diurnal characteristics of the region, daytime hours (1200-2400 GMT) and nighttime hours (0000-1200 GMT) were analyzed separately. In order to investigate any differences in the frequency and timing of lightning activity in the eastern and western parts of the study area, the study area was divided into two equal halves. Daytime and night- time hourly frequencies were calculated separately for both regions with FORTRAN programs. For graphic output Harvard Graphics was employed. The wide range of frequency values over the course of the examined months required some adjust- ment of the vertical scale. Note that the vertical scale is 100 strikes per hour in April, 160 flashes in May and September; June through August have vertical scales of 300 strikes per hour. CHAPTER 6 SPATIAL ANALYSIS OF LIGHTNING ACTIVITY: RESULTS In general, lightning activity decreases from south to north across the MDR subgrid. Note in Figure 15 that all grid cells except for a few along the extreme edges of the analysis area reported lightning strikes from January 1989 to October 1990. The lack of data in the extreme corners of the analysis area may indicate that these MDR grid cells lie outside the effective range of the SUNYA lightning detection network. Numbot oi Strikes 1 lo 14 1510 29 30 lo 44 45 to 59 60 lo 84 85(0109 110(0139 140l0173 Illlflfiflfl Fig. 15. Total Number of Cloud-to-Ground Strikes for the Study Period 80 81 Two axes of somewhat increased lightning activity can be observed. The first is oriented in southwest-northeast di- rection and extends from Iowa into lower Michigan. The sec- ond has a similar orientation and extends from southern Indiana into Ohio. The highest cumulative lightning frequen- cies with 140 strikes or more per cell during the study pe- riod are found in southwestern Iowa, along the Indiana- Illinois border, and in Ohio. The highest number of strikes that was reported for a single cell was 178. H Ii] M . I' . E I' i N' ill' I' ll . E l' W The spatial analysis by month was conducted for night- time (0000-1200 GMT) and daytime (1200-2400 GMT) hours sepa- rately. For each monthly nighttime/daytime period, the cumu- lative number of lightning events and the number of lightning days per grid cell were plotted. In addition, the spatial analysis is augmented by the calculation of hourly frequen- cies of ground strikes accumulated over the entire month for the eastern and western halves of the study region. Variations in the timing and frequency of lightning activity in the eastern and western portions of the study area are ev- ident. These variations are presented in detail in the fol- lowing discussion. 82 April 1989 In April 1989, a considerably higher number of strikes was recorded during the night (0000-1200 GMT) than during the day over the entire study area (1075 flashes compared to 396 strikes). Note in Figure 16a that the flash density during the night is also larger than the daytime equivalent (Figure 17a). During the night, a maximum of 34 CG strikes per grid cell is found along the Ohio/Indiana border embedded in a band of relatively high frequencies stretching from Illinois to Ohio (Figure 16a). A secondary frequency maximum can be identified from the southern portion of Lake Michigan through southern Michigan to Ohio. In general, lightning activity was infrequent during all daytime periods of April 1989. Only 396 strikes were re- ported for the grid system and the maximum number of light- ning events per individual grid cell was fourteen strikes. As Figure 17a shows, the area of maximum frequency extends from Ohio into West Virginia. A secondary frequency maximum can be observed in Iowa and Minnesota. Similar distinct max- ima are not evident in the plot of lightning days in April 1989 (Figures 17b) suggesting that the patterns of daytime lightning events likely result from only one or two storm events with fairly large lightning densities. (Tr in other words, the poor correspondence between the number of strikes per grid cell and the number of lightning days suggests that centers of daytime lightning frequency in April 1989 probably 83 Number of Strikes to4 109 01017 8(028 to45 IIINU ‘ a (n .e 29 Fig. 16a. Nighttime (0000 - 1200 GMT) Lightning Strikes. April 1989. Number 01 Days IIIEU outrun- 2 4 6 8 1 88888 0 Fig. 16b. Number 01 Days with Nighttime Lightning Activity, April 1989. 84 Number 01 Strikes 1t04 5109 101017 181028 291045 IEIND Fig. 1711. Daytime (1200 - 2400 GMT) Lightning Strikes. April 1939. Number of Days 8888 IIIND ONUIW‘ 4000‘” a-e 0 Fig. 17b. Number 01 Days with Daytime Lightning Activity. April 1989. 85 result from single events with relatively high lightning ac- tivity. Differences in the diurnal phasing of convection during April 1989 are evident in the hourly frequency totals for the western and eastern portions of the study area. In the west (Figure 18a), lightning activity increased fairly steadily throughout the day and into the night. The frequency maximum was recorded at 0500 GMT, after which the curve shows a sharp decline in lightning frequency. The east (Figure 18b), on the other hand, shows decreasing lightning frequencies over the course of the day after fairly high frequencies during most hours of the night. The highest hourly values, also at 0500 GMT, are also preceded by an increase in lightning frequency. An additional interesting, but as yet difficult to ex- plain, observation is that during April 1989 positive light- ning flashes were more common during the night, especially in the western portion of the study area. May 1989 In comparison to April, the lightning events of May 1989, both nighttime and daytime (Figures 19a and 20a), are more frequent farther north in the study area. As for April, nighttime activity in May 1989 far exceeds daytime activity. Between 0000 and 1200 GMT, 1337 CG flashes were reported. High frequencies are found in a triangular region extending from Wisconsin through northeastern Ohio to the southern edge 86 DPoeltlve Strikes ‘0 _______________________ Negative Strikes o ‘ . 121314 15 13171319 20 2122 23 Time (6M1) Time (GMT) Fig. 18a. Hourly Flash Frequencies for the Western Portion of the Study Area, April 1989. I I .1 it ' DPositive Strikes 21.1.1 ENegetive Strikes .. . (mu. t i :i L ”a h :i , .01 L. d“ 4.1: 121314161317131920212223 T|me(GMT) Fig. 18b. Hourly Flash Frequencies for the Eastern Portion of the Study Area, April 1989. 87 May 1989 In comparison to April, the lightning events of May 1989, both nighttime and daytime (Figures 19a and 20a), are more frequent farther north in the study area. As for April, nighttime activity in May 1989 far exceeds daytime activity. Between 0000 and 1200 GMT, 1337 CG flashes were reported. High frequencies are found in a triangular region extending from Wisconsin through northeastern Ohio to the southern edge of Indiana (Figure 19a). In addition, a few grid cells in northwestern Iowa reported higher values than the surrounding areas. Despite the fact that almost twice as many ground strikes were recorded during the night than during the day, the maximum frequency was only slightly higher with 18 CGs per cell as compared to 14 flashes per cell during the day. The different lightning density values suggest that nighttime activity extended over a broader area than daytime activity. There is only a small correspondence between the number of nighttime CG flashes per grid cell and the number of light- ning days (Figure 19b). Within the above-described triangle of highest frequency, only a few cells from eastern Wisconsin and western Michigan reported lightning activity on three or more days. For the remaining grid cells, including north- western Iowa where a comparatively high number of events was recorded, lightning occurred on only one or two days during the month. On the other hand, in southwestern Illinois and across Lake Erie, the number of lightning events was small in spite of a relatively larger number of lightning days. 88 Number of Strikes Fig. 193. Nighttime (0000 - 1200 GMT) Lightning Strikes. May 1989. Number oi Days IIINU «cumu- 88888 2 4 6 8 1 O Fig. 19b. Number of Days with Nighttime Lightning Activity, May 1989. 