sign in sign up
<<
Home , Landfall, Rainband, Typhoon Dujuan (2003)

NOVEMBER 2012 Z H A N G E T A L . 3573

Lightning Distribution and Eyewall Outbreaks in Tropical during

WENJUAN ZHANG Laboratory of Lightning Physics and Protection Engineering, Chinese Academy of Meteorological Sciences, and Graduate University of Chinese Academy of Sciences, Beijing,

YIJUN ZHANG Laboratory of Lightning Physics and Protection Engineering, Chinese Academy of Meteorological Sciences, and State Key Laboratory of Severe Weather, Chinese Academy of Meteorological Sciences, Beijing, China

DONG ZHENG Laboratory of Lightning Physics and Protection Engineering, Chinese Academy of Meteorological Sciences, Beijing, China

XIUJI ZHOU Laboratory of Lightning Physics and Protection Engineering, Chinese Academy of Meteorological Sciences, and State Key Laboratory of Severe Weather, Chinese Academy of Meteorological Sciences, Beijing, China

(Manuscript received 23 November 2011, in final form 10 April 2012)

ABSTRACT

Cloud-to-ground lightning data and intensity data ( and central pressure) for 33 northwest Pacific tropical cyclones were used to analyze lightning distributions during the period of landfall in China. Lightning activities varied enormously from storm to storm with an average flash rate over 500 km of radius from 3 to 3201 flashes per hour, and no obvious relationship between average intensity and average flash rate occurred. The maximum flash density shifted from the eyewall region (0–60 km) to outer (180–500 km) as the intensity level increased. The average ratio of flash density in the eyewall to outer 2 was highest (1:0.5) for with the level of a tropical storm (17.2–24.4 m s 1) and lowest (1:8.6) 2 for severe (41.5–50.9 m s 1). After storm landfall, flash density in the rainband decreased more 2 rapidly in severe typhoons than in severe tropical storms (24.5–32.6 m s 1) and typhoons, but increased in 2 tropical depressions (10.8–17.1 m s 1) and tropical storms. With the strength of intensity level, lightning in the outer rainband gradually weakened after the storm landfall. Lightning outbreaks were identified in a consistent manner for all tropical cyclones to inspect the re- lationship of eyewall flashes to the changes of structure and intensity. Eyewall flash outbreaks were found during the period of intensity change (15% of outbreaks in intensification and 43% in weaken), and the period of maximum intensity (15% of outbreaks) of storms. A new result of this analysis found that 10% of the outbreaks occurred prior to and during periods of storm turning, which is potentially important for the tra- jectory change forecasting of tropical cyclones.

1. Introduction in tropical cyclones. Dun (1951) suggested that thun- derstorms occurred frequently on the outer region but Our knowledge about lightning activities in tropical not the center of hurricanes. Jorgensen et al. (1985) in- cyclones (TCs) has gone through several stages. In the early dicated the strongest hurricane updraft cores were weak times, there was a lack of research on storm electrification in comparison with that in midlatitude . Willoughby et al. (1985) found seeding for supercooled water was ineffective in hurricanes because of little su- Corresponding author address: Wenjuan Zhang, Laboratory of Lightning Physics and Protection Engineering, Chinese Academy percooled water above the melting level. Black and Hallett of Meteorological Sciences, Beijing 100081, China. (1986), on the basis of aircraft observations, also found E-mail: [email protected] a limited amount of supercooled water above the melting

DOI: 10.1175/MWR-D-11-00347.1

Ó 2012 American Meteorological Society Unauthenticated | Downloaded 10/03/21 11:14 AM UTC 3574 MONTHLY WEATHER REVIEW VOLUME 140 level in hurricanes. The above work suggests that inter- ionosphere to determine lightning strike locations (Pessi actions between supercooled water, large ice aggregates, et al. 2009). As the detection range of LLDN is an order and small ice particles for noninductive collision charge of magnitude greater than that of the NLDN, data from separation (e.g., Takahashi 1978; Saunders and Peck 1998) the network were used for a more complete documen- are not abundant in TCs, and the necessary microphys- tation of lightning for the evolution of hurricanes over ical conditions may not be sufficient to produce high could- the open ocean (Squires and Businger 2008; Leary and to-ground lightning rates in tropical cyclones. Black et al. Ritchie 2009; Demetriades et al. 2010). Moreover, the (1986) took some of the first steps to increasing our knowl- World Wide Lightning Location Network (WWLLN; edge about lighting activities in tropical storms (Samsury Rodger et al. 2005, 2006) has also been used to study and Orville 1994). On the basis of study on Hurricane lightning in tropical storms. Price et al. (2009) and Pan Diana (1984), they found lightning to be frequent in trop- et al. (2010) examined lightning activities in TCs during ical cyclones. Venne et al. (1989) and Lyons et al. (1989) the entire lifetime over the globe and the northwest Pa- also found lightning stroke frequently in hurricanes and cific, respectively. Abarca et al. (2011) studied 24 At- mainly located in the eyewall and outer rainbands. lantic basin TCs with WWLLN data and suggested flash With the development of lightning detecting tech- density in the inner core was potential for distinguishing nologies in the last two decades, observations have de- intensifying versus nonintensifying TCs. Although the termined that lighting in tropical cyclones may be a more DE and LA of the WWLLN are still low, it can be used common event than originally thought and studies began to overcome the spatiotemporal limitations that other to investigate frequency and distribution of lightning ground-based networks have, thus allowing global, real- in tropical cyclones. Land-based lightning detection net- time monitoring of lightning in tropical storms. Despite works that operate primarily in very low-frequency to efforts to investigate the various aspects of lightning in low-frequency (VLF/LF) bands allow lightning moni- TCs noted above, the magnitude and radial distribution toring within a few hundred kilometers of the coast. of the flash rate in TCs have varied significantly among Samsury and Orville (1994), using the National Light- the studies because of the different technology used in ning Detection Network (NLDN; Cummins et al. 1998; the ground-based network. Cummins and Murphy 2009) data, described the distri- Satellites provide a common means for monitoring bution of lightning in Hurricanes Hugo and Jerry (1989) storms over the oceans and were used to investigate total as they made landfall, and found the majority of flashes lightning [-to-ground (CG) plus intracloud (IC)] located in the right-front and right-rear quadrants. Molinari and passive microwave ice scattering magnitudes in trop- et al. (1994, 1999), also with the NLDN data, found a ical cyclones. Observations of total lightning from the common radial distribution of lightning in nine North Optical Transient Detector (OTD) were compared with Atlantic basin hurricanes: a weak maximum in the - TC intensity by Cecil and Zipser (1999). With Lightning wall region, a minimum in the inner rainbands (80–100- Imaging Sensor (LIS) data from the Tropical Rainfall km radius), and a strong maximum in the outer rainbands Measuring Mission (TRMM) satellite, Cecil et al. (2002a, (210–290-km radius). Lyons and Keen (1994) examined 2002b) studied 261 overpasses of 45 hurricanes and found lightning in four tropical storms in the Gulf of Mexico radial distribution of lightning was similar to Molinari using Lightning Position and Tracking System (LPATS) et al. (1994, 1999). They proposed that supercooled cloud data, and found lightning was common within the outer water occurred preferentially in outer rainbands com- rainbands while absent within the interior of mature TCs. pared to other tropical oceanic . Yokoyama More recently, Fierro et al. (2011) provided a first insight and Takayabu (2008), also using LIS data from TRMM, into the three-dimensional electrical activity with the Los however, found a larger flash rate and more vigorous Alamos Sferics Array (LASA; Shao et al. 2006) and in- convective activity in the inner core (0–60 km) compared vestigated the inner-core lightning with changes of storm to the rainband (60–500 km). intensity in Hurricanes Katrina and Rita (2005). The The predictive values of CG lightning to structure and above land-based networks have been the most conve- intensity changes of TCs have been discussed recently. nient platform to observe lightning in TCs with high de- Studies have shown that lightning outbreaks in the eye- tection efficiency (DE) and location accuracy (LA) in the wall often occur prior to or during periods of storm in- domain, but their coverage are limited to continent and tensification, such as the case studies of Hurricanes Diana adjacent ocean for the DE and LA decreasing sharply (1984) and Florence (1988) by Lyons and Keen (1994), with the distance from the shore. (1992) by Molinari et al. (1994), and The Long-Range Lightning Detection Network Hurricanes Rita and Katrina (2005) by Shao et al. (2005), (LLDN), operated by the Vaisala Group, Squires and Businger (2008), and Thomas et al. (2010). detects VLF electromagnetic waves reflected from the Lightning outbreaks were usually recognized as sharp

