historical tropical cyclone activity and impacts in the ... · cyclones have occurred during el...

443

Historical Tropical Cyclone Activity and Impacts in the Cook Islands1

Fes A. de Scally2

Abstract: Analysis of a recently completed database of 143 tropical cyclones inthe Cook Islands revealed a minimum average frequency of 0.8 cyclones per cy-clone season between 1820 and 2006, with a more-precise frequency of 1.8 cy-clones per season with the beginning of satellite monitoring of cyclones in 1970.Since 1970, 31% of cyclones have reached hurricane intensity. The SouthernCooks have been more than twice as frequently affected by cyclones as theNorthern Cooks, with the island of Palmerston having the greatest numberof encounters. Since 1820, 96% of cyclones have occurred during the officialNovember–April cyclone season, with February alone accounting for 29%.Since 1970, 46% of cyclones achieving hurricane status have occurred in Febru-ary. Nevertheless, Cyclone Martin in October–November 1997 demonstratedthe dangers of a cyclone occurring outside the official season. An increase incyclone occurrences since the mid-1970s is probably attributable to satellitemonitoring, but it is noteworthy that all six cyclones known for certain to haveachieved major hurricane status have occurred since 2002. Since 1970, 56% ofcyclones have occurred during El Nino events, an increase of 15% from the1870–1969 period. Since 1891, cyclones with moderate and major human im-pacts have occurred on average at least every 3.8 and 8.8 yr, respectively, withthe Southern Cooks more than twice as frequently affected as the NorthernCooks. However, past cyclone disasters in the latter group suggest that risk tohuman life is greater there due to the potential for inundation of the atolls bystorm surges. Half of cyclones with human impacts have occurred during ElNino events, with weak to moderate El Ninos almost as important in this re-spect as strong El Ninos. Only 13% of cyclone impacts have occurred duringLa Nina events.

Historical records of natural hazard im-pacts are an important tool for disaster pre-paredness (Asian Development Bank 1991,Tobin and Montz 1997, Zillman 1999), espe-cially because human memory is frequentlymuch shorter than the recurrence intervalof many hazard phenomena (Granger 1999).Without such records the impact of a natural

hazard is extremely difficult to evaluate (To-bin and Montz 1997). Several printed reportsand Internet databases catalog historicaltropical cyclone (henceforth simply calledcyclone) activity in the Southwest PacificOcean, providing details on meteorologicalparameters such as tracks, intensities, andwind speeds. However, only a few provideany details at all about the cyclones’ impactson the inhabitants and communities of thisregion’s island nations. Even fewer provideany information at all about cyclones andtheir impacts in the Cook Islands specifically(Visher 1925, Visher and Hodge 1925, Kerr1976, Basher et al. 1992). As a result, al-though historical studies of cyclone impactshave been undertaken in some Pacific nations(e.g., Shaw 1981, Spennemann 2004), no up-to-date or even incomplete record of cycloneactivity exists for the Cooks. Yet this smallnation is regularly affected by these storms,

Pacific Science (2008), vol. 62, no. 4:443–459: 2008 by University of Hawai‘i PressAll rights reserved

1 Financial support for this research was providedby University of British Columbia Okanagan researchgrants. Manuscript accepted 8 October 2007.

2 Associate Professor of Geography, I. K. BarberSchool of Arts and Sciences, University of British Co-lumbia Okanagan, 3333 University Way, Kelowna, Brit-ish Columbia, Canada V1V 1V7 (e-mail: [email protected].).

and after recent calamities such as deadly Cy-clone Martin in October–November 1997and four destructive cyclones in February–March 2005 there is a recognition that a his-torical database of cyclones and their impactsis urgently required. The purpose of this pa-per is to briefly describe the recent comple-tion of such a database and then to providesome results from its analysis.

The Cook Islands consist of 15 small is-lands scattered over 2 million km2 of theSouthwest Pacific (Figure 1). The SouthernCooks, officially called the Southern Group,consist of the high volcanic island of Raro-tonga; almost-atoll of Aitutaki; makatea(upraised coral) islands of ‘Atiu, Ma‘uke,Mangaia, and Mitiaro; atolls of Palmerstonand Manuae; and coral cay of Takutea. The

Figure 1. Cook Islands location map, showing the 200-nautical-mile (370-km) Exclusive Economic Zone (EEZ).

444 PACIFIC SCIENCE . October 2008

Northern Cooks (Northern Group) islandsare all atolls with the exception of Nassau,a coral cay. The climate of both groups isdominated by the extensive and persistenttrade-wind zone of the South Pacific Ocean(Thompson 1986a,b). The position and in-tensity of the South Pacific ConvergenceZone (SPCZ), an area of convergence be-tween low-latitude easterly trade winds andthe higher-latitude southeast trades, have amajor influence on the weather including cy-clone formation, as discussed later. The dryseason occurs from May to October, and thewet season and large majority of cyclones oc-cur from November to April. The role of theEl Nino–Southern Oscillation (ENSO) phe-nomenon in cyclone occurrence is discussedlater. About 90% of the Cook Islands’ cur-rent population of 21,000 lives in the South-ern Group, with Rarotonga containing about70% of the total population. The islands ofManuae, Takutea, and Suwarrow are cur-rently uninhabited.