89 As Figure 20a shows, the 766 recorded daytime ground flashes have their main concentration in a triangle extending from southern Lake Michigan through southern Michigan to Indiana. A secondary concentration is found in southeastern Minnesota and central Wisconsin. Overall, more grid cells reported daytime ground strikes in May than in April although the typical number of strikes per cell remains relatively low (1-4 CG flashes). Again, the distribution of lightning events is not reflected in the distribution of lightning days (Figure 20b). Most grid cells within the area of main con- centration from Michigan to Indiana reported lightning on only one or two days in May except for one cell along the Michigan-Indiana border and another one in Indiana along Lake Michigan, where daytime lightning activity was recorded on more than three days. Some differences in daytime flash den- sity are evident in the western and eastern portions of the study area. In the east, most grid cells recorded lightning on only one or two days, whereas cells in Wisconsin and along the Wisconsin-Minnesota border also had a larger number of lightning days (3 to 4). As in April, the hourly frequency distributions vary considerably for the eastern and western portions of the study area. In the west (Figure 21a), the diurnal curve of lightning frequency shows similarities to the April 1989 curve, as lightning activity increases throughout the day to a peak at 0500 GMT. In contrast to the fairly steady in- crease in frequency during the afternoon hours evident in 90 Fig. 20a. Daytime (1200 - 2400 GMT) Lightning Strikes. May 1989. Fig. 20b. Number of Days with Daytime Lightning Activity. May 1989. Number of Strikes IIIND E «sumo- a. .. Lauren) 9. i? '3 88888 0 91 April, hourly frequency values in May decrease between 2100 and 0000 GMT. The sharp decline after 0500 GMT is even more pronounced than in April. In the east (Figure 21b), the curves for April and May are very different. While fairly high hourly values are evident throughout the nighttime pe- riod in April (note the difference in the vertical scale be- tween April and May), a distinct frequency maximum in May at 0100 GMT is followed by a steady decline until 1800 GMT with very low (less than ten strikes) hourly frequencies between 1200 and 1700 GMT. After 1800 GMT the hourly frequencies in- crease sharply. The positive and negative lightning strikes frequencies do not reveal any regional or temporal prefer- ences . 92 CI Positive Strikes Negative Strikes ‘ .._ ‘,;:_ .1 g ., . f ‘i ‘ . 0 . . .- . . 0 . 01234537381011 121314151317131920212223 Time(GMT) TIme(GMT) Fig. 21a. Hourly Flash Frequencies for the Western Portion of the Study Area, May 1989. UPaMnflMu Negetlve Strikes . . . .' .. . 0 . , 01 234537391011 121314151317131920212223 nmomun nmomun Fig. 21b. Hourly Flash Frequencies for the Eastern Portion of the Study Area, May 1989. 93 June 1989 Compared to April and May, lightning strikes for the en- tire study area during June 1989 were more frequent during the day than at night. The fewer nighttime strikes (1075 to- tal events) are concentrated along the Iowa-Nebraska border and in a band stretching from the Iowa-Missouri border into southwest Ontario (Figure 22a). The maximum frequency per grid cell (34) is found in northern Missouri. This value is considerably higher than the daytime maximum value (21 strikes) despite the larger total number of strikes during the day. Except for small areas in Illinois and western Iowa, nighttime lightning activity occurred on only one or two days during June 1989 (Figure 22b). This suggests that the cumulative flash totals resulted from one or two major storm systems tracking through the area, as was also evident in April and May. Some correspondence between lightning fre- quencies and the number of lightning days per cell is evident in the southwest corner of the study area. However, while the maximum number of events is clustered in the center of that region, the maximum number of lightning days is located near the periphery. The southwest-northeast oriented axis of increased lightning activity from Iowa to Ontario is not mir- rored in the spatial distribution of the number of lightning days. In June 1989, the large increase in daytime lightning strikes (1541 total events) compared to April and May is ac- companied by a distinct concentration of a higher number of 94 Number of Strikes Fig. 22a. Nighttime (0000 - 1200 GMT) Lightning Strikes. June 1989. Number of Days IIINE] outnu- 2 4 6 8 1 88888 0 Fig. 22b. Number 01 Days with Nighttime Lightning Activity June. 1989. 95 strikes in the eastern half of the study area (Figure 23a). This concentration of ground flashes in the eastern half of the study area is in general agreement with studies by Wallace (1975) and Winkler (1987) who reported afternoon max- ima of precipitation and thunderstorm activity for the east- ern United States and a nocturnal maximum for the central part of the country. Compared to the two previous months, a better correlation between regions of increased daytime lightning activity and regions with higher number of light- ning days (Figure 23b) is evident in June 1989. Note that both, the highest frequencies and number of lightning days, are found in the eastern part of the study area suggesting that the significantly higher frequency of lightning events in the eastern portion of the study area is a result of the larger number of days on which lightning occurred rather than the result of a few, larger convective systems with unusually high lightning activity. The hourly flash frequencies for the western portion of the study region show an increase in lightning activity from 2000 GMT to a maximum at 0300 GMT (Figure 24a) after which the hourly frequencies decrease again (note the scale change between May and June). As in May, the eastern portion of the study region (Figure 24b) displays a strong increase in hourly frequencies in the early afternoon hours. Compared to May, however, the increase in hourly values starts two hours earlier in June (1700 GMT compared to 1900 GMT) and ends fairly abruptly at 2200 GMT, whereas the curve in May 96 Number of Strikes Fig. 23a. Daytime (1200 - 2400 GMT) Lightning Strikes, June 1989. Number 01 Days ..INU outrun- 88888 2 4 6 8 10 Fig. 23b. Number 01 Days with Daytime Lightning Activity, June 1989. 150 Fig. 24a. ........................ ........................ 11me (GMT) 97 250 150 100 . .1 g “a . . :3. o.J£WkafimuflfihihfikJ‘ ........................ ------------------------ eeeeeeeeeeeeeeeeeeeeeee 121314151317131920212223 Time (GMT) of the Study Area, June 1989. 150 Fig. 24b. ........................ ........................ eeeeeeeeeeeeeeeeeeeeeeeee ‘0 0 1011 Tlme(GMT) 300 250 ........................ 2m .................... F; ii? ‘50 ............ F: 3;: - 1.. .z: A: 1‘1 .' ' ’ 100 ----------- f. l ‘3 3: so -a ------- g . r5 _ ‘ .; .- J1. ‘.‘ .I OEQIK'HWI it?! 3;; M? ‘6; ’ D Positive Strikes 8 Negative Strikes Hourly Flash Frequencies for the Western Portion Cl Positive Strikes EMMnmnmmMi 12 13 14 15 1617 1a 10 20 21 Time (GMT) of the Study Area, June 1989. Hourly Flash Frequencies for the Eastern Portion 98 increases until a maximum at 0100 GMT and then gradually de- clines. No diurnal trend in the frequency of positive light- ning flashes in either region is evident in Figures 24a and 24b. July 1989 As for June, most ground flashes in July 1989 occurred during the day. During the nighttime period (0000-1200 GMT), the west received a slightly higher number of events (720 CGs) than the east (714 CGs). Three distinct clusters of in- creased lightning activity, western Iowa, southern Lake Michigan, and western Lake Erie, are evident in Figure 25a. Especially intriguing are the clusters over the lakes, as this activity is occurring over the summertime cooler lake waters. Although there was some indication of frequency max- ima in these areas on the June plot, the maxima are more dis- tinct and include a larger number of cells in July. The recorded events during the daytime period in Ju1y 1989 also support the findings by Wallace (1975) and Winkler (1987). Note in Figure 26a the significantly higher number of lightning events in the eastern portion of the study area (1515 flashes) than in the west (543 events). However, ele- vated daytime lightning activity of 5-9 strikes per cell is also found in the western region from southern Iowa to east- ern Illinois. In general, a west to east axis of maximum ac- tivity from eastern Iowa to southeast Ohio is evident. In Lower Michigan, fewer flashes were reported in July 1989 99 Number of Strikes IIIED Fig. 258. Nighttime (0000 - 1200 GMT) Lightning Strikes. July 1989. Nunber of Days IIIND require- 88888 duos» 0 Fig. 25b. Number of Days with Nighttime Lightning Activity. July 1989. 100 Number of Strikes 1to4 5t09 10 to 17 18 to 28 29to 45 IIIND Fig. 26a. Daytime (1200 - 2400 GMT) Lightning Strikes, July 1989. Number of Days IIIED ONUIID-i 2 4 6 8 1 88888 0 Fig. 26b. Number 01 Days with Daytime Lightning Activity, July 1989. 101 compared to June, suggesting a more southerly storm track for July 1989. As for June, daytime lightning activity was re- ported on a significantly larger number of days in the east- ern portion of the study region, particularly in southeastern Ohio, compared to the western portion (Figure 26b). The much larger number of events in the eastern region, and especially during the day, is clearly evident in Figures 27a and 27b. The hourly plots show a much smaller amplitude in the frequency curve for the western portion (Figure 27a) of the study region than for the east. Lower values in the morning hours (1300-1800 GMT) are followed by fairly steady frequencies in the afternoon and throughout the night (1900- 1200 GMT). Despite the lower total number of strikes, the plot of positive lightning strikes reveals a slight prefer- ence for the western part of the study region. In the east (Figure 27b), on the other hand, the ten- dency for a steep increase in hourly values in the afternoon hours, apparent in May and June, is continued in July. Starting an: 1700 GMT, hourly frequencies sharply increase until 1900 GMT when they level off. The decrease is pro- longed one hour compared to June, and is most pronounced be- tween 2300 and 0200 GMT. 102 ‘ so ________________________ El Positive Strikes E] Negative Strikes 121314151317 13 19 20 2122 23 Time (GMT) Time (GMT) Fig. 27a. Hourly Flash Frequencies for the Western Portion of the Study Area, July 1989. 