Unauthenticated | Downloaded 10/03/21 11:14 AM UTC NOVEMBER 2012 Z H A N G E T A L . 3575 increases of flash density in time series. Nagele (2010) proposed a quantitative definition of outbreak consid- ering both the increase in flashes from one time period to the next and the threshold on absolute flash count. Squires and Businger (2008) analyzed the morphology of eyewall lightning in two category-5 hurricanes and found each hurricane produced eyewall outbreaks dur- ing the period of rapid intensification (RI), replacement cycles (RC), and the period encompassed the maximum intensity (MI). Molinari et al. (1999) proposed the in- dications of eyewall lightning outbreaks in the following: an outbreak in a weakening, steady, or slowly deepening hurricane might indicate a rapidly intensification; an out- break in a hurricane deepening for some time may in- dicate the imminent end or reversal of intensification; the FIG. 1. Sensor locations of the lightning location system (GDLLS) lack of inner-core flashes may indicate little change in of the Electric Power Company. the hurricane. Price et al. (2009) observed a statistical increase of lightning activity approximately one day be- but also strong winds and disasters, which fore the peak winds. Fierro et al. (2011) revealed an in- cause large fatalities and economic losses. In addition, crease in discharge heights during RI and suggested that China has been constructing the national lightning de- the IC narrow bipolar events (NBEs) were useful in tection network since the 1990s and has now obtained tracking the evolution of strong convection during in- ground-based lightning data for more than one decade. tensification. Furthermore, increasing CG flash rates could In this work, we use lightning data from a regional light- also accompany or precede a weakening of weak TCs ning detection network in , along with storm and/or an imminent change in its track (Willoughby 1990; track and intensity data from the China Meteorological Fierro et al. 2007). It was suggested by Black and Hallett Administration (CMA), to study the characteristics (1999) that lightning outbreaks within the eyewall should of lightning activity during the periods of landfall be rare when compared to continent thunderstorms. A (prelandfall, landfall, and postlandfall). We examine the high electric field was collocated with strong updraft and following three fundamental questions: 1) What are the high supercooled water concentrations. The rarity of su- frequency and distribution of lightning in TCs during percooled water and inactive interaction of graupel and landfall? 2) What are the differences for lightning ac- ice crystals would prevent the charging process within tivity in TCs before and after landfall? 3) Does eyewall the eyewall. By this reason, lightning would occur in the lightning outbreak during the landfall? If so, what are eyewall only when convection and strong updrafts are the predictive values for eyewall lightning to TC changes? rapidly growing. To take full advantage of the predictive value of lightning in tropical cyclones discussed above, modelers have capitalized on the assimilating methods 2. Data and methods of lightning data into numerical models, to help improve a. CG lightning data forecasts of small-scale convective structure of the inner core and, hence, the potential imminent intensity changes Lightning activities of 33 tropical cyclones that made of the storms (Marks and Shay 1998; Fierro et al. 2007; landfall in Guangdong province, south China, from 1999 Fierro and Reisner 2011). to 2010, were analyzed in this study. The lightning data Overall, the above studies showed variation of spatial were from a lightning location system [Guangdong and temporal characteristics of lightning activity in Lightning Location System (GDLLS); Chen et al. 2004], tropical cyclones, and indicated the potential of using which was put into operation in 1997 in the grid of data from lightning detection network to predict changes Guangdong Electric Power Company. The network of storm intensity and evolution of structure. This study consists of 16 sensors (Fig. 1) and provides real-time CG looks more closely at the lightning characteristics of lightning locations for the whole area of Guangdong northwest Pacific tropical cyclones during their landfall province and its vicinity covered by the combined tech- in China. China is one of the most severely - nique of magnetic direction finding (DF) and time of ar- affected countries in the world with about seven to eight rival (TOA). It provides information of date, time, latitude, typhoons making landfall each year. The landfalling ty- longitude, peak current, polarity, and number of strokes phoons bring not only severe flooding to coastal areas, per flash. Chen et al. (2002) evaluated the performance

