外文文献word

Atmospheric Research 79 (2006) 1 14Geographical and seasonal characteristics of therelationship between lightning ground flash densityand rainfall within the continent of AustraliaE.R. Jayaratnea, Y. Kuleshovb,*aGPO Box 2434, Brisbane, QLD 4001, AustraliabVictoria 3001, AustraliaReceived 12 January 2005; accepted 22 March 2005AbstractGround-based observations of annual rainfall and lightning incidence collected over periodsranging from 9 to 22 years at 23 stations around the continent of Australia were used to computevalues of drain yieldT, defined as the mass of rain produced per lightning ground flash (units: kgf l 1) over a given area of ground. The rain yield was found to vary considerably withgeographical location, season and climatic conditions. Of the 23 stations, 5 were mid-continentaland these showed a mean rain yield of 2.64 108 kg f l 1 in contrast to the coastal and near-coastal stations that showed a corresponding mean value of 9.91 108 kg f l 1. The difference wasstatistically significant at the confidence level of 95%. When the stations were classified accordingto seasonal climate zones, the winter and winter-dominant rainfall stations showed a rain yield of1.28 109 kg f l 1 while the summer and summer-dominant rainfall stations showed asignificantly lower value of 5.44 108 kg f l 1. Again the difference was statistically significantat the 95% confidence level. Every one of the 23 stations showed mean winter rain yields thatwere significantly higher than the summer values. These differences are attributed to surfaceheating which controls such parameters as cloud base height and convective available potentialenergy in the atmosphere. In terms of the behaviour of the rain yield with geographical, seasonal* Corresponding author. Tel.: +613 9669 4896; fax: +613 9669 4760.E-mail address: y.kuleshovbom.gov.au (Y. Kuleshov).0169-8095/$ - see front matter. Crown Copyright D 2005 Published by Elsevier B.V. All rights reserved.doi:10.1016/j.atmosres.2005.03.004International Laboratory for Air Quality and Health, Queensland University of Technology,National Climate Centre, Australian Bureau of Meteorology, GPO Box 1289K, Melbourne,2E.R. Jayaratne, Y. Kuleshov / Atmospheric Research 79 (2006) 114and climatic conditions, the Australian observations are in good agreement with studies in otherparts of the world.Crown Copyright D 2005 Published by Elsevier B.V. All rights reserved.Keywords: Lightning; Rainfall; Ground flash; Rain yield1. IntroductionThe close association between rain and lightning has been recognized since timeimmemorial. The Roman philosopher Lucretius observed the correlation in 58 BC andconcluded that thunder caused rainfall. Robert Hooke noted a relationship betweengushes of rain at the ground and overhead lightning in 1664. More recently, Lord Kelvinand Faraday both made references to the phenomenon and hypothesized that falling rainmay cause lightning. However, systematic scientific studies were not carried out until thelatter half of the last century. Battan (1965) and Piepgrass et al. (1982) related numbercounts of cloud to ground (CG) lightning from nearby thunderstorms to rain gaugereadings and found them to be well correlated. Many studies have found the intensity oflightning to be positively correlated to rainfall estimated from radar measurements(Kinzer, 1974; Reap and MacGorman, 1989; Williams et al., 1992; Cheze andSauvageot, 1997). Intense falls of rain associated with nearby CG lightning have beendocumented by Shackford (1960), Moore et al. (1962), Piepgrass et al. (1982) andJayaratne et al. (1995).2. Rain yieldsAlthough there is a high positive correlation between rainfall and lightning, the ratioof rain mass to CG lightning flash count over a common area, with units of kilogramsof rain per flash, quantitatively defined as the brain yieldQ, varies considerable withlocation. In general, heavy rain associated with monsoon or oceanic convection showrain yields of the order of 1091010 kg fl 1, while continental convective thunder-storms show much smaller values of 107108 kg fl 1 (Williams et al., 1992; Petersenand Rutledge, 1998). Zipser (1994) found that the number of thunder days associatedwith heavy rain in tropical monsoon and oceanic storm regions was significantly lowerthan that in continental rainfall regimes. Observing the results of several studies atvarious geographical locations, Petersen and Rutledge (1998) concluded that the rainyield varied by a factor of 10 or more at any given location and by a factor of up to103 between different locations and rainfall regimes. At the lower end, values ofaround 5 107 kg fl 1 were found