25° ........................ 25° ........................ m ........................ 2m .............. — 2." ..' a El Positive Strikes 150 -------------- . g ’ 1 1: NegatNe Strikes 100 ------------ 1‘ run— I. o , . 4 V 121314151317131920212223 nmomMn Fig. 27b. Hourly Flash Frequencies for the Eastern Portion of the Study Area, July 1989. 103 August 1989 Fewer nighttime (1358 C65) than daytime strikes (1813 CGs) were also reported during August 1989. Little regional differentiation in the frequency of lighting activity is evi— dent for the nighttime hours (Figure 28a). Note that several spatially dispersed clusters of maximum activity are evident, including southern Iowa, southeastern Ohio, and southern Lake Michigan. Missing are the clusters of enhanced activity in western Lake Erie and southwestern Iowa evident in the two previous months. The only cells with five or six nighttime lightning days in August 1989 are found in southern Iowa (Figure 28b). This area corresponds to a main cluster of lightning activity. The other clusters of higher nighttime lightning frequencies are not reflected in the distribution of lightning days. During the daytime period, the spatial concentration of ground flashes shows a clear preference for the eastern half of the grid system (Figure 29a). As for July 1989, higher daytime lightning activity in August is concentrated along a southwest-northeast oriented axis, although this axis is lo- cated farther south (from southeastern Illinois to southern Ohio) compared to July. Both June and July had daytime max- ima in lightning frequency, but the highest cumulative strike frequencies per grid cell were recorded during the nighttime period, whereas in August maximum frequencies per cell oc- curred during the daytime period. The maximum cell frequency value of 36 events exceeds those for June and July. It 104 Number of Strikes Fig. 28a. Nighttime (0000 - 1200 GMT) Lightning Strikes. August 1989. Number 01 Days IIIED muons- 88888 Human Fig. 28b. Number of Days with Nighttime Lightning Activity, August 1989. 105 Fig. 29a. Daytime (1200 - 2400 GMT) Lightning Strikes, August 1989. Number of Strikes Number 01 Days IIIEU ONUIU-I d00‘M 88888 c Fig. 29b. Number of Days with Daytime Lightning Activity, August 1989. 106 approximates the nighttime values for the preceding two months (36 and 37 flashes per cell, respectively). However, the increase in daytime lightning frequency in August as com- pared to July was not accompanied by a significant increase in the number of lightning days (Figure 29b). Most cells fall in category II (3-4 lightning days per cell), a few fall in category III (5-6 lightning days per cell), and one cell recorded lightning on seven or more days. Considerably more cells in July were assigned to category III. The axis of maximum daytime lightning frequency in the southern and southeastern portions of the study area generally experienced five to six lightning days in August 1989. The hourly frequency curves for both the western and eastern portions of the study area resemble the June (Figures 24a and 24b) rather than the July (Figures 27a and 27b) dis- tribution . Similar features in the western curve (Figure 30a) include a steady increase in hourly frequency toward a nocturnal peak that is followed by a decrease in hourly val- ues to a udnimum around noon. While the June maximum oc- curred at 0300 GMT, the August maximum was delayed by four hours. The hourly frequency distribution for the eastern portion of the study area (Figure 30b) again shows the strong increase in lightning activity in the afternoon hours that was observed in May through July. Hourly frequency decrease beginning at approximately 2300 GMT. An interesting feature concerning’ the jpolarity' of lightning strikes is evident in the west, especially in the 150 100 Fig. 30a. ........................ eeeeeeeeeeeeeeeeeeeeeeee eeeeeee 250 150 100 107 ........................ ........................ ........................ ; _ g H} . e r . .“ ‘-:‘ l 'I h > “. ‘ 7:? 9* Pal-x] >51 " in. r‘ ‘~ [3 Positive Strikes El Negative Strikes 12 13 14 15 13 17 18 19 20 21 22 23 nmemMn of the Study Area, August 1989. 150 Fig. 30b. eeeeeeeeeeeeeeeeeeeeeeee ........................ e o n e e n .................. ego-pouauo-eooce- ....... . ',.‘ . 1. ~,- _ ' ".1 ‘3 11 is" “'0! ti;- . 3 91011 Time (GMT) 2w ........................ 2m ................... : - . .‘i: I": ‘1 — If? , '7: c - 150 -------------- . .1 , 7‘. i" ‘. {1 '97 -r . I Z '1" e 100 ------------ z. . 4 fl \ r ’t i 3'? 1 0 . ‘ 4 K a ; :4 l P 1. m . ...... , ' g I L" '4 ‘, '- 1 . ‘ , .i .1 s H ‘ Q‘- ‘ ~“ - t t "‘74 H 4'8 . L ‘ . ,. . ‘ _ ‘ . ‘ A“ o ‘4 11:1 2.1 . F {‘42 ‘ ,... 71 in: ,1, g. '1 Hourly Flash Frequencies for the Western Portion ilfimmemMu Negative Strikes 12 13 14 15 13 17 13 19 20 21 22 23 nmemun of the Study Area, August 1989. Hourly Flash Frequencies for the Eastern Portion 108 evening hours. Despite the lower hourly frequency values in the western portion of the study area as compared to the east, the percentage of positive strikes is considerably higher. Thus, the prevalence of positive strikes in the west that had already been observed. in July is even more pronounced in August. September 1989 After three months with daytime maxima in lightning ac- tivity, in September 1989 the nighttime lightning events (1305 total flashes) exceed the daytime events. This shift in temporal preference is a result of a sharp decline in the number of daytime events (645 compared to 1813 in August). The higher nighttime totals are accompanied by a higher maximum frequencies per cell (Figure 31a). While a maximum of 33 CG strikes per cell was recorded for nighttime activity in September, the equivalent figure for daytime activity is 18 CG strikes per cell. Nighttime lightning activity shows a clear concentration in the western half of the study area (1032 strikes compared to 273 in the east). Clusters of higher lightning activity in the west can be identified along the Iowa-Nebraska border, in northeast Iowa, and in central Illinois. Two of these spatial clusters, Iowa-Nebraska and central Illinois, are also evident in the distribution of lightning days (Figure 31b). In Iowa, on the other hand, strikes were recorded on only two days during all the night- time intervals in the month of September. Weaker frequency Number of Strikes Fig. 31a. Nighttime (0000 - 1200 GMT) Lightning Strikes. September 1989. Number of Deye IIIEE] 04010-1) 33333 HOODAM Fig.31b. Number of Days with Nighttime Lightning Activity, September 1989. 110 maxima in both, the number of lightning events and lightning days, are evident over southern Lake Michigan and eastern Lake Erie. The dramatic decrease in the frequency of daytime events from August to September is accompanied by a more uniform distribution of lightning strikes across the study area com— pared to the previous months, although there is a slight preference for the eastern half of the study area. Apart from a slight concentration extending from Illinois into Indiana, the daytime spatial distribution does not reveal any trends (Figure 32a). In regards to the number of lightning days, the cluster in Illinois-Indiana is duplicated only by a fraction of the affected cells (Figure 32b). Four grid cells along the Illinois-Indiana border are the only cells that re- ported 5-9 CG strikes The hourly frequency curves for the eastern and western portions of the study area differ substantially. While the hourly frequency of CG flashes was fairly uniformly dis- tributed throughout the daytime hours in the western region with approximately 30 CGs per hour, lightning activity was almost three times more frequent in the nighttime period be- tween 0300 and 0800 GMT (Figure 33a). In the eastern por- tion, maximum activity occurred in a limited period during the afternoon and early evening hours from 2000 GMT to 2300 GMT (Figure 33b). In both regions the shapes of the diurnal curves resemble those for August, in spite of the large dif- ference in strike frequency (note the scale difference 111 Number 0! Strikes 1to4 5t09 10 to 17 18 to 28 29to 45 IIIED Fig. 328. Daytime (1200 - 2400 GMT) Lightning Strikes, September 1989. Number 0! Days IIIED GDNOIu-e to 2 to 4 to 6 to 8 to 10 Fig. 32b. Number oi Days with Daytime Lightning Activity, September 1989. 112 between August and September). While the daytime maximum in the eastern portion of the study area stands out in August, it is the nighttime maximum in the west that is more apparent in September. The nighttime lightning events in the west are also characterized by a relatively high percentage of posi- tive strikes. Unlike the two previous months, the nighttime period is not the period of small lightning frequencies. Thus the previously proposed inverse relationship between the total number of strikes and the number of positive strikes is not evident in September. It rather seems that there may ex- ist a temporal and spatial preference for positive lightning strikes. The tendency of the largest number of positive strikes to be recorded in western portion of the study area, mainly during the nighttime period, can also be observed in 1990 as will be seen in the following discussion. 