Unauthenticated | Downloaded 10/03/21 11:14 AM UTC 3576 MONTHLY WEATHER REVIEW VOLUME 140 of GDLLS with the data of recorded lightning faults on of super typhoons was observed when lightning occurred transmission lines in 1997–99, and found the DE is as during the period of study. high as 86% and the median LA is about 1.3 km. Com- Each storm was divided into three regions based on parisons with field studies on artificially triggered light- electrical characteristics: eyewall, within 60 km of the ning in 2007 and 2008, Chen et al. (2010) showed that the storm center; inner rainband, 60–180 km from the cen- LA was 760 m and the DE for flash was 93% at the field ter; and outer rainband, 180–500 km from the center. site within the network. The motivation for dividing the storm was based largely For the purpose of minimizing variation of location on the work of Molinari et al. (1994, 1999) who found error and detection efficiency with distance, range was three broad categories of convection regimes on the basis limited to 400 km away from TC center to at least one of lightning activity. This study used a similar method, sensor in this study. This method was also used by Molinari except the distance from the center for each region was et al. (1999) in interpreting the time variation of flashes slightly different. When analyzing radial distributions of in nine Atlantic basin hurricanes. Although the network flash density for TCs, lightning flashes were divided into provides information on individual strokes that made up 20-km annular rings beginning from the storm center to a flash, only flashes as a whole were examined in this outward to 500 km, and the radial distributions of flash study. Additionally, small positive discharges with peak density were counted. currents less than 10 kA were regarded as cloud discharges (Cummins et al. 1998, 2006) and were not included in the c. Eyewall lightning outbreak definition dataset. Biagi et al. (2007) found a more updated thresh- By inspecting lightning activities in the eyewall region, old of 15 kA for cloud discharges; however, there is no defined as within 60 km of the storm center, an eyewall unique threshold for classifying a small positive report as lightning outbreak was determined if it satisfied the fol- a CG stroke and the exact value is still in debate. lowing two conditions: 1) a 50% increase in flash rate b. Tropical data (flashes per hour) from one hour to the next, and 2) the absolute increased count of flash in one hour was larger The information on TC track and intensity were than the mean flash rate of the storm during the time of obtained from the Yearbook of [China study. This method was also used by Nagele (2010) to Meteorological Administration (CMA) 1999–2010] pub- analyze lightning activity in 16 hurricanes making landfall lished by the CMA for 1999–2010. The yearbook gives along the Gulf of Mexico. By using the above method, the 6-hourly intervals data of center latitude, longitude, eyewall flash outbreaks in all the storms were identified maximum sustained surface speed, and minimum to examine their relationships with changes of storm in- central pressures. Hourly center position and intensity tensity and track. were obtained by spline interpolation. Landfall is de- fined as the time the center of the eye first crosses the coast. The period of landfall begins when the storm center 3. Results is 400 km away from at least one sensor in the network, a. Overall lightning distribution in tropical cyclones and ends when the center is without 400 km off the net- work after the landfall. Table 1 lists the 33 TCs examined Table 1 presents a summary of CG flash rate within in this study and the times their centers were within 400 km 500 km of the center for the 33 storms studied in this of at least one sensor of the detection network. Figure 2 work. The table also includes information on the begin- shows tracks of the TCs during the periods when lightning ning and end time, the landfall time, and the storm peak data were examined. intensity. The average flash rate in Table 1 means the According to the Yearbook of Tropical Cyclone,TC number of flashes occurred in the period of interest di- intensities are classified into six levels in China de- vided by the study hours. The flash rate on land means pending on the mean maximum wind speed: tropical the number of flash occurred during postlandfall divided 2 depression (TD; 10.8–17.1 m s 1), tropical storm (TS; by hours the storm was on land. Table 1 shows that the 2 17.2–24.4 m s 1), severe tropical storm (STS; 24.5– average flash rate varied largely from storm to storm. 2 2 32.6 m s 1), typhoon (TY; 32.7–41.4 m s 1), severe ty- Some of the storms had intense lightning activities with 2 2 phoon (STY; 41.5–50.9 m s 1), and super typhoon (Super flash rates of more than 2000 h 1 [e.g., Typhoon Nuri 2 TY; larger than 51.0 m s 1). Because of changes in in- (2008) and Typhoon Molave (2009)], while some storms tensity throughout the storm’s life cycle, TC intensity had low flash rate of near zero [e.g., Typhoon Leo (1999) levels in this study are determined not by peak intensi- and Tropical Storm Mekkhala (2002)]. ty (i.e., mean maximum wind speed) in the lifetime, but No obvious relationship between the average flash by the central wind speed when lightning occurs. None rate and the storm peak intensity (maximum wind speed

Unauthenticated | Downloaded 10/03/21 11:14 AM UTC NOVEMBER 2012 Z H A N G E T A L . 3577 and minimum central pressure in the lifetime) was found. a ratio of flash densities between the eyewall and outer Severe (2006) had a large maximum rainband of 1:2.9. When the storm strengthened to se- 2 wind speed of 46 m s 1 but the average flash rate was vere typhoon intensity, the vast majority of lightning oc- 2 only 21 h 1, whereas Typhoon Nuri (2008) had a maxi- curred in the outer region and the ratio of flash densities 2 mum wind speed of 38 m s 1, but the flash rate was up between the eyewall and outer rainband reached the 2 to 3201 h 1, much higher than some severe typhoons lowest (1:8.6). (Table 1). Samsury and Orville (1994) found CG flash The ratios of positive to CG flashes in the eyewall and rate for Hurricanes Jerry (1989) was an order of mag- outer rainband were also shown in Table 2. The percent nitude higher than for Hugo, a more intense hurricane. of positive lightning within the eyewall for all intensity Molinari et al. (1994, 1999), with the study of nine Atlantic levels were higher than that in the outer rainband. The basin hurricanes, also found the overall flash rate had no highest percent of positive lightning was 30.4% in the clear relationship with the averaged storm intensity. Hor- eyewall regions of storms with typhoon intensity, and izontal wind speeds that characterized the storm intensity the lowest was 6.6% in the outer rainband with tropical were not a good predictor of lightning activity in TCs, and depression intensity. The average percent of positive the active electrical activity in storms was more related to lightning in the eyewall and outer rainband for the entire the presence of strong updraft (Black and Hallett 1999). sample were 23% and 8%, respectively. Different char- acteristics of positive and total lightning activity in the b. Lightning distribution in tropical cyclones at eyewall and outer rainband might be due to convective different intensity levels regimes in a tropical storm. Holle et al. (1994) showed Figure 3 shows the radial distributions of flash density that in a mesoscale convective system (MCS), the con- for TCs of different intensity levels. Tropical storms and vective region tended to have almost (97%) exclusively severe typhoons had distributions of single peak feature negative flashes, whereas in the trailing stratiform re- with maximum flash densities in the eyewall and outer gion the percent of positive flashed was averagely 24%, rainband. Tropical depressions and severe tropical storms up to 30%. From this aspect, the eyewall of tropical storm had more intense lightning activities in the outer - has similarities to the stratiform region and the outer band and showed a bimodal distribution of flash density rainband to the convective region of MCSs. Previous in the radius. One peak was in the eyewall region similar studies interpreting microphysics and electrification mech- to tropical storm and the other was located in the outer anisms in TCs found: (i) the eyewall has attributes similar rainband. The radial distributions of lightning in typhoons to deep, weakly electrified, monsoonal convection and also showed a bimodal feature, but the two maximum has a relatively low flash rate as a result of weak vertical were both located in the outer rainband. Flash densities velocities and lack of necessary graupel particles above in the eyewall of severe typhoons and typhoons were much the melting level (Black and Hallett 1999); (ii) the outer lower than that in severe tropical storms and tropical rainband, with attributes similar to continental thun- storms. This result was consistent with the study of derstorm, contains the majority number of flashes in the Atlantic basin hurricanes by Molinari et al. (1999), who storm and has a low ratio of positive flash (Molinari et al. found the ground flash density maximum in the core was 1994, 1999; Orville and Coyne 1999; Cecil et al. 2002a,b). 2 larger in marginal (32–35 m s 1 mean maximum sus- 2 c. Lightning distribution during the period of pre- tained surface wind speed) than in strong (56 m s 1) and postlandfall hurricanes. Although flash densities in the eyewall and outer rainband enormously varied with storm intensity A tropical storm goes through many changes of con- level, low flash density in the inner rainband occurred in vective structure during the period of landfall, thus light- all TC levels (Fig. 3). Table 2 presents the ratios of flash ning activity during this time would also seem to change. densities between the eyewall and outer rainband for Figure 4 shows the average flash rates of each storm five TC levels. Lightning had the same flash density in during the period of prelandfall and postlandfall. The the eyewall and outer rainband in tropical depressions. flash rate of prelandfall/postlandfall was obtained by us- Tropical storms had the largest ratio (1:0.5) of flash den- ing the number of flash occurred before/after landfall sityamongallintensitylevelsandthemajorityoflightning divided by hours the storm over water/land. Lightning in storms at this level occurred in the core region. Al- activities became more intense in 15 storms and weaker though the maximum of flash density was also located in 18 storms after the storm made landfall. Flash rates of within the eyewall, lightning in the outer rainband of postlandfall did not change significantly in some storms severe tropical storms became intense and the ratio of compared to prelandfall [e.g., Typhoon Maggie (1999) flash densities between the eyewall and outer rainband and Severe (2003)]. However, light- decreased to 1:0.6. Storms at the typhoon level showed ning activities changed enormously in some TCs [e.g.,