in the arid south-western United States. A widesection of the mid-continental United States showed remarkably stable values clusterednear 108 kg fl 1, as did a landlocked station in Botswana within the Africansubcontinent. In tropical locations, the rain yields increased systematically from acontinental value of 4 108 kg fl 1 to a maritime value of 1010 kg f l 1 in the westernPacific Ocean. Williams et al. (1992) identified two distinct rainfall regimes in DarwinE.R. Jayaratne, Y. Kuleshov / Atmospheric Research 79 (2006) 1143in continental northern Australia. Rain yields for tropical continental break periodthunderstorms and tropical oceanic thunderstorms differed by almost an order ofmagnitude, being 3 108 kg fl1 and 2 109 kg fl1, respectively. Similarly, rainyields for break and monsoon period convection that occurred offshore over Melvilleand Bathurst Islands near Darwin showed values of 8 108 kg f l1 and 8 109 kgfl1, respectively.Williams et al. (1992) attributed the contrasting lightning activity in the two types ofrainfall regimes to differences in convective available potential energy (CAPE). Manystudies have demonstrated a strong increase in lightning activity with CAPE (Williams etal., 1992; Petersen et al., 1996). This is not surprising as CAPE bears a strong relationshipto the potential wet bulb temperature, Tw,a parameter that increases with temperatureand humidityboth of which lead to an increase in lightning activity (Williams andRenno, 1993). Williams et al. (1992) showed that a 1 8C change in Tw resulted in a changein CAPE of about 1 kJ kg1. The mean daily maximum surface Tw in Darwin during the19881989 wet season dropped by about 2 8C from the break periods to the monsoonperiods. Highly active lightning storms occurred during the break periods, while relativelylittle lightning was observed in the monsoonal storms. The mean values of CAPE duringthe two periods were 2000 J kg1 and 800 J kg1, respectively. The correspondinglightning flash rates observed over an area of 40,000 km2 were about 1000 and 100 perday, respectively.Continental land surface is systematically hotter than the sea. This gives rise to greaterCAPE, atmospheric instability and stronger air motions that are vital for deep convectionand thunderstorm formation. Although the total rainfall is about the same, lightningactivity over land is an order of magnitude greater than over the oceans (Orville andHenderson, 1986). Thus, maritime stations, in general, have a higher rain yield thancontinental stations.An alternative hypothesis for the landocean contrast in lightning is based ondifferences in boundary layer aerosol concentrations (Rosenfeld and Lensky, 1998).Continental air is more polluted than ocean air and contains more cloud condensationnuclei. Typical concentrations range from 100 to 200 cm3 over the oceans to valuesgreater than 1000 cm3 over land. The resultant larger numbers of smaller clouddroplets at continental locations give rise to a dominance of diffusional droplet growthand a suppressed coalescence. This leads to a reduction in rainfall and allows liquidwater to ascend to the higher mixed phase region of thunderclouds where strongelectrification takes place. The net result is increased lightning activity, reducedrainfall and reduced rain yields at continental stations when compared to maritimestations.A further possible explanation for the contrast in lightning and rainfall characteristicsbetween land and ocean thunderstorms is based on cloud base height (Williams andStanfill, 2002). They argue that higher cloud base heights provide larger updraught widthsand reduced dilution by mixingtwo factors that promote lightning activity. It has beenshown that lightning flash rate increases with cloud base height (Williams et al., in press).Typically, cloud base heights over the maritime and continental locations are about 500 mand 3000 m, respectively, and the associated lightning flash rates between these twolocations differed by an order of magnitude.4E.R. Jayaratne, Y. Kuleshov / Atmospheric Research 79 (2006) 1143. Lightning detectionThe Australian Bureau of Meteorology (ABM) maintains a network of about 40lightning sensors scattered widely around Australia. The sensor used is the CIGRE-500(CIGREInternational Conference on Large Electric Systems, 500 Hz ground-flashcounter). These counters are specifically designed to detect negative ground flashes andhave been used extensively to provide estimates of ground flash density (Barham, 1965;Prentice, 1972). The antenna used is a vertical aluminium tube of dimensions andelectrical characteristics conforming to CIGRE standards. The number