150 ........... too ------ :1 Fig. 33a. eeeeeeeeeeee 113 Ci Positive Strikes O Negative Strikes .00 ........................ jfl it flflfl iii—Tina 12 13 14 15 16 17 18 19 20 21 22 23 Time (GMT) Hourly Flash Frequencies for the Western Portion of the Study Area, September 1989. ‘50 ........................ ‘m ........................ nt‘ ’- 1r g ..,- 13.. ,~ V 1' O 0112 3165 e 7 hmemun Fig. 33b. A [- t». g" i“ 1m ................ ‘m .I ................ w ................ 7 H "i f" ' 01JTWJFTIEHFT' ”(‘1 eeeeeeee 121314151617181920212223 Time (GMT) El Positive Strikes E Negative Strikes Hourly Flash Frequencies for the Eastern Portion of the Study Area, September 1989. 114 May 1990 The large amount of interannual variability in storm tracks and the possible impact of high activity from single systems on the frequency of lightning strikes are evident when comparing the May 1989 and May 1990 plots of nighttime and daytime activity. Both nighttime and daytime lightning frequencies for May 1990 (Figures 34 and 35) are concentrated along a southwest-northeast axis from Nebraska into southern Lake Michigan and Indiana, whereas in May 1989 (Figures 19 and 20) nighttime events were most frequent in Michigan and Indiana, and daytime events were most frequent in Wisconsin, southwest Michigan and northern Indiana. Lightning activity in May 1990 was much more frequent at night (1530 occurrences) than during the day (723 events). Also, the largest strike frequency per cell is considerably higher for the nighttime period (29 strikes) than daytime pe- riod (10 strikes). This axis of lightning activity during the nighttime period is better observed in the distribution of lightning days (Figure 34b) than in the distribution of lightning events (Figure 34a). The larger number of night- time lightning occurrences in May 1990 is accompanied by a larger number of lightning days. For example, the region of frequent nighttime lightning activity along the Nebraska-Iowa border reported ground flashes on up to eight days. Two main concentrations of daytime lightning activity (Figure 35a) and lightning days (Figure 35b) in southwestern Iowa and in northwestern Illinois can be identified within 115 Fig. 348. Nighttime (0000 - 1200 GMT) Lightning Strikes. May 1990. Number of Strikes Number of Days IIINEI @NUIG-fi .eDO‘N 38838 c Fig. 34b. Number of Days with Nighttime Lightning Activity. May 1990. Number oi Strikes Fig. 35a. Daytime (1200 - 2400 GMT) Lightning Strikes, May 1990. Z c 5 ¥ 9.. U m ‘3 IIIN! Fig. 35b. Number of Days with Daytime Lightning Activity, May 1990. 117 the band extending from Nebraska into southern Lake Michigan and Indiana. An additional area of frequent activity, appar- ent only on the plot of lightning events is found in southern Indiana and southwestern Ohio. Here CG strokes were reported on only one to two days, unlike the two other centers of lightning activity that reported daytime lightning activity on up to six days in May 1990. The prevalence of nighttime events is particularly evi- dent in the western portion of the study area where over 120 events were reported between 0400 and 0500 GMT (Figure 36a). The hourly frequency distribution in the west is character— ized by a minimum in lightning activity between 1500 and 1800 GMT followed by a steady increase in activity until 0400 GMT and a subsequent decrease. In the eastern portion of the study area (Figure 36b) the number of events is fairly uni- formly distributed throughout the daytime hours. The night- time period, on the other hand, shows three peaks from 2300- 0100 GMT, 0500-0700 GMT, and 1000-1100 GMT. Not only does the spatial distribution vary between May 1989 and May 1990, but also the diurnal curves vary significantly. The previ- ously noted marked period of minimum activity from 1100 to 1800 GMT in May 1989 (Figure 21b) is not present in May 1990. Also, the late morning minimum evident in May 1989 for the eastern region is not as well defined in 1990. The curve for the western part of the study area resembles to some extent the April 1989 (Figure 21a) and September 1989 (Figure 33a) curves, however, in that all three curves display a 118 1 m ........................ Cl Positive Strikes B Negative Strikes “12131415161715192’1'321222; nmemun nmemmn Fig. 36a. Hourly Flash Frequencies for the Western Portion of the Study Area, May 1990. 1m ........................ 1w ........................ DPositiveStrtkes .Negative Strikes w ........................ '5 'flflflflflflfiflfifl 01284567891011 "121314151617181920212223 Tlme (GMT) T'ime(GMT) Fig. 36b. Hourly Flash Frequencies for the Eastern Portion of the Study Area, May 1990. 119 pronounced late night maximum (at 0400 GMT, respectively at 0500 GMT). With respect to the polarity of lightning events, a distinct preference of positive strikes for the western portion of the study area during the nighttime period is evi- dent. June 1990 Unlike June 1989, which had a daytime frequency maximum, June 1990 showed a pronounced nighttime maximum in lightning activity (3257 flashes compared to 1697). The distinct pref- erence for nighttime events to occur in the western half of the study area and daytime events in the eastern half of the study area described in June 1989 (Figures 22a and 23a) is not evident in June 1990 (Figures 37a and 38a). In June 1990, lightning activity is most frequent in the western por- tion of the study area in both time periods, particularly in southern Iowa and western Illinois. Also, a secondary fre- quency maximum is evident in the eastern portion of the study area in central and southwestern Ohio. The larger number of events is also reflected in the plot of lightning days (Figures 37b and 38b). In both day— time and nighttime periods, the number of days with lightning activity is larger in the western portion, in particular in central Iowa, where some grid cells experienced 9 to 10 lightning days. The nighttime maximum in lightning activity is evident in the hourly frequency distributions, too. The June 1990 120 Fig. 37a. Nighttime (oooo - 1200 GMT) Lightning Strikes. June 1990. Number of Strikes E 3 ¥ 9. S’ ’5 .HINU ONUI‘D-P 38338 ..naabn 0 Fig. 37b. Number of Days with Nighttime Lightning Activity. June 1990. Number of Strikes 1to4 5t09 10t017 18t028 29to45 IIINU Fig. 38a. Daytime (1200 - 2400 GMT) Lightning Strikes. June 1990. Number oi Days E11102 3to4 I 5t06 I 71118 I 9to10 Fig. 38b. Number of Days with Daytime Lightning Activity, June 1990. 122 (Figure 39a) distribution for the western portion of the study area closely resembles the May 1990 distribution (Figure 36a). Both curves have considerably higher hourly values in the nighttime period with the time of maximum fre- quency occurring around 1-2 hours before local midnight (at 0400, respectively 0500 GMT). While May 1990 shows a drop in lightning frequencies between 1500 and 1800 GMT, the hourly values in June maintain a fairly steady level during the day (note the difference in scale between the May and June plots). The plots for the eastern portion of the study area for May (Figure 36b) and June (Figure 39b) do not display similarities. Lightning activity ixiihnua 1990 shares some features with the June 1989 distribution. Despite the much larger am— plitude in 1990, both curves display a nighttime maximum at 0300 GMT, respectively at 0400 GMT, for the western part of the study area” .An additional consistent feature between June 1989 (Figure 24b) and 1990 (Figure 39b) is the distinct period of relatively infrequent lightning activity in the eastern portion of the study area from approximately 1200 to 1600 GMT (late morning and midday). In June 1990 this period is prolonged until 1900 GMT. For the eastern region, the in- crease in lightning activity in the afternoon hours that had been reported for May through September 1989 is evident again. Also apparent is the preference of positive lightning for the western portion of the study area (Figure 39b). 123 E] Positive Strikes E Negative Strikes o - 121314151617181920212228 nmemun Fig. 39a. Hourly Flash Frequencies for the Western Portion of the Study Area, June 1990. m ........................ 2m ........................ _ _ _ ClPositive Strikes 15° BNegative Strikes 0 121814151817181920212223 nmemun Fig. 39b. Hourly Flash Frequencies for the Eastern Portion of the Study Area, June 1990. 