Unauthenticated | Downloaded 10/03/21 11:14 AM UTC 3578 MONTHLY WEATHER REVIEW VOLUME 140

TABLE 1. List of tropical cyclones examined in this study with their flash rates and the times their centers were within 400 km of the lightning location system (GDLLS) of the Guangdong Electric Power Company.

Peak intensity Intensity level Pressure Wind Average flash Flash rate on 2 2 2 in lifetime Storm Year Begin/end Hours (hPa) (m s 1) rate (h 1) Landfall land (h 1) Severe typhoon Imbudo 2003 0800 UTC 23 Jul 36 947 47 455 0200 UTC 24 Jul 23 2000 UTC 24 Jul Dujuan 2003 1900 UTC 1 Sep 35 949 45 86 1200 UTC 2 Sep 96 0600 UTC 3 Sep 1300 UTC 2 Sep 1500 UTC 2 Sep Chanchu 2006 1400 UTC 16 May 41 949 46 21 1815 UTC 17 May 9 1700 UTC 18 May Fengshen 2008 2200 UTC 23 Jun 74 984 28 1609 2130 UTC 24 Jun 896 0000 UTC 27 Jun Hagupit 2008 2100 UTC 22 Sep 50 939 51 1266 2245 UTC 23 Sep 100 2300 UTC 24 Sep Typhoon Leo 1999 0500 UTC 30 Apr 61 970 36 3 1400 UTC 2 May 6 1800 UTC 2 May Maggie 1999 2300 UTC 5 Jun 67 969 36 145 1400 UTC 6 Jun 147 1800 UTC 8 Jun 1500 UTC 7 Jun Sam 1999 0600 UTC 21 Aug 80 975 33 265 0900 UTC 22 Aug 218 1400 UTC 24 Aug Durian 2001 1600 UTC 30 Jun 56 970 35 200 1900 UTC 1 Jul 107 0000 UTC 3 Jul Utor 2001 0000 UTC 5 Jul 64 965 35 563 0000 UTC 6 Jul 818 1600 UTC 7 Jul Yutu 2001 2300 UTC 23 Jul 61 974 34 1253 1600 UTC 25 Jul 726 1200 UTC 26 Jul Nari 2001 2100 UTC 17 Sep 75 985 28 284 0200 UTC 20 Sep 468 0000 UTC 21 Sep Krovanh 2003 0100 UTC 24 Aug 45 960 40 38 2200 UTC 24 Aug 13 2200 UTC 25 Aug Prapiroon 2006 0500 UTC 2 Aug 65 974 33 298 1120 UTC 3 Aug 151 2200 UTC 4 Aug Neoguri 2008 2300 UTC 17 Apr 43 959 40 42 0615 UTC 19 Apr 98 1800 UTC 19 Apr Nuri 2008 0000 UTC 21 Aug 54 965 38 3201 1500 UTC 22 Aug 5205 0600 UTC 23 Aug Molave 2009 1900 UTC 17 Jul 41 965 38 2179 1650 UTC 18 Jul 527 1200 UTC 19 Jul Chanthu 2010 0300 UTC 21 Jul 52 969 35 383 0545 UTC 22 Jul 401 0700 UTC 23 Jul Severe tropical York 1999 0000 UTC 14 Sep 78 975 30 1379 1000 UTC 16 Sep 335 storm 0600 UTC 17 Sep Maria 2000 1500 UTC 29 Aug 87 980 28 351 2000 UTC 31 Aug 531 0600 UTC 2 Sep Kammuri 2002 1600 UTC 2 Aug 86 980 28 266 2200 UTC 4 Aug 134 0600 UTC 6 Aug Vongfong 2002 2100 UTC 18 Aug 27 979 30 25 1200 UTC 29 Aug 22 0000 UTC 20 Aug Sanvu 2005 1300 UTC 12 Aug 24 980 30 1311 0445 UTC 13 Aug 2245 1300 UTC 13 Aug Pabuk 2007 1800 UTC 7 Aug 90 980 30 1221 1100 UTC 10 Aug 1597 1200 UTC 11 Aug Kammuri 2008 1700 UTC 4 Aug 70 979 25 1037 1415 UTC 6 Aug 132 1500 UTC 7 Aug Goni 2009 0600 UTC 3 Aug 150 970 28 787 0110 UTC 5 Aug 363 1200 UTC 9 Aug

Unauthenticated | Downloaded 10/03/21 11:14 AM UTC NOVEMBER 2012 Z H A N G E T A L . 3579

TABLE 1. (Continued)

Peak intensity Intensity level Pressure Wind Average flash Flash rate on 2 2 2 in lifetime Storm Year Begin/end Hours (hPa) (m s 1) rate (h 1) Landfall land (h 1) Tropical storm Wendy 1999 0200 UTC 3 Sep 43 990 20 59 2000 UTC 3 Sep 79 2100 UTC 4 Sep Mekkhala 2002 2000 UTC 24 Sep 82 991 24 9 0000 UTC 28 Sep 9 0600 UTC 28 Sep Nameless 2004 1400 UTC 26 Jul 16 989 23 60 0300 UTC 27 Jul 85 0600 UTC 27 Jul Jelawat 2006 1800 UTC 27 Jun 30 995 20 431 2340 UTC 28 Jun 200 0000 UTC 29 Jun Higos 2008 0200 UTC 3 Oct 76 998 18 201 0910 UTC 4 Oct 308 0600 UTC 6 Oct Nangka 2009 0100 UTC 26 Jun 23 988 24 320 1450 UTC 26 Jun 424 0000 UTC 27 Jun Soudelor 2009 1800 UTC 10 Jul 43 994 18 481 0020 UTC 12 Jul 31 1300 UTC 12 Jul lightning frequency in Severe Tropical Storm York (1999) lightning in the outer rainband decreased dramatically decreased by 76%, while in Typhoon Nuri (2008), it in- after the landfall. In storms at intensity of typhoon and creased by 63% after landfall (Table 1)]. severe typhoon, the outer rainband contained the vast Because of the changes in roughness between the majority of ground flashes before landfall. When the ocean and land surface, there is an increase in frictional storms were on land, flash density decreased rapidly in the convergence on the onshore side of storms as they ap- outer rainband but increased in the eyewall and inner proach the coast (Powell 1987). Figure 5 presents the rainband. Parrish et al. (1982) examined Hurricane radial distribution of flash density during the period of Fredric (1979) and found that convection in the eyewall pre- and postlandfall for five intensity categories. Light- increased by more than 50% when it approached the ning is mostly located in the eyewall and outer rainband coastline. The changes of convection would increase in storms at TD and TS intensity when over water. After lightning activity in the eyewall during the period of the storms make landfall, flash density decreased in the postlandfall. Figure 5 shows that the ground flash den- eyewall region but slightly increased in the inner and sity in the rainband decreased more rapidly in strong TCs outer rainband. A severe tropical storm showed a larger (STY) than in marginal TCs (STS and TY), but increased lightning frequency both in the eyewall and outer rain- in weak TCs (TD and TS) during the period of post- band then a tropical storm during prelandfall, but landfall. When storms strengthen to STS intensity after