materials and methods

The analysis in this paper is based on a re-cently completed database of historical cy-clone activity and its impacts in the CookIslands (de Scally et al. 2006). This databasewas compiled from a wide variety of printedmeteorological reports, Internet cyclonedatabases and reports, and nonmeteorologicalsources of information. The most importantof the printed meteorological reports arethose of Visher (1925), Visher and Hodge(1925), Hutchings (1953), Kerr (1976), Revell(1981), Basher et al. (1992), Thompson et al.(1992), and cyclone reports of the Fiji Mete-orological Service. For more recent cycloneactivity, important Internet sources are thedatabases of the U.S. Navy (2004), Joint Ty-phoon Warning Center (2005), Herold et al.(2006), and Bath and Deguara (2007), as wellas the archived annual reports of the JointTyphoon Warning Center. An enormous va-riety of nonmeteorological sources were vi-tally important for obtaining information oncyclone impacts (fatalities, injuries, damageto infrastructure and crops, and other eco-

nomic and social impacts): they include suchdiverse material as memoirs of nineteenth-century London Missionary Society priests(e.g., Williams 1841, Sunderland and Buza-cott 1866, Gill 1894), annual reports of theCook Islands Administration published inthe Appendices to the Journals of the Houseof Representatives of New Zealand and dat-ing back to 1891, periodicals such as PacificIslands Monthly and Cook Islands News, and on-line reports of disaster-relief agencies such asthe International Federation of Red Crossand Red Crescent Societies, United NationsDepartment of Humanitarian Affairs, andOffice for Coordination of Humanitarian Af-fairs. The database itself contains a detaileddescription of the information sources usedfor creating it.

Two criteria were employed for includingcyclones in the database (de Scally et al.2006). First, any mention of impacts in anyof the islands of the Southern and NorthernGroups (Figure 1) warranted that cyclone’sinclusion in the database. This ensured theinclusion of damaging cyclones even if theydid not meet the second criterion. Becausethe intention was also to include cyclonesthat passed close to the Cook Islands whetheror not they had any impact, a second criterionwas employed: any cyclone spending a por-tion of its life, no matter how briefly, insidethe Cook Islands’ Exclusive Economic Zone(EEZ) was included in the database (Figure1). For simplicity the 200-nautical-mile (370-km) EEZ circles in Figure 1 ignore the EEZof adjoining jurisdictions and therefore prob-ably differ slightly from the official boundary.Nevertheless this provided an objective crite-rion by which to include cyclones in the data-base. A tropical cyclone for this purpose wasdefined as any nonfrontal synoptic-scale, cy-clonic system of tropical origin possessingsustained winds of at least gale force or 62km/hr (Holland 1983, Henderson-Sellers etal. 1998) (Table 1). If it could be determinedfrom best-track data that a cyclone’s sustainedwinds did not achieve gale intensity while in-side the EEZ, it was excluded from the data-base.

An in-depth discussion of the quality of

Tropical Cyclones in the Cook Islands . de Scally 445

the resulting record is contained within thedatabase (de Scally et al. 2006). Worth notinghere is the changing quality of cyclone datain the historical sources described earlier, in-cluding the differing criteria employed foridentifying cyclones, especially in the earlyrecords. More important, the availability oflimited satellite imagery in the South Pacificbefore 1980 and the nonavailability of imag-ery before 1970 means that the real numberof cyclones before the satellite era may neverbe known (Steve Ready, quoted by Maunder1995). In the database, the first recorded useof satellite imagery to identify and locate cy-clones occurs in December 1967, and so it isassumed that after 1970 all cyclones havebeen recorded. The lack of satellite data be-fore about 1970 also makes the classificationof cyclone intensity extremely difficult. Thelack of satellite observations can be partly cir-cumvented in populated archipelagos by theavailability of ground-based observations,allowing at least important cyclones to berecorded. Between 1891 and 1970, detailedreporting by the Cook Islands Administrationand periodicals such as Pacific Islands Monthlyand Cook Islands News ensures that most or allimportant or damaging cyclones are includedin the database. In contrast, the cyclone rec-ord before 1891 is patchy.

results

The cyclone database contains 166 entriesdating back to approximately 1600. Of this

total, 143 are classified as tropical cycloneswith a high degree of certainty; the remaining23 are excluded from the analysis in this pa-per. The potential value of Cook Islanders’oral histories for gaining insights into cy-clones before Europeans’ arrival is suggestedby three events in ca. 1590, 1665, and 1785as written down by Gill (1894) and Beagle-hole and Beaglehole (1938). The ca. 1590 cy-clone produced a storm surge that killedalmost all of the estimated 1,000 to 2,000 in-habitants of Pukapuka (Beaglehole and Bea-glehole 1938). However, given the difficultyof separating myth from fact in such oral his-tories, these cyclones are also excluded fromthe analysis. Cyclone entries subsequent tothese, which began in ca. 1820, include de-scriptions of several parameters. For manynineteenth-century cyclones this includesonly the date, a subjective classification of in-tensity, and brief descriptions of the originand track, although some also have informa-tion available on the specific islands affectedand the impacts. The impacts of some majorcyclones, including ones that struck Raro-tonga in December 1831 and March 1846,are described in great detail. For more recentcyclones the parameters for which informa-tion is available usually include the date,cyclone name and identification numbers,intensity category (Table 1), precise originand track, maximum sustained wind speedand occasionally gust speed, minimum baro-metric pressure, islands affected, impacts,and other miscellaneous information. Theanalysis reported in this paper focuses on spa-tial and temporal patterns of cyclone activity,and cyclone intensity and impacts.