124 Unlike May, no clear temporal preference can be identified, however. July 1990 In July 1990, lightning activity was more evenly dis- tributed in the nighttime and daytime periods (1748 and 1467 strikes) compared to June 1990. However, the maximum number of strikes per cell was greater for the nighttime period. During the night, the cell with the highest strike totals re— ported 34 C63; the highest daytime strike density, on the other hand, was 19 CGS per cell. The nighttime lightning flashes during July 1990 were more frequent in the western part of the study area than in the east (Figure 40a). The area of higher frequencies extends northward from Iowa into southern Minnesota and eastward into Illinois. The number of lightning days is also large for this area (Figure 40b). A major difference between the nighttime distribution for July 1989 (Figure 25a) and July 1990 is the relatively infrequent occurrence of flashes over southern Lake Michigan and western Lake Erie compared to July 1990. As for July 1989, daytime lightning activity is more frequent in the eastern half of the study area although the number of strikes reported in 1989 (Figure 26a) was consider- ably larger than in 1990 (Figure 41a). The highest concen— tration of daytime lightning events is observed in Ohio. A slight southwest-northeast oriented axis of greater activity is evident from Indiana across Ohio into West 125 Fig. 40a. Nighttime (0000 - 1200 GMT) Lightning Strikes. Juiy 1990. Number of Strikes 104 ms 0tot7 81028 9to45 IIIND 1 5 1 1 2 Number of Days IIIND UNUIO-b d00‘” 83833 c Fig. 40b. Number of Days with Nighttime Lightning Activity, July 1990. 126 Number of Strikes 1to4 5to9 10 to 17 18 to 28 291045 IEINEJ Fig. 41a. Daytime (1200 - 2400 GMT) Lightning Strikes. July 1990. Number of Days IIIEEJ outrun- 2 4 5 8 1 88388 0 Fig. 41b. Number of Days with Daytime Lightning Activity. July 1990. 127 Virginia. The frequency maximum in Ohio also corresponds with the relative maximum in the number of lightning days per grid cell (Figure 41b). Grid cells with the highest number of lightning days recorded strikes on five or six days. A secondary axis of higher daytime ground flash frequencies in July 1990 extends from Indiana into central lower Michigan. On the other hand, the diurnal distributions of light- ning activity by hour are very similar for the July 1989 and 1990, particularly in the eastern portion of the study area. In both months, a period of elevated activity is evident from approximately 1900 to 0200 GMT (late afternoon and early evening) in the eastern portion (Figures 27b and 42b), whereas in the western portion (Figures 27a and 42a) light- ning activity is frequent over a broader period (approximately 2100-1100 GMT). In July 1990, very few posi- tive lightning strikes were recorded. 128 1 5° ________________________ [:1 Positive Strikes Negative Strikes o'. ‘ _ .L' . 12 1a 14 15 1e 17 1e 19 20 21 22 2: hmemmn Fig. 42a. Hourly Flash Frequencies for the Western Portion of the Study Area, July 1990. Dmmme&Mu E] Negative Strikes o L" L 121314151617101920212223 nmemun Fig. 42b. Hourly Flash Frequencies for the Eastern Portion of the Study Area, July 1990. 129 August 1990 Lightning activity was considerably more frequent in August 1990 compared to August 1989. Unlike August 1989, a clear nighttime maximum in lightning frequency was evident in August 1990. Areas of higher frequency extended further north during August 1990 than in the previous months of June and July. IDs the nighttime period, lightning activity was most frequent in northern Iowa, southwestern Wisconsin and in eastern Indiana (Figure 43a). These cores of high lightning frequencies, however, are surrounded by cells with fairly high strike frequencies. In fact, the nighttime period of August 1990 has the largest areal extent of grid cells with five or more strikes of all examined months. For the daytime hours, two main areas of concentration can be identified in western Iowa and northern Indiana (Figure 44a). Apart from these two main centers of lightning activity with considerable areal extend, several smaller clusters can be located in Ohio, in southern Ontario, and in the central portion of Michigan. The highest cumulative strike frequency per cell was recorded in northern Indiana. This particular cell, however, was affected by CGs on a maxi- mum of two days in the whole month. The number of lightning days for daytime activity (Figure 44b) is comparatively low with only two cells falling into category II (5-6 lightning days). The 3340 daytime ground strikes have a slightly higher maximum cell frequency (43 CG strikes per cell) than the 1835 nighttime events (41 CG strikes per cell). 130 Number oi Strikes to4 m9 0tot7 8t028 29to45 Ole-e IZIND —. -. Fig. 43a. Nighttime (0000 - 1200 GMT) Lightning Strikes, August 1990. Number of Days IIINU 0“de 38888 dam.” Fig. 43b. Number of Days with Nighttime Lightning Activity August 1990. Number 01 Strikes Fig. 44a. Daytime (1200 - 2400 GMT) Lightning Strikes. August 1990. Number of Days E] 1 to 2 I'll owe-u 8388 ..mma O Fig. 44b. Number of Days with Daytime Lightning Activity, August 1990. 132 The shape of the diurnal curve for the western portion of the study area for August 1990 (Figure 45a) is similar to the one previously presented for August 1989 (Figure 30a) while the eastern curve (Figure 45b) more closely resembles the July 1990 (Figure 42b) curve. As for August 1989, the hourly fre- quency distribution for the west shows lightning activity to be least around noon local time. Both curves show a decline in lightning activity between 1500 and 1900 GMT. Lightning activity subsequently picks up and displays a peak shortly after midnight LST (0700 GMT), after which it drops off again. The amplitude of the diurnal curve in August 1990 is about double that of August 1989. In the east, both July 1990 and August 1990 show a marked increase in hourly frequency values around noon LST. While the July curve peaks at 2300 GMT and subsequently de- creases, however, the August curve maintains fairly high hourly values until 0300 GMT and drops off then. As in the previous months, a higher percentage of positive strikes was recorded in the west than in the east. With respect to tem- poral characteristics, a slight preference for the nighttime period can be detected in the record of positive strikes in the western portion of the study area. 133 .012345676 11me(GMT) "12131415 1617161920 212223 nmemun D Positive Strikes [3 Negative Strikes Fig. 45a. Hourly Flash Frequencies for the Western Portion of the Study Area, August 1990. -~5FQFQF1FHF71 1234567391011 nmemun u121314151317161920212223 nmemun [3 Positive Strikes E] Negstm Strikes Fig. 45b. Hourly Flash Frequencies for the Eastern Portion of the Study Area, August 1990. 134 September 1990 Nighttime activity was more frequent in September 1990 than daytime events (1320 strikes compared to 1041). For the nighttime period, a northwest-southeast band of increased ac- tivity can be identified from northwestern Wisconsin into the northeast corner of Ohio (Figure 46a). A striking feature about the nighttime distribution of September 1990 is the oc- currence of both higher lightning frequency and a higher num~ ber of lightning days (Figure 46b) along the northern border of the study region, an area with little activity in the pre- vious months. Maximum nighttime strike frequency per cell was 25 CG flashes compared to 24 CG flashes for the daytime period. During both, night and day, the eastern part of the study area received higher numbers of CG strikes. While there is a fairly good distinction between regions at night with 810 eastern strikes versus 510 strikes in the west, the preference for the eastern section of the study area is very distinct during the day with 865 strikes in the east versus 176 western events. Daytime lightning flashes during September 1990 were concentrated in a triangle extending from southwest Michigan into Pennsylvania and into Ohio (Figure 47a). Generally, the plot of lightning days (Figure 47b) does not correspond well with the distribution of events, indicating that the band of higher lightning frequencies from southwest Michigan into Pennsylvania is most likely the result of one or two major Number of Strikes [11:04 5109 .10t017 I18to28 I 29t045 Fig. 46a. Nighttime (1200 - 2400 GMT) Lightning Strikes. September 1990. Number oi Deye IIIED outrun- 2 4 B 8 1 88888 0 Fig. 46b. Number of Days with Nighttime Lightning Activity, September 1990. 136 Number of Strikes 1to4 5t09 10 to 17 1a to 28 29to 45 IIIED Fig. 47a. Daytime (1200 - 2400 GMT) Lightning Strikes. September 1990. Number oi Days IIIND OVUICQ-h ..OOAN 88888 a Fig. 47b. Number of Days with Daytime Lightning Activity, September 1990. 137 storm systems, while the grid cells in Ohio were affected by several thunderstorms. Inn Ohio lightning activity was re- corded on up to 6 days. While there are few similarities in the frequency distributions of September of both years, September 1990 has some resemblance with. the .August 1990 curve (Note the difference in scale between the two months). In the west (Figures 47a and 48a), little lightning activity occurred around noon and into the early afternoon in both August and September. The period of small lightning activity is even more extended in September (1500-2200 GMT unlike 1500-1900 in August). The largest lightning activity was recorded around local midnight. In September 1990, the period of strongest lightning activity extended from 0600 GMT to 1200 GMT. As for August 1990 (Figure 47b), the frequency distribu- tion for September in the east (Figure 48b) displays a strong increase in lightning activity starting at 1800 GMT and peak— ing at 2300 GMT. While the September curve subsequently de- creases, August shows a prolonged period of elevated hourly frequencies until 0300 GMT. The period of weaker activity in both months extends from 0600 GMT through 1800 GMT. While September does not show much variation in this time interval, the hourly values in August decrease even further to a mini- mum between 1200 GMT and 1800 GMT. Consistent with September 1989, the nighttime interval in the west stands out as the time and place of preferred occurrence of positive lightning strikes. [3 Positive Strikes E Negative Strikes o H 121314151617131920212223 nmemmn Fig. 48a. Hourly Flash Frequencies for the Western Portion of the Study Area, September 1990. El Positive Strikes Negative Strikes o M ‘ “:4 . 