FIG. 2. Center positions of the 33 tropical cyclones examined in FIG. 3. Radial distribution of CG flash density in five TC intensity this study for the periods of within 400 km of at least one locating levels, computed by averaging periods of storms with the same sensor. intensity.

Unauthenticated | Downloaded 10/03/21 11:14 AM UTC 3580 MONTHLY WEATHER REVIEW VOLUME 140

TABLE 2. The ratio of flash density between the eyewall and outer rainband and percentage of positive CG lightning for the eyewall and outer rainband of five TC intensity levels, computed by periods of storm with the same intensity as in Fig. 3. The eyewall region was within 60 km of the storm center and the outer rainband was 180–500 km from the center as described in section 2.

Intensity level Tropical depression Tropical storm Severe tropical storm Typhoon Severe typhoon Ratio of flash density 1:1 1:0.5 1:0.6 1:2.9 1:8.6 1CG% in eyewall 27.2% 17.4% 15.4% 30.4% 25.2% 1CG% in outer rainband 6.6% 11.6% 7.2% 8.6% 9.1% landfall, lightning in the outer rainband gradually outbreaks at approximately 7.1 h. The storm made weakened with the increase of intensity level. Our re- landfall at an average of 18.6 h after the eyewall sults indicate the difference of lightning activity within outbreaks in the intensification period (Table 3). the eyewall and outer rainband for storms at five in- 2) During the weakening stages of storm intensity. A tensity levels during the period of pre- and postlandfall. total of 27 outbreaks occurred during the periods of It is possible that the distributions are not fit for the weakened intensity: 7 in prelandfall and 20 in post- characteristic of every storm, but they give an overall landfall, which accounted for 43% of the total number picture of lightning activity in storms at different in- of outbreaks. The ratio of positive to total CG flashes tensity levels before and after landfall. of 17.1% was higher in weakening periods than in other periods (Table 3). Thomas et al. (2010), with the d. Eyewall lightning outbreaks analysis of North Atlantic hurricanes, also indicated an By using the method described in section 2c, a total of increase in the relative number of positive CG lightning 62 eyewall flash outbreaks were identified in the sample. in the inner core prior to and during periods of storm Table 3 shows the analysis results of flash outbreaks that weakening, and suggested a potential value of eyewall occurred in the eyewall region. Eyewall lightning bursts lightning for hurricane intensity change forecasting. occurred mainly in the following four time periods. 3) During the periods of maximum intensity. Nine out- breaks occurred during the times that encompassed 1) During the intensification period over water. Nine the MI and accounted for 15% of total bursts. The outbreaks occurred during the period of rapid deep- average ratio of positive to total CG flashes during ening before the storm landfall and accounted for the MI period was 10.3%. The time when storms 15% of all outbreaks. The peak value of maximum reached MI was at an average of 7.8 h ahead of land- sustained wind was preceded by the eyewall flash fall (Table 3).

FIG. 4. The average flash rates of each storm listed in Table 1 during the period of prelandfall and postlandfall.

Unauthenticated | Downloaded 10/03/21 11:14 AM UTC NOVEMBER 2012 Z H A N G E T A L . 3581

FIG. 5. Radial distribution of flash density during the period of pre- and postlandfall, computed by averaging periods of storms with the same intensity level: (a) tropical depression, (b) tropical storm, (c) severe tropical storm, (d) typhoon, and (e) severe typhoon.

Figure 6 shows an example of eyewall lightning time York reached MI, a third eyewall outbreak outbreaks during the above three periods. The figure occurred, but it was much smaller than the first two in displays the time variation of eyewall lightning, super- flash rates. York made its landfall in Guangdong at 2 imposed on hourly interpolations of the maximum 1000 UTC 16 September at the wind speed of 27 m s 1. 2 sustained wind speed for Severe Tropical Storm Then the winds dropped rapidly from 27 to 12 m s 1 York (1999). York produced a total of five eyewall in 20 h and the central pressure increased from 977 flash outbreaks during the period of study. The first to 1000 hPa. Two eyewall outbreaks occurred during two occurred during the rapid intensification which this weakening stage after the storm making landfall. began at 0000 UTC 14 September and ended at 0200 4) During the periods of recurvature. Eyewall outbreaks UTC 15 September. The maximum sustained surface were also observed during a steady period of storm 2 winds increased from 20 to 30 m s 1 over the 29-h with no changes of winds and pressure. The out- period of intensification, and then kept the peak wind breaks were associated with turnings of the storm 2 of 30 m s 1 for 26 h (i.e., the MI stage). During the track. In total, there were 10 outbreaks in 6 storms

Unauthenticated | Downloaded 10/03/21 11:14 AM UTC 3582 MONTHLY WEATHER REVIEW VOLUME 140

TABLE 3. Number of eyewall flash outbreaks, percentage of positive CG, and average hour of the outbreak to landfall in different stages for all the storms listed in Table 1.

Flash outbreaks 1CG% Hours to landfall Stage No. % Mean (min–max)With landfall No. of outbreak Mean (min–max) Strengthen 9 15 13.9 (3.0–51.9) Prelandfall 9 18. 6 (7–58) Weaken 27 43 17.1 (1.8–81.3) Prelandfall 7 16.9 (1–48) Postlandfall 20 7.9 (0–35) MI 9 15 10.3 (3.2–17.7) Prelandfall 9 7.8 (2–17) Steady Recurvature 10 16 12.8 (1.1–20.5) Prelandfall 7 25.6 (5–36) Postlandfall 3 15.3 (7–25) Uncertain 7 11 8.0 (1.9–21.0) Prelandfall 7 19.4 (3–44) Tot 62 100