of flashes isregistered on a mechanical counter that increments once for every flash detected. Multiplestrokes within a flash are eliminated by a 1 s dead time interval introduced by the circuitryafter every first-stroke. The best estimate of the effective horizontal range of the counter inAustralia is 30 km (Prentice and Mackerras, 1969) corresponding to a detection area of2827 km2.4. ClimatologyThe continent of Australia contains a diverse range of climatic zones. The tropicalnorthern and eastern coastal rim is generally humid and experiences heavy rainfall in thesummer. The continental interior is largely arid and the southern regions are mostlytemperate. The average rainfall in Australia is 450 mm. Around 80% of the landmass has amedian rainfall less than 600 mm per year with 50% less than 300 mm. Large arealpockets within South and West Australia have less than 150 mm. The vast interior of thecontinent has a median annual rainfall of less than 200 mm. This region is not normallyexposed to moist air masses for extended periods and rainfall is irregular. However, infavourable synoptic situations, which occur infrequently over extensive parts of the region,up to 400 mm of rain may fall within a few days and cause widespread flooding. Theregion with the highest annual rainfall is the east coast of Queensland near Cairns, withsome stations recording over 3000 mm per year.Owing to its low relief, compared to other continents, Australia causes little obstructionto the atmospheric systems that control the climate. However, as outlined earlier, therainfall pattern is strongly seasonal in character, with a winter rainfall regime in the southand a summer regime in the north. During the southern hemisphere winter (MayOctober),huge anticyclonic high pressure systems transit from west to east across the continent andmay remain almost stationary over the interior for several days. Northern Australia is thusinfluenced by mild, dry south-east winds, while southern Australia experiences cool, moistwesterly winds. During the winter, frontal systems passing from the west to the east overthe southern ocean have a controlling influence on the climate of southern Australia,causing rainy periods. In the summer months (November April), the anticyclones move ina more southerly track along the coast, directing easterly winds over the continent andproviding fine, hot weather in southern Australia. During this season, northern Australia isheavily influenced by the intertropical convergence zone. The associated intrusion ofwarm moist air gives rise to hot and humid conditions. Heavy rain may be prevalent for 2to 3 weeks at a time due to tropical depressions caused by monsoonal low-pressureE.R. Jayaratne, Y. Kuleshov / Atmospheric Research 79 (2006) 1145troughs. Thus, in contrast to the wet summer/dry winter typical of Darwin and Brisbane,Adelaide and Perth show the wet winter/dry summer pattern whereas Sydney, Melbourne,Canberra and Hobart show a relatively uniform pattern of rainfall throughout the year.A rainday is defined as a 24-h period, usually from 9 am to 9 am the next day, whenmore than 0.2 mm of rain is recorded. The frequency of raindays does not necessarilycorrelate well with the annual rainfall. For example, the frequency exceeds 150 per year inparts of the north Queensland coast where the annual rainfall is over 2000 mm, as well asin much of southern Victoria and in the extreme south-west of Western Australia where itis not more than about 600 mm. Over most of the continent the frequency is less than 50raindays per year. In the high rainfall areas of northern Australia, the number of raindays isabout 80 per year, but much heavier falls occur in this region than in the southern regions.5. MethodsAlthough most of the lightning sensors at the ABM stations have been operating for1020 years, not all of them have provided complete data sets; there being somesignificantly long gaps in the records at many sites, mainly due to instrument and batteryfailure. Of the 40 stations, 23 were selected for their reliability and availability of lightningand rainfall data over sufficiently long periods of observation and to represent a widegeographical distribution across the continent. These sites are shown on the map in Fig. 1.Table 1 lists the stations, arranged according to state. Station identification numbers aregiven in column 1. Columns 3, 4 and 5 show the latitude and longitude of each site and thenumber of years over which reliable data were available for use in this study.A method of deriving lightning ground flash density from CIGRE-500 counterregistrations has been described by Kuleshov