121314151317131920212223 nmemun Fig. 48b. Hourly Flash Frequencies for the Eastern Portion of the Study Area, September 1990. 139 Summary: The eastern and western portions of the study area dis- play different temporal preferences of lightning activity. While in 1989 the whole study area reported more daytime strikes than nighttime strikes in June, July, and August, in 1990 July was the only month with more daytime than nighttime flashes. Apart from August 1989 and July 1990 the maximum flash frequencies per cell were reported for the nighttime interval. Areal preferences of lightning activity vary by months. During the day, the east received more lightning strikes than the west in all months except May 1990 and June 1990. The nighttime distribution is not as clear cut. An equal number of months recorded an eastern maximum as a west- ern maximum. Eastern maxima were recorded in April 1989, May 1989, August 1989, August 1990, and September 1990. CHAPTER 7 SUMMARY AND CONCLUSION This thesis was designed as a preliminary examination of cloud-to-ground lightning observations for the southern Great Lakes region. It was part of a larger, ongoing research pro- ject in the Geography Department at Michigan State University concerned with a cdimatology of convection in the southern Great Lakes region. The southern Great Lakes region lies astride the boundary between the nocturnal thunderstorm and precipitation regime in the central U.S. and the afternoon regime in the eastern U.S. Therefore, it is ideally suited for a systematic comparison of characteristics of nocturnal and daytime lightning activity and convection. In the first part of the study, the diurnal and seasonal variations of lightning activity were examined with respect to frequency of ground strikes, number of return strokes, peak amplitude, and polarity. One objective was to investi- gate the timing of peaks in amplitude and return strokes and their interrelationship. Another objective was to examine temporal variations in the occurrence of positive and nega- tive strikes. The second part focused on spatial variations in light- ning characteristics across the study region. One goal was 140 141 to determine whether the eastern and western portions of the study area displayed differences in the timing and frequency of lightning events similar to previously identified spatial and diurnal variations of precipitation. Also of interest were temporal and spatial variations in the occurrence of positive lightning flashes. A final objective was the devel- opment and/or assessment of computer software suitable for spatial and temporal analysis of lightning observations. Methodelm For the analysis of temporal trends in lightning charac- teristics both monthly totals and five-minute maxima of the number of return strokes and peak amplitudes were employed. Cloud-to-ground strike frequencies were calculated for the nighttime (0000-1200 GMT) and the daytime (1200-2400 GMT) pe- riod of each month separately. Return stroke and amplitude characteristics were first analyzed separately and then examined for corresponding tem- poral trends by means of five-minute maxima. The choice of five-minute intervals was made because they preserve a fairly high degree of diurnal variability. The data for the respec- tive months of both years were combined and each successive five-minute period (00:00-00:05, 00:05-00:10, etc.) was exam- ined for (1) the strike with the highest number of return strokes and (2) the positive and the negative strike with the largest amplitude. In the calculation of the maximum number of return strokes per five-minute period, no distinction was 142 made with respect to polarity since, as previously reported by other authors (e.g. Rust et al. , 1981), a high percentage of positive flashes have only one return stroke. The differ- ent amplitude characteristics of positive and negative flashes, however, required that a distinction by polarity be made in the analysis of five-minute maxima of amplitude. A grid system commonly used for radar observations (MDR grid) was employed for the analysis of spatial variations in lightning characteristics. A 32x19 cell subgrid of the 113x89-cell MDR matrix extending over the conterminous United States was chosen as the analysis area for the southern Great Lakes region. Lightning locations in latitude and longitude were translated to the MDR grid cells (grid spacing is 47.625 km at 60'N) using software programs provided by the National Meteorological Center. Lightning frequencies and the number of lightning days per cell were calculated monthly for the nighttime (0000-1200 GMT) and daytime (1200-2400 GMT) peri- ods. Atlas Mapmaker was employed in the analysis of grid frequencies and to map grid cell values. In addition, hourly flash frequencies for positive and negative flashes were cal- culated for the eastern and western halves of the study area for both times of day. The graphic output for hourly flash frequencies was produced using Harvard Graphics. 143 The annual distribution of lightning strikes is in gen— eral agreement with other studies of cloud-to-ground light- ning characteristics (e.g. Orville et al. 1987) with the most frequent lightning activity in the summer months and little activity in the winter. There is considerable variation in the frequency and distribution of lightning events between 1989 and 1990, although the number of days with lightning oc- currences is comparable. In 1990, more lightning events (25,549 flashes) were reported than in 1989 (22,168 flashes) despite the fact that in 1990 data were only available until 25 October. The period of "increased lightning activity" in 1989 includes the months of April through September and in 1990 lasts from May through September. For the entire study area more lightning strikes were recorded during the night- time (0000-1200 GMT) than during the daytime (1200-2400 GMT) period. In 1989, 53 percent of cloud-to-ground lightning strikes were reported during the night; in 1990, the compa- rable figure was 60 percent. While overall more strikes were recorded during the night, some months, typically occurring during the warm season (June 1989, July 1989, October 1989, April 1990, and July 1990), reported more strikes during the day. The higher total number of events and the higher per- centage of nighttime events in 1990 suggest different cloud characteristics for both years. One might speculate that 144 nighttime storms are of larger areal extent and/or longer du- ration and/or generate higher flash-density rates. A small scale investigation of areal extent, duration, and flash-density rates would be needed to determine whether differences in cloud systems, i.e. a few major storms or a number of smaller storms, are responsible for the differences in flash frequencies between the nighttime/daytime period and between the two years. The characteristics of return strokes and flash ampli- tude did not display distinct seasonal differences, although some monthly variations were apparent. For example, the max- imum number of return strokes per flash generally was largest in the warm months. The largest number of return strokes for negative flashes (26) was recorded in August 1990; for posi— tive flashes the maximum value of 11 return strokes was recorded in September 1989. The number of return strokes varied markedly for posi- tive and negative flashes. In general agreement with previ- ous studies (e.g. Rust et al., 1981), a high percentage (70- 84 percent) of positive cloud-to-ground flashes had only one return stroke, although the percentage was lower than that reported by Orville et a1 (1987). Also unlike Orville et al's findings, the percentage of negative flashes with only one return stroke did not fluctuate seasonally but rather fell between 29 and 47 percent throughout the year. The peak amplitude of lightning flashes also displayed little diurnal or monthly variation. Contrary to previous 145 findings, the peak amplitudes for negative cloud-to-ground flashes tended to be smaller in the cool season. The highest amplitude values were recorded for negative flashes in the warm season; nine lightning flashes >300kA were recorded in June 1989 and July 1989. The strongest recorded amplitude was 617kA. No direct relationship between the peak current of posi- tive and negative flashes and the number of return strokes was detected in this climatological analysts. Due to the short length of the data set, undue influence by individual storms may overshadow temporal trends. Therefore, case stud- ies of individual storms may be needed to further examine the temporal characteristics of flash densities, peak currents, and