[i.e., Typhoon Sam (1999), Typhoon Nari (2001), Severe (bursts 1, 2, and 3) at 0800, 1600, and 1700 UTC 4 October, Tropical Storm Pabuk (2007), Severe Typhoon respectively (Fig. 8a). These outbreaks occurred during Fengshen (2008), Tropical Storm Higos (2008), and the storm recurvature on 4 October, and the moving Severe Tropical Storm Goni (2009)] that occurred direction of the storm changed from north to northeast during this period, and accounted for 16% of all during this time (Fig. 8b). outbreaks. 4. Discussion Figures 7a,b show the temporal evolution of eyewall a. Variations of lightning pattern in TCs at different outbreaks and the storm track of Typhoon Nari (2001), intensity levels respectively. Nari experienced a total of six eyewall flash outbreaks during the period of study. The first two out- Radial distribution of flash density in TCs at different breaks (bursts 1 and 2) occurred during a steady period intensity levels shows that the maximum of lightning ac- 2 for the winds kept at 18 m s 1 and the pressure kept at tivity generally shifted from the eyewall region to outer ;998 hPa for 36 h (Fig. 7a). Actually these two outbreaks rainband as the storm strengthened. Our result showed were associated with the storm recurvature that began the ratios of flash density between the eyewall and outer at 0000 UTC 18 September and ended at 1200 UTC rainband were the highest (1:0.5) in TS and the lowest 19 September 2001 (Fig. 7b). In the period of study, Nari (1:8.6) in STY. When the intensity level strengthened made its largest turning angle at 0000 UTC 19 September, from TS to STY, radial flash density transferred from with the moving direction changed from southwest to bimodal distribution with maxima at the eyewall and west. The second outbreak at 2200 UTC and a great in- outer rainband to single peak distribution with maxima crease of flash rate at 2300 UTC 18 September presented 2 h ahead of the sharp turning. From Figs. 7a,b it was shown that the eyewall lightning became intense at the beginning of recurvature and increased to a flash peak 1 h ahead of the biggest turning point. Nari intensified from TS to STS during the period of 1200 UTC 19 September– 0000 UTC 20 September subsequently and the third and fourth outbreaks (bursts 3 and 4) occurred prior to and during this intensification. The storm intensity decreased rapidly from STS to TS and then to TD after Nari made landfall at 0200 UTC 20 September. The fifth and sixth eyewall outbreaks (bursts 5 and 6) were associated with this weakening stage. While Nari made an erratic looping track in the first few days from 6 to 15 September of its genesis, lightning data for this period were unavailable because of the detection range of the network. Figure 8 FIG. 6. Temporal evolution of lightning within 60 km of the center of Severe Tropical Storm York (1999) (bars) superimposed give another example of which eyewall flash breaks oc- on hourly interpolations of the speed curred during the period of recurvature. Severe Tropical (line with circles). Eyewall lightning outbreaks are indicated by Storms Higos (2008) experienced three eyewall outbreaks arrows with serial numbers.

Unauthenticated | Downloaded 10/03/21 11:14 AM UTC NOVEMBER 2012 Z H A N G E T A L . 3583

FIG. 8. As in Fig. 7 but for Severe Tropical Storms Higos (2008). The numbers next to the open circles indicate the dates from 29 Sep FIG. 7. Eyewall lightning outbreaks of Typhoon Nari (2001). (a) 2008. The period of study is between 0200 UTC 3 Oct and 0500 UTC Temporal evolution of eyewall lightning, superimposed on hourly 6 Oct 2008. maximum sustained wind speed. (b) The 6-hourly best-track po- sitions for the life cycle. The black dotted line denotes tropical depression status, the green dash–dot line denotes a tropical storm, was more electrified. Once a strong storm developed, the blue dashed line denotes a severe tropical storm, and the red solid line denotes a typhoon. Open circles mark the 0000 UTC lightning in the eyewall region became episodic and did positions of each day. The numbers next to the open circles indicate not occur for a long time (Molinari et al. 1999), thus, re- the dates in September 2001. Arrows with serial numbers in both sulting in the reduction of lightning density in the eyewall figures indicate the 6 times of eyewall outbreaks. The period of region. study is between 0000 UTC 18 Sep and 0000 UTC 21 Sep. A number of studies compared lightning activity in the eyewall with outer rainband. Molinari et al. (1999), at only the outer rainband and contained a higher flash using lightning data from NLDN, found that the ratio of density. Molinari et al. (1999) investigated prehurricane flash density between the eyewall and outer rainband in stages of three Atlantic basin storms and found that hurricane was ;1:2.0–6.0. In contrast to Molinari et al. none of them had the characteristic radial distribution (1999) who suggested that the outer bands dominated seen at the hurricane stage. Demetriades and Holle (2006) the lightning distribution, Cecil et al. (2002a,b) with a also indicated that lightning activity in tropical depres- study of 45 TCs with data from Lightning Imaging Sensor sions and tropical storms did not show a preferential spa- (LIS), found that the ratio was closer to 1:1. Squires and tial pattern, but these weaker systems generally produced Businger (2008) found the ratios were 1:0.2 and 1:1 for more lightning than hurricanes. While as the weakest Hurricanes Rita and Katrina, respectively, with LLDN storm, the ratio of flash density in TD was 1:1, not the data. Our result showed the ratios of flash density between highest among the five intensity levels. Although weak the eyewall and outer rainband were ;1:0.5–8.6. The dif- storms were unlikely to have a fully developed eyewall ferences among results of the above studies were partly structure, the core convection that occurred in these storms due to the large interstorm variability in flash density and