and Jayaratne (2004). This method takes intoaccount the detection efficiency of the instrument and makes further corrections toeliminate a small number of falsely counted intracloud flashes and to include the smallfraction of ground flashes (4%6%) that carry a net positive charge and are not counted bythe instrument. Therefore, the ground flash density reported in this paper is the totalnumber of ground flashes (negative and positive) per square kilometre per year.The annual rainfall data were obtained from the ABM National Climate Centre archive.Readings were averaged over the number of years of data availability shown in column 5of Table 1.6. Results and discussion6.1. Lightning incidenceyrColumn 6 of Table 1 shows the mean annual ground flash density, N g in units of km 21Geraldton and Moora show the lowest lightning incidence (0.25 and 0.29, respectively),much lower than the dry mid-continental locations such as Tennant Creek (1.80) andKalgoorlie (0.94). The two north coast stations of Darwin and Kununurra have the highestfor each of the 23 stations. Note that the two coastal west Australian towns ofFig. 1. Map of Australia showing the six major seasonal rainfall zones and the locations of the 23 stations used in the analysis. Refer to Table 1 for the station identificationnumbers. The stations are listed according to rainfall zone in Table 2.6E.R. Jayaratne, Y. Kuleshov / Atmospheric Research 79 (2006) 114E.R. Jayaratne, Y. Kuleshov / Atmospheric Research 79 (2006) 114Table 1List of stations used in the study with associated details and data7Station ID1234567891011121314151617181920212223LocationGeraldton WAKalgoorlie WAMeekatharra WAMoora WAPerth WAPort Hedland WAThree Springs WAKununurra WADarwin NTTennant Creek NTCenter Is NTCeduna SAMt Isa QLDBrisbane QLDTownsville QLDBowraville NSWCobar NSWCoffs Harbour NSWLismore NSWNowra NSWBallarat VICMelbourne VICWhitlands VICLatitude(deg S)28.830.826.630.631.920.429.515.812.319.615.732.120.627.519.130.731.530.328.835.037.537.736.9Longitude(deg E)114.7121.5118.5116.0116.0118.6115.8128.7131.0134.2136.8133.7139.5153.0146.5152.8145.8153.1152.3150.5143.8144.8146.3Years ofdata22222017212216921209191915131218201714212120N g(km2 yr1)0.250.941.140.290.410.960.407.147.151.802.270.352.731.740.832.471.161.851.960.440.400.811.41Rainfall(mm yr 1)432.1291.5275.7463.0744.7319.2384.7705.31801.8453.7923.7273.4440.21118.61108.91159.0410.01663.51223.81118.0662.5518.91407.5Yield(kg f l 1)1.76E + 093.10E + 082.41E + 081.59E + 091.81E + 093.33E + 089.52E + 089.87E + 072.52E + 082.53E + 084.07E + 087.90E + 081.61E + 086.42E + 081.34E + 094.70E + 083.55E + 089.00E + 086.26E + 082.54E + 091.67E + 096.37E + 081.00E + 09See the map in Fig. 1 for the geographical locations of the stations on the continent.lightning ground flash densities (7.15 and 7.14, respectively), well above all the otherstations. It is interesting that Centre Island, which is also a north coast station but lies withinthe Gulf of Carpentaria, has an N g value of only 2.27 which is three times less than forDarwin and Kununurra. To confirm the veracity of this difference, we derived values of thetotal flash density, N t (ground flash + intracloud flash) from worldwide satellite remotesensing data gathered by NASA instruments for lightning detectionOptical TransientDetector (OTD) and Lightning Imaging Sensor (LIS) for these three localities. The N t valuesfor Kununurra, Darwin and Centre Island were 26.14, 22.87 and 8.14 f l km2 yr1,respectively. Thus, the total lightning activity as derived from the satellite data for Darwinand Kununurra is also three times higher than for Centre Island, which confirms our findings.Column 7 of the table gives the mean annual rainfall in mm. Column 8 gives the rainyield calculated in kg of rain per flash (kg f l1).6.2. Rain yield geographical distributionIn Fig. 2, we plot the mean annual rainfall of the stations against the mean annualground flash density. There is a positive linear correlation of r = 0.43 ( P = 0.04) betweenthe two parameters (best line not shown). However, what is more important is the apparentclustering of stations of similar climatic conditions. The two straight lines show the8E.R. Jayaratne, Y. Kuleshov / Atmospheric Research 79 (2006) 1141.E+101.E+091.E+080.11.010.0Ground Flashes (km-2 yr-1)Fig. 2. The rain yield plot showing the annual rainfall versus the ground flash density for the 23 stations. Stationidentification numbers refer to column 1 of Table 1. The five inland stations (Table 3) are shown as triangles (E).All other stations, shown as squares (n), are coastal or near-coastal. The two solid straight lines represent const