number of return strokes. MW: Over the entire study area, cloud-to-ground lightning was more frequent during the nighttime period (0000-1200 GMT) than during the day (1200-2400 GMT) except for some warm months. However, the eastern and western portions of the study area displayed different diurnal characteristics. In 1989, the eastern half of the study area recorded more strikes than the west during all daytime periods. During the nighttime periods in 1989, only April, May, and August recorded a relative maximum in the east. In 1990, the begin- ning of the warm season (May and June) displayed a relative maximum in lightning frequency in the western portion of the 146 study area for both the daytime and nighttime periods. After July, a transition month with a daytime maximum in the east and a nighttime maximum in the west, the later part of summer (August and September) showed relative maxima in the east for both times of day. Little correspondence between the cells of greatest lightning frequencies and the number of days with recorded lightning activity in these cells indicated great differences in flash density for individual storms. Also, the time of day when the largest number of flashes for individual cells was recorded did not always correspond with the time of day of a higher total number of strikes. June and July 1989, for example, with a relative daytime maximum for the entire study area, reported higher maximum flash frequencies per cell dur- ing the night. The axis of largest flash frequency differed consider- ably between months and the same months of different years indicating the impact of individual storm tracks. A slight tendency for a southwest-northeast axis of higher lightning frequency could be detected in both the plot of all lightning events for the period of "increased lightning activity" and in the plots for individual months (e.g. in the daytime plots for July 1989, August 1989, May 1990 and in the graphs for the nighttime period of April 1989, June 1989, September 1989, and May 1990). The southwest-northeast orientation of the area of highest lightning frequencies that could be ob- served both during the day and at night suggests that a 147 different areal division for the analysis of daytime versus nighttime characteristics for the eastern and western halves of the study area might be appropriate. The more frequent occurrence of positive lightning flashes in the western region, particularly at night, may again be indicative of different cloud systems. Positive lightning has previously been reported to occur more fre— quently in regions of stratiform precipitation as opposed to the convective regions of individual storm systems. Additional analyses of the spatial distribution of positive strikes on an individual storm scale are therefore warranted. In some months slightly elevated lightning frequencies are apparent over southern Lake Michigan and western and cen- tral Lake Erie. The increased lightning activity is espe- cially apparent in the plot of lightning days for the night- time periods of May 1989, July 1989, and July 1990 and the daytime period of September 1990. mm For the analysis of the lightning observation data set, a number of FORTRAN programs were written by the author. Other software packages that were employed in the temporal and spatial analyses and graphic output included (1) Harvard Graphics for bar charts of five-minute maximum values of re- turn strokes and peak amplitude values and (2) Atlas MapMaker for the analysis of grid cell frequencies and the mapping of flash frequencies and lightning days within the MDR subgrid. 148 Thirdly, Aldus Freehand was used in combination with Atlas Mapmaker to produce the map of the whole study area (Figure 1). W The most constraining limitation of this climatological analysis was the short length (less than two years) of the data set, especially since the examined 22 nmmths revealed considerable differences in lightning frequency and lightning characteristics between months and the same months of both years. Any seasonal and diurnal trends may be overshadowed by the influence of individual storms. Therefore, the re- sults from the temporal analysis of lightning characteris— tics, especially the five-minute maximum values of the number of return strokes and peak amplitude, have to be treated with caution. Different study regions were employed in the first and second part of the thesis. Due to the difference in areal size of the two study regions and the resulting smaller data set in part two (Figure 14), results from the first part of the study could not be incorporated in the discussion on spa- tial variations. W The preliminary examination of the cloud-to-ground lightning observation data set has shown considerable differences in lightning frequencies between the two examined 149 years. For a climatological study of convection the analysis of a longer data set is warranted. Also, analyses of areal extent, duration, and flash fre- quency of storm systems for different times of day, by sea- son, and for the eastern and western portions of the study area are needed to answer questions about the nature of lightning producing storms. Of special interest in this con- text is whether the cloud-to-ground strikes are produced by few major storms or a number of smaller systems. To further the compatibility of cloud-to-ground light- ning observations with observations of pmecipitation, de- tailed analyses of the relationships between flash-frequen— cies, number of return strokes, peak amplitudes, polarity, cloud characteristics, and precipitation amounts over the du- ration of individual storms are needed. Since this study has suggested different cloud and lightning characteristics for nighttime versus daytime lightning-producing storms, case studies of daytime and nighttime storm systems are strongly recommended. L I ST OF REFERENCES LI ST OF REFERENCES Balling, R. C. Jr. 1985. "Warm Season Nocturnal Precipitation in the Great Plains of the United States. " Q, gljm. Appl. Met. 24, 1383- 1387. Brook, M., M. Nakano, and P. Krehbiel. 1982. "The Electrical Structure of the Hokuriku Winter Thunderstorms." JL_G£QDh¥Si_BaS. 87:C2, 1207-1215. Engholm, C.D., E.R. Williams, and R.M. Dole. 1990. "Meteorological and Electrical Conditions Associated with Positive Cloud-to-Ground Lightning." MQnL_Wea*_Bey. 118, 470-487. Fuquay, D.M. 1982. "Positive Cloud-to-Ground Lightning in Summer Thunderstorms." J*_Gegphys+_3es. 87:C9, 7131-7140. Goodman, S.J. 1983. "Lightning Activity Associated with Severe Storms Embedded within a Mesoscale Convective Storm Complex." MW in Tulsa, Ok, October 17—20, 1983, by the Am. Met. Soc., 29—32. , S.J., D.R. MacGorman. 1986. "Cloud-to-Ground Lightning Activity in Mesoscale Convective Complexes." MW 114:12, 2320-2328- Holle, R.L., R.E. Lopez, W.L. Hiscox, and D. Rosenfeld. 1984. "Cloud-Ground Lightning Associated with Radar Returns in South Florida." W W. in Miami, Fl. January 9-13, 1984, by the Am. Met. Soc., 479-484. Idone, V.P., R.E. Orville and R.W. Henderson. 1984. "Ground Truth: A Positive Cloud-to-Ground Lightning Flash." J. Clim. Appl. Met. 23, 1148-1151. Krehbiel, P. R. , M. Brook, R. L. Lhermitte, and C. L. Lennon. 1983. "Lightning Charge Structure in Thunderstorms, " in W. ed L H Ruhnke and J. Latham (Hampton, Va.: Deepak, 1983), 408- 410. 150 151 Krider, E.P., R.C. Noggle, A.E. Pifer, and D.L. Vance. 1980. "Lightning Direction-Finding Systems for Forest Fire Detection." W. 61:9, 980-986. Lopez, R.E. and R. L. Holle. 1986. "Diurnal and Spatial Variability of Lightning Activity in Northeastern Colorado and Central Florida during the Summer." W Rex. 114:7, 1288-1312. , W. D. Otto, R. Ortiz, and R. L. Holle. 1990. "The Lightning Characteristics of Convective Cloud Systems in Northeastern Colorado " Ersnri_Sixteenth_£onfi_on_§exere it or or t it. '0 eooos- .' '. W by the Am Met Soc., 727- 731. Maier, L.M., E.P. Krider, and M.W. Maier. 1984. "Average Diurnal Variation of Summer Lightning over the Florida Peninsula. W. 112:6, 1134—1140. Orville, R.E. 1981. "Global Distribution of Midnight Lightning-~September to November 1977." W. 109, 391-395. , R.E., R.W. Henderson, and R.E. Pyle. 1990. "The National Lightning Detection Network--Severe Storm Observations " MW ii 00 on 4 iiO‘ o eooero .‘ ’- . t - W, by the Am Met Soc., J27- J30. , R.E., R.W. Henderson, and L.F. Bosart. 1988. "Bipole Patterns Revealed by Lightning Locations in Mesoscale Storm Systems." W. 15:2, 129- 132. , R.E., R.A. Weisman, R.E. Pyle, R.W. Henderson, and R.E. Orville, Jr. 1987. "Cloud-to-Ground Lightning Flash Characteristics From June 1984 Through May 1985." J_._ W. 92 :DS, 5640-5644 . , R.E., R.W. Henderson, and L.F. Bosart. 1983. "An East Coast Lightning Detection Network." BullL_Am+_MeL‘ 59;. 64:9, 1029-1037. Piepgrass, M.V. and E.P. Krider. 1982. "Lightning and Surface Rainfall During Florida Thunderstorms." J_._ W. 87, 11, 193-11,201- Rust, W.D., D.R. MacGorman and R.T. Arnold. 1981. "Positive Cloud-to-Ground Lightning Flashes in Severe Storms." W. 8:7: 791-794 . 152 Rutledge, S.A. and D.R. MacGorman. 1988. "Cloud-to-Ground Lightning Activity in the 10-11 June 1985 Mesoscale convective System Observed during the Oklahoma-Kansas PRE- STORM Project. W 116:7: 1393-1403- Speheger, D.A., D.J. Shellberg, J. R. Gibbons, J. A. DeToro, and T. P. Grazulis. 