Unauthenticated | Downloaded 10/03/21 11:14 AM UTC 3584 MONTHLY WEATHER REVIEW VOLUME 140 the use of different lighting data and their limited range of with the highest value of 3201 flashes per hour and detection. lowest of 3 flashes per hour during the period of land- fall. The average flash rate did not present a positive b. Implications of eyewall flashes to storm intensity relationship with the peak storm intensity (maximum and recurvature sustained surface wind speed and minimum central Lightning activities that were positively relevant to pressures in the lifetime). convective strength could provide valuable information 2) Radial distribution of flash densities varied in storms on changes of intensity and track (Squires and Businger at different intensity levels. The maximum of light- 2008; Molinari et al. 1999). The development of convec- ning density generally shifted from the eyewall to the tive bursts within the eyewall could accompany signifi- outer rainband as the strength of storm intensity varied. cant changes of storm intensity (Heymsfield et al. 2001). The ratio of flash density between the eyewall and The eyewall lightning outbreaks over water always oc- outer rainband was largest in TS and lowest in STY. curred during times of intensification (Molinari et al. The percent of positive lightning within the eyewall 1994; Squires and Businger 2008), and hurricanes in a were higher than that in the outer rainband. Different quasi-steady state always showed weak lightning activity characteristics of positive and total lightning activity (Molinari et al. 1999). Our results showed that in all the in the eyewall and outer rainband might be due to 62 eyewall outbreaks, 15% occurred prior to or during convective regimes in tropical storms. the period of intensification and 15% occurred during 3) Spatial distribution of lightning varied from the period the MI (Table 3). Eyewall lightning outbreaks occurred of prelandfall to postlandfall. Ground flash density in during these circumstances were consistent with the past the rainband decreased more rapidly in strong TCs studies. A new result of our analysis indicated that a sig- (STY) than in marginal TCs (STS and TY), but in- nification portion (43%) of outbreaks occurred during the creased in weak TCs (TD and TS) during the period of weakening stages after the storm made landfall, with an postlandfall. When storms strengthened to STS inten- increase in the relative number of positive CG lightning. sity after landfall, lightning in the outer rainband grad- Moreover, eyewall lightning outbreaks (16%) could also ually weakened with the increase of intensity level. accompany with storm recurvatures. 4) Eyewall lightning could be a potential predictor to Some previous studies have discussed the mechanisms changes of storm intensity and track. An eyewall out- of lighting bursts during the period of storm strength- break in a weak and deepening storm could indicate ening. The vigorous updrafts that were enhanced during a rapid intensification. The outbreak of eyewall light- the period of intensification accelerated rates of charge ning was averagely 7.1 h before the peak of storm separation and charge concentration, thus promoting in- intensity before landfall. An outbreak in a mature, tense lightning activities within the eyewall (Lyons and steady storm could indicate it was about to reach the Keen 1994; Black and Hallett 1999). These suggested that MI. An increase in the percent of positive lightning in eyewall lightning outbreaks might be a useful forecast the eyewall during an outbreak may suggest a coming tool to predict imminent changes of storm intensity. Nev- period of storm weakening. An eyewall flash outbreak ertheless, questions about why lightning in the eyewall in a storm with steady state and little changes in wind bursts frequently during the weakening stages after land- speed and central pressure may indicate an imminent fall and what happens to the convective structure of recurvature. storms that caused eyeall outbreaks during the period of In the near future, we intend to combine our results of recurvature are not well understood yet. eyewall lightning outbreaks with data from land-based radar and the TRMM satellite to obtain the internal pre- 5. Conclusions cipitation structure for storms that experienced eyewall outbreaks, and to investigate the relationship between Cloud-to-ground lightning data from a regional light- eyewall lightning and convection morphology during the ning detection network in China have been used to ana- stages of storm weakening and recurvature. lyze lightning activities in tropical cyclones at different intensity levels during the period of landfall, and to in- vestigate the predictive value of eyewall flash outbreaks Acknowledgments. We thank Prof. Ying Li of the State to changes of intensity and trajectory in storms. The main Key Laboratory of Severe Weather, Chinese Academy of results of this study were summarized as follows: Meteorological Sciences and Prof. Shanchang Tao of the University of Science and Technology of China for their 1) Analysis of 33 tropical cyclones showed that the constructive suggestions to this study. We also thank two average flash rate varied largely from storm to storm, anonymous reviewers for their useful comments on this

Unauthenticated | Downloaded 10/03/21 11:14 AM UTC NOVEMBER 2012 Z H A N G E T A L . 3585 manuscript. This work was supported by the Basic Sci- https://ams.confex.com/ams/Annual2006/techprogram/paper_ entific Research and Operation Fund of Chinese Academy 105142.htm.] of Meteorological Sciences (2010Z004), the R&D Special Demetriades, N. W. S., and R. L. Holle, 2006: Long range lightning nowcasting applications for tropical cyclones. Preprints, 19th Fund for Public Welfare Industry (GYHY201006005), and Int. Lightning Detection Conf., Tucson, AZ, Amer. Meteor. the Natural Science Foundation of China (41005006). Soc., P2.15. [Available online at https://ams.confex.com/ams/ pdfpapers/99183.pdf.] ——, ——, S. Businger, and R. D. Knabb, 2010: Eyewall lightning REFERENCES outbreaks and tropical cyclone intensity change. Preprints, Abarca, S. F., K. L. Corbosiero, and D. Vollaro, 2011: The World 29th Conf. on Hurricanes and Tropical , Tucson, Wide Lightning Location Network and convective activity in AZ, Amer. Meteor. Soc., 16D.3. [Available online at https:// tropical cyclones. Mon. Wea. Rev., 139, 175–191. ams.confex.com/ams/29Hurricanes/techprogram/paper_168543. Biagi, C. J., K. L. Cummins, K. E. Kehoe, and E. P. Krider, 2007: htm.] National Lightning Detection Network (NLDN) performance Dun, G. E., 1951: Tropical cyclones. Compendium of Meteorology, in southern Arizona, Texas, and Oklahoma in 2003–2004. T. F. Malone, Ed., Amer. Meteor. Soc., 887–901. Fierro, A. O., and J. M. Reisner, 2011: High-resolution simula- J. Geophys. Res., 112, D05208, doi:10.1029/2006JD007341. tion of the electrification and lightning of Black, P. G., R. A. Black, J. Hallett, and W. A. Lyons, 1986: during the period of rapid intensification. J. Atmos. Sci., 68, Electrical activity of the hurricane. Preprints, 23rd Conf. on 477–494. Radar Meteorology and Conf. on Cloud Physics, Snowmass, ——, L. Leslie, E. Mansell, J. Strake, D. MacGormand, and CO, Amer. Meteor. Soc., J277–J280. C. Ziegler, 2007: A high resolution simulation of the micro- Black, R. A., and J. Hallett, 1986: Observation of the distribution physics and electrification in an idealized hurricane-like vor- of ice in hurricanes. J. Atmos. Sci., 43, 802–822. tex. Meteor. Atmos. Phys., 98, 13–33. ——, and ——, 1999: Electrification of the hurricane. J. Atmos. Sci., ——, X. M. Shao, T. Hamlin, J. M. Reisner, and J. Harlin, 2011: 56, 2004–2028. Evolution of eyewall convective events as indicated by intra- Cecil, D. J., and E. J. Zipser, 1999: Relationships between tropical cloud and cloud-to-ground lightning activity during the rapid cyclone intensity and satellite-based indicators of inner core intensification of hurricanes Rita and Katrina. Mon. Wea. convection: 85-GHz ice-scattering signature and lightning. Rev., 139, 1492–1504. Mon. Wea. Rev., 127, 103–123. Heymsfield, G. M., J. B. Halverson, J. Simpson, L. Tian, and T. P. ——, ——, and S. W. Nesbitt, 2002a: Reflectivity, ice scattering, Bui, 2001: ER-2 Doppler radar investigations of the eyewall and lightning characteristics of hurricane eye-walls and rain- of during the Convection and Moisture bands. Part I: Quantitative description. Mon. Wea. Rev., 130, Experiment-3. J. Appl. Meteor., 40, 1310–1330. 769–784. Holle, R. L., A. I. Watson, R. E. Lopez, D. R. MacGorman, ——, ——, and ——, 2002b: Reflectivity, ice scattering, and light- R. Ortiz, and W. D. Otto, 1994: The life cycle of lightning and ning characteristics of hurricane eye-walls and rainbands. Part severe weather in a 3–4 June 1985 PRE-STORM mesoscale II: Intercomparison of observations. Mon. Wea. Rev., 130, convective system. Mon. Wea. Rev., 122, 1798–1808. 785–801. Jorgensen, D. P., E. J. Zipser, and M. A. LeMone, 1985: Vertical Chen, L. W., Y. J. Zhang, W. T. Lu, Z. H. Huang, and J. Yu, 2010: motions in intense hurricanes. J. Atmos. Sci., 42, 839–856. Comparison analysis between location records of two LLS for Leary, A. L., and E. A. Ritchie, 2009: Lightning flash rates as an artificially triggered lightning. Proc. 30th Int. Conf. on Light- indicator of tropical cyclone genesis in the eastern North Pa- ning Protection, Cagliari, Italy, 1257. cific. Mon. Wea. Rev., 137, 3456–3470. Chen, S. M., Y. Du, L. M. Fan, H. M. He, and D. Z. Zhong, 2002: Lyons, W. A., and C. S. Keen, 1994: Observations of lightning in Evaluation of the Guang dong lightning location system with convective within tropical storms and hurricanes. transmission line fault data. IEE Proc. Sci. Meas. Technol., Mon. Wea. Rev., 122, 1897–1916. 149 (1), 9–16. ——, M. G. Venne, P. G. Black, and R. C. Gentry, 1989: Hurricane ——, ——, and ——, 2004: Lightning data observed with lightning lightning: A new diagnostic tool for tropical storm fore- location system in Guang-dong province, China. IEEE Trans. casting? Preprints, 18th Conf. on Hurricanes and Tropical Power Deliv., 19 (3), 1148–1153. Meteorology, San Diego, CA, Amer. Meteor. Soc., 113–114. China Meteorological Administration, 1999–2010: Yearbook of Marks, D. F., and L. K. Shay, 1998: Landfalling tropical cyclones: Tropical Cyclone (in Chinese). China Meteorological Press. Forecast problems and associated research opportunities. Bull. Cummins, K. L., and J. M. Murphy, 2009: An overview of lightning Amer. Meteor. Soc., 79, 305–323. locating systems: History, techniques, and data uses, with an Molinari, J., P. K. Moore, V. P. Idone, R. W. Henderson, and A. B. in-depth look at the U.S. NLDN. IEEE Trans. Electromagn. Saljoughy, 1994: Cloud-to-ground lightning in Hurricane An- Compat., 51, 499–518. drew. J. Geophys. Res., 99 (D8), 16 665–16 676. ——, M. J. Murphy, E. A. Bardo, W. L. Hiscox, R. B. Pyle, and ——, ——, and ——, 1999: Convective structure of hurricanes as A. E. Pifer, 1998: A combined TOA/MDF technology upgrade revealed by lightning locations. Mon. Wea. Rev., 127, 520–534. of the U.S. National Lightning Detection Network. J. Geo- Nagele, D., 2010: Analysis of cloud-to-ground lightning within phys. Res., 103 (D8), 9035–9044. tropical cyclones. M.S. thesis, Dept. of Geosciences, Texas ——, J. A. Cramer, C. Biagi, E. P. Krider, J. Jerauld, M. A. Uman, Tech University, 16 pp. and V. A. Rakov, 2006: The U.S. National Lightning De- Orville, R. E., and M. J. Coyne, 1999: Cloud-to-ground lightning in tection Network: Post-upgrade status. Preprints, Second tropical cyclones (1986–1996). Preprints, 23rd Conf. on Hur- Conf. on Meteorological Applications of Lightning Data, ricanes and Tropical Meteorology, Dallas, TX, Amer. Meteor. Atlanta, GA, Amer. Meteor. Soc., 6.1. [Available online at Soc., 194.