1990. "A Climatology of Severe Thunderstorm Events in Indiana. " in W 0 O 0 I ‘ ‘ II C I 0 O 4 11. . 4 Hananaskis_RarhL_AltaiL_Canadar_92tober_22;262_1290 by the Am. Met. Soc. , 18- 23. Stolzenburg, Maribeth. 1990. "Characteristics of the Bipolar Pattern of Lightning Location Observed in 1988 Thunderstorms." B_11,§I.,.‘l._._Am_.__Mej.;_.__fi_g25;_L 71:9, 1331-1338. Takagi, N., T. Takeuti, and T. Nakai. 1986. "On the Occurrence of Positive Ground Flashes." Jl_fiegnh¥SL_ReSi 91:D1, 9905-9909. Wallace, J.M. 1975. "Diurnal Variations in Precipitation an Thunderstorm Frequency over the Conterminous United States." W. 103, 406-419. Winkler, J. A. 1987. "Diurnal Variations of Summertime Very Heavy Precipitation in the Eastern and Central United States. Physical_fieogranh¥ 8: 3, 210- 224. , E.R. Skeeter, and P.D. Yamamoto. 1988. "Seasonal Variations in the Diurnal Characteristics of Heavy Hourly Precipitation across the United States. hfiunJL__fleathl Bay. 116, 1641-1658. APPENDICES 00OOOOOOOOOOOOOOOOOOOOOOOO 10 123 APPENDIX A FORTRAN_Erogram_Grid PROGRAM GRID December 1991 - C. Gunreben Program to take file of lightning strikes with parameters month, day, year, hour, min, sec, latitude, longitude, # of return strokes, polarity, amplitude and put them into a 113x89 grid according to MDR standards. GRIOOOlO GRIOOOZO GRIOOOBO GRI00040 GRIOOOSO GRIOOO6O GRIOOO70 The conversion program provided by the Nat. GRI00080 Meteorol. Center is used to translate the coor-GRIOOO9O dinates from latitude/longitude to the grid coordinate system overlaid on a polar stereo- graphic map projection true at 60 degrees. The original record is read and the pro- spective grid cell is determined. The grid cell is added at the end of each record as variable of i and j. Variables: radpd earthr earthradius xmeshl meshlength of grid cells at 60 Degr. N orient grid centered along 105 Degr. W INTEGER I,J INTEGER MONTH,DAY,YEAR,HOUR,MIN,SEC,MULT REAL ELAT,ELONG,STRENGTH,RADPD,XMESHL,ORIENT, REAL EARTHR,RE,XI,XJ REAL WLONG,R,XLAT RADPD=0.01745329 EARTHR=6371.2 XMESHL=47.625 ORIENT=105.0 READ (9,123,END=85) MONTH,DAY,YEAR,HOUR,MIN, *SEC,ELAT,ELONG,MULT,STRENGTH GRIOOlOO GRIOOllo GRI00120 GRIOOlBO GRIOOl4O GRIOOISO GRI00160 GRI00170 GRI00180 GRI0019O GRIOOZOO GRIOOZlO GRIOOZZO GRIOOZBO GRI0024O GRIOOZSO GRI00260 GRI00270 GRI00280 GRIOOZ90 GRI00300 GRIOO310 GRIOOBZO GRIOO330 GR10034O GRIOOBSO GRI00360 GRI00370 GRIOO380 FORMAT (6(I2,1X),1X,F6.3,2X,F7.3,1X,i2,2X,F6.1)GR100390 ELONG=ABS(ELONG) RE=(EARTHR*1.86603)/XMESHL 153 GRIOO4OO GRIOO410 124 85 154 XLAT=ELAT*RADPD WLONG=(ELONG+180.0-ORIENT)*RADPD R=(RE*COS(XLAT))/(1.0+SIN(XLAT)) XI=R*SIN(WLONG) XJ=R*COS(WLONG) XI=XI+41 XJ=XJ+161 I=XI =XJ WRITE (10,124) MONTH,DAY,YEAR,HOUR,MIN,SEC, *BLAT,ELONG,MULT,STRENGTH,I,J FORMAT (6(I2,1X),1X,F6.3,2X,F7.3,1X,i2,2X, *F6.1,2X,I2,2X,I2) goto 10 WRITE (*,*) 'DONE' STOP END GRIOO420 GRIOO43O GRIOO44O GRIOO450 GRIOO46O GRIOO47O GRIOO48O GRIOO49O GRIOOSOO GRIOOSlO GRIOOSZO GRIOOS3O GRI0054O GRIOOSSO GRI00560 GRIOOS7O GRIOOSBO 0000 35 25 15 10 123 20 85 75 65 55 45 80 APPENDIX B W PROGRAM MMULTMAX MMU00010 program to read file of five-minute maxima of MMU00020 the number of strikes (Fivemult) and create MMU00030 files of five-minute maxima for the same MMUOOO40 months of both years combined MMUOOOSO INTEGER MONTH,DAY,YEAR,HOUR,MIN,SEC,MULT MMU00060 INTEGER I,J,K,L,M,AMULT(12,0:23,12) MMUOOO70 REAL ELAT,ELONG,STRENGTH MMU00080 DO 15, I=1,12 MMU00090 DO 25, J=0,23 MMU00100 DO 35, K=1,12 MMUOOllO AMULT(I,J,K)=O MMU00120 CONTINUE MMU00130 CONTINUE MMU00140 CONTINUE MMU00150 READ (9,123,END=80) MONTH,DAY,YEAR,HOUR,MIN, MMU00160 *SEC,ELAT,ELONG,MULT,STRENGTH MMU0017O FORMAT (6(I2,1X),1X,F6.3,2X,F7.3,1X,I2,2X,F6.1)MM000180 DO 45, I-1,2 MMUOOl90 DO 55, J=1,12 MMU00200 DO 65, K=1,31 MMU00210 DO 75, L=0,23 MMU00220 DO 85, M=1,12 MMU00230 IF (MONTH.EQ.J.AND.HOUR.EQ.L.AND. MMU00240 * MIN.LT.(M*5)) THEN MMUOOZSO IF (AMULT(J,L,M).LT.MULT) THEN MMU00260 AMULT(J,L,M)=MULT MMU00270 ENDIF MMU00280 READ (9,123,END=80) MONTH,DAY,YEAR,MMU0029O * HOUR,MIN,SEC,ELAT,ELONG,MULT, MMU00300 * STRENGTH MMU0031O IF (MONTH.EQ.J.AND.HOUR.EQ.L.AND. MMU00320 * MIN.LT.(M*5)) THEN MMU00330 GOTO 20 MMU00340 ENDIF MMU00350 ENDIF MMU00360 CONTINUE MMU00370 CONTINUE MM000380 CONTINUE MMU00390 CONTINUE MMU00400 CONTINUE MMU00410 DO 95, I=1,12 MMUOO420 155 124 115 105 95 156 D0 105, J=0,23 DO 115, K=1,12 WRITE (I+10,124) I,J,K,AMULT(I,J,K) FORMAT (12,3X,12,3X,I2,3X,I2) CONTINUE CONTINUE CONTINUE WRITE (*,*) 'DONE' STOP END MMUOO430 MMUOO44O MMU00450 MMUOO460 MMU00470 MMU00480 MMUOO490 MMUOOSOO MMUOOSlO MMU00520 0000000 O000000000000000000000000 APPENDIX C FORTRAN__Erogram_Da¥s PROGRAM DAYS DAYOOOlO December 1991 - C. Gunreben DAY00020 program to calculate the number Of lightning DAY00030 days per grid cell (32x19 MDR grid) by day/ DAY00040 nighttime (0000-1200 GMT / 1200-2400 GMT) and DAY00050 put them into monthly files; output adjusted DAY00060 according tO Atlas MapMaker data file require- DAYOOO7O ments DAY00080 INTEGER MONTH,DAY,YEAR,HOUR,MIN,SEC,MULT DAYOOO9O INTEGER CYEAR,CMONTH,CDAY,TIME DAYOOlOO INTEGER I,J,K,L,N,P DAYOOllO INTEGER COUNT(89:90,12,2,57:88,48:66) DAY00120 INTEGER FLAG(2,57:88,48:66) DAY00130 REAL ELAT,ELONG,STRENGTH DAY00140 DAY00150 variables: DAY00160 month,day,year,hour,min,sec = time variables DAY00170 original data set DAY00180 elat/elong = latitude/longitude original data DAY00190 set DAY00200 mult = number of return strokes original DAY00210 data set DAY00220 strength = amplitude original data set DAY00230 i/j = location within 113/89 MDR grid DAY00240 system calculated by Fortran DAY00250 Program GRId (C. Gunreben, 1991) DAY00260 l = variable tO adjust output files DAY00270 n/p = variables to adjust location with-DAY0028O in 113/89 MDR grid tO 32x19 grid DAY00290 time = variable for distinction daytime/ DAY00300 nighttime in array DAY00310 1=0000-1200 GMT /2=1200-2400 GMT DAY00320 count = array to count number of lightningDAY00330 days per cell DAY00340 count(year,month,time,i,j) DAY00350 flag = marker to reject further strikes DAY00360 for one cell for the same day DAY00370 DAY00380 set arrays to 0 DAY00390 DO 5, CYEAR=89,90 DAY00400 DO 10, CMONTH=1,12 DAY00410 157 4O 30 20 10 15 123 68 67 66 0000 0000 158 DO 20, TIME=1,2 DO 30, I=57,88 DO 40, J=48,66 COUNT(CYEAR,CMONTH,TIME,I,J)=O FLAG(TIME,I,J)=O CONTINUE CONTINUE CONTINUE CONTINUE CONTINUE read first record READ (9,123,END=85) MONTH,DAY,YEAR,HOUR,MIN, *SEC,ELAT,ELONG,MULT,STRENGTH,I,J read next record if outside grid IF (I.GT.88.0R.I.LT.57.0R.J.GT.66.0R.J.LT.48) *THEN GOTO 15 ENDIF FORMAT (6(12,1X),1X,F6.3,2X,F7.3,1X,i2,2X, *F6.1,2(2X,IZ)) set array DO 50, CYEAR=89,9O DO 60, CMONTH=1,12 DO 65, CDAY=1,31 DO 66, TIME=1,2 DO 67, K=57,88 DO 68, L=48,66 FLAG(TIME,K,L)=0 CONTINUE CONTINUE CONTINUE IF (YEAR.EQ.CYEAR.AND.MONTH.EQ.CMONTH. * AND.DAY.EQ.CDAY) THEN IF (HOUR.LT.12) THEN TIME=1 check grid cell, if no strike has been recorded for that day for 0000-1200 GMT - advance daycount for nightly hours by one IF (FLAG(TIME,I,J).EQ.O) THEN COUNT(CYEAR,CMONTH,TIME,I,J)=COUNT * (CYEAR,CMONTH,TIME,I,J)+1 FLAG(TIME,I,J)=1 ENDIF ELSE TIME=2 check grid cell, if no strike has been recorded for that day for 1200-2400 GMT, advance daycount for daytime hours by one IF (FLAG(TIME,I,J).EQ.O) THEN COUNT(CYEAR,CMONTH,TIME,I,J)=COUNT * (CYEAR,CMONTH,TIME,I,J)+l FLAG(TIME,I,J)=1 DAY00420 DAYOO43O DAYOO44O DAYOO450 DAYOO46O DAYOO47O DAY00480 DAYOO49O DAYOOSOO DAYOOSlO DAYOOSZO DAY00530 DAY00540 DAYOOSSO DAY00560 DAY0057O DAY00580 DAY00590 DAYOOGOO DAY00610 DAY00620 DAY00630 DAY00640 DAY00650 DAY00660 DAY00670 DAYOO68O DAY00690 DAY007OO DAY00710 DAY00720 DAYOO730 DAYOO740 DAYOO75O DAY00760 DAY0077O DAY00780 DAYOO79O DAYOOBOO DAY00810 DAY00820 DAY00830 DAY00840 DAY00850 DAY00860 DAY0087O DAY00880 DAYOO89O DAY00900 DAY00910 DAY00920 DAY00930 DAY00940 DAYOO950 65 6O 50 85 000 126 127 90 80 110 100 150 159 ENDIF ENDIF read next record READ (9,123,END=85) MONTH,DAY,YEAR, HOUR, MIN, SEC, ELAT, ELONG, MULT, STRENGTH, I,J IF (I.GT.88.0R.I.LT.57.0R.J.GT.66.0R. J.LT.48) THEN GOTO 25 ELSE goto 35 ENDIF ENDIF CONTINUE CONTINUE CONTINUE DO 100, CYEAR=89,9O DO 110, CMONTH=1,12 IF (CYEAR.eq.90) THEN l adjusts output to 22 monthy files L=CMONTH+12 ELSE L=CMONTH ENDIF output adjusted so it can be read directly into Atlas Mapmaker boundary file created for this grid system by C. Gunreben (1991) WRITE (10+L,126) 'M',’/','D',',','DAYAM', ',','DAYPM' FORMAT (A1,A1,A1,A1,A5,A1,A5) DO 80, I=57,88 DO 90,J=48,66 N=I-56 P=J-47 WRITE (10+L,127) N,'/',P,',',COUNT (CYEAR,CMONTH,1,I,J), ',',COUNT(CYEAR,CMONTH,2,I,J) FORMAT (I2,Al,IZ,A1,I4,A1,I4) CONTINUE CONTINUE IF (L.EQ.22) THEN GOTO 150 ENDIF CONTINUE CONTINUE WRITE (*,*) 'DONE' STOP END DAYOO960 DAYOO970 DAYOO980 DAYOO990 DAYOlOOO DAYOlOlO DAYOlOZO DAY01030 DAY0104O DAYOlOSO DAY01060 DAY01070 DAY01080 DAY01090 DAYOllOO DAYOlllO DAY01120 DAY01130 DAY01140 DAY01150 DAY01160 DAY01170 DAY01180 DAY01190 DAY01200 DAY01210 DAY01220 DAY01230 DAY01240 DAY01250 DAY01260 DAY01270 DAY01280 DAY01290 DAY01300 DAY01310 DAY01320 DAY01330 DAY0134O DAY01350 DAY01360 DAY01370 DAY01380 DAY01390 DAY01400 DAY01410 DAY01420 DAY01430