Unauthenticated | Downloaded 10/03/21 11:14 AM UTC 3586 MONTHLY WEATHER REVIEW VOLUME 140

Pan, L. X., X. S. Qie, D. X. Liu, D. F. Wang, and Y. Yang, 2010: The crystal/graupel collisions. J. Geophys. Res., 103 (D12), 13 949– lightning activities in super typhoons over the Northwest Pa- 13 956. cific. Sci. Chin. Ser. D, 53 (8), 1241–1248. Shao, X. M., and Coauthors, 2005: Katrina and Rita were lit up with Parrish, J., R. Burpee, F. Marks, and R. Grebe, 1982: Rainfall lightning. Eos, Trans. Amer. Geophys. Union, 86 (42), 398, patterns observed by digitized radar during the landfall of doi:10.1029/2005EO420004. Hurricane Fredric (1979). Mon. Wea. Rev., 110, 1933–1944. ——, M. Stanley, A. Regan, J. Harlin, M. Pongratz, and M. Stock, Pessi, A. T., S. Businger, K. L. Cummins, N. W. S. Demetriades, 2006: Total lightning observations with the new and improved M. Murphy, and B. Pifer, 2009: Development of a long-range Los Alamos Sferic Array (LASA). J. Atmos. Oceanic Tech- lightning detection network for the Pacific: Construction, nol., 23, 1273–1288. calibration, and performance. J. Atmos. Oceanic Technol., 26, Squires, K., and S. Businger, 2008: The morphology of eyewall 145–166. lightning outbreaks in two category 5 hurricanes. Mon. Wea. Powell, M. D., 1987: Changes in the low-level kinematic and Rev., 136, 1706–1726. thermodynamic structure of (1983) at land- Takahashi, T., 1978: Riming electrification as a charge generation fall. Mon. Wea. Rev., 115, 75–99. mechanism in thunderstorms. J. Atmos. Sci., 35, 1536–1548. Price, C., M. Asfur, and Y. Yair, 2009: Maximum hurricane in- Thomas, N. J., N. N. Solorzano, S. A. Cummer, and R. H. Holzworth, tensity preceded by increase in lightning frequency. Nat. Geo- 2010: Polarity and energetics of inner core lightning in three sci., 2, 329–332. intense North Atlantic hurricanes. J. Geophys. Res., 115, Rodger, C. J., J. B. Brundell, and R. L. Dowden, 2005: Location A00E15, doi:10.1029/2009JA014777. accuracy of VLF World-Wide Lightning Location (WWLL) Venne,M.G.,W.A.Lyons,C.S.Keen,P.G.Black,andR.C. network: Post-algorithm upgrade. Ann. Geophys., 23, 277– Gentry, 1989: Explosive growth: A possible indicator 290. for tropical storm intensification? Preprints, 24th Conf. on Radar ——,S.Werner,J.B.Brundell,E.H.Lay,N.R.Thomson,R.H. Meteorology, Tallahassee, FL, Amer. Meteor. Soc., 545–548. Holzworth, and R. L. Dowden, 2006: Detection efficiency of the Willoughby, H. E., 1990: Temporal changes of the primary circu- VLF World-Wide Lightning Location Network (WWLLN): lation in tropical cyclones. J. Atmos. Sci., 47, 242–264. Initial case study. Ann. Geophys., 24, 3197–3214. ——, D. P. Jorgensen, R. A. Black, and S. L. Rosenthal, 1985: Samsury, E., and R. E. Orville, 1994: Cloud-to-ground lightning in Project STORMFURY: A scientific chronicle 1962–1983. tropical cyclones: A study of Hurricanes Hugo (1989) and Bull. Amer. Meteor. Soc., 66, 505–514. Jerry (1989). Mon. Wea. Rev., 122, 1887–1896. Yokoyama, C., and Y. N. Takayabu, 2008: A statistical study on Saunders, C. P. R., and S. L. Peck, 1998: Laboratory studies of the rain characteristics of tropical cyclones using TRMM satellite influence of the rime accretion rate on charge transfer during data. Mon. Wea. Rev., 136, 3848–3862.

Unauthenticated | Downloaded 10/03/21 11:14 AM UTC