With at least 143 cyclones having affectedthe Cook Islands between 1820 and 2006, aminimum average frequency of 0.8 cyclonesper cyclone season is derived, correspondingto a recurrence interval of 1.3 yr if recurrenceinterval is taken as the reciprocal of fre-quency (Table 2). Twenty-one percent ofthese are known with a high degree of cer-tainty to have achieved hurricane or majorhurricane status (Table 1), probably an un-derestimate given the lack of satellite moni-toring before 1970. In the satellite era after1970 65 cyclones have affected the Cooks,

TABLE 1

Classification of Tropical Cyclone Intensity (from Revell1981)

Equivalent MaximumSustained Winds

Beaufort ScaleDescription knots m/sec km/hr

Below gale (aforce 7)a <34 <17.2 <62Gale (force 8 or 9) 34–47 17.2–24.4 62–88Storm (force 10 or 11) 48–63 24.5–32.6 89–117Hurricane (force 12) >63 >32.6 >117Major hurricane >90 >46.3 >167

a Systems with winds below gale force are not consideredtropical cyclones.


giving a considerably higher and more preciseaverage frequency of 1.8 cyclones per seasonand corresponding recurrence interval of 0.6yr (Table 2). This includes 20 hurricanes ormajor hurricanes (31% of all cyclones), givingan average frequency of 0.6 per season (re-currence interval ¼ 1.8 yr) during the 1970–2006 period for cyclones achieving these in-tensities.

Figure 2 shows 104 cyclones since 1820 forwhich sufficient track information is avail-able for plotting. The dominant northwest tosoutheast track pattern just to the south ofthe Southern Group islands is apparent. Themajority of cyclones affecting the Cook Is-lands have originated in the region of Samoa,northern Tonga, and Wallis and Futuna, al-though many others have originated withinthe Northern Group and subsequentlytracked roughly southwest. Occasionally cy-clones have tracked to the Cooks from as farwest as the Solomon Islands.

Table 2 shows cyclone activity by groupbased on documented impacts on individualislands and, in some cases, effects inferredfrom the proximity of the cyclone track tospecific islands. The Northern Group tallydoes not include several tropical disturbancesand depressions that originated within thatgroup’s EEZ (Figure 1) but failed to achievetropical cyclone status until after they hadtracked into the Southern Group’s EEZ. Be-tween 1820 and 2006 at least 119 cyclonesaffected the Southern Group, giving a mini-mum average frequency of 0.6 cyclones per

season (recurrence interval ¼ 1.6 yr). Only42 cyclones are recorded to have affected theNorthern Group during that period, giving aminimum average frequency of 0.2 cyclonesper season (recurrence interval ¼ 4.4 yr).Since 1970, 58 cyclones have occurred in theSouthern Group and 25 in the NorthernGroup, giving average frequencies of 1.6(recurrence interval ¼ 0.6 yr) and 0.7 (recur-rence interval ¼ 1.4 yr) per season, respec-tively (Table 2). Palmerston (Figure 1) is theisland most frequently affected by cycloneactivity, as shown by the statistic in Table 2that is based on both reported effects and ef-fects inferred from the proximity of cyclonetracks to Palmerston (Figure 2).

Figure 3 shows the monthly distribution ofcyclone activity during the 1820–2006 period,with five of the 143 cyclones excluded be-cause the month of occurrence is unknown.In cases where a cyclone tracked from theend of one month into the next, it is countedas an occurrence in both months. The excep-tion to this rule is if the first day of cycloneoccurrence was on the last day of a month,or the last day of occurrence was on the firstday of a month: in this case the cyclone iscounted only in the other month, followingthe convention of Kerr (1976), Revell (1981),and Thompson et al. (1992). Ninety-six per-cent of all cyclones have occurred during theofficial Southern Hemisphere tropical cy-clone season from 1 November to 30 April,with February the single most active monthwith 29% of all cyclone occurrences (Figure

TABLE 2

Cyclone Occurrences for the Country as a Whole and by Group

1820–2006 1970–2006

AreaTotal

CyclonesAverage Cyclones

per Season R.I.aTotal

CyclonesAverage Cyclones

per Season R.I.a

Cook Islands 143 0.8 1.3 65 1.8 0.6Southern Groupb 119 0.6 1.6 58 1.6 0.6Northern Groupb 42 0.2 4.4 25 0.7 1.4Group unknown 6 — — 0 — —Palmerston 53 0.3 3.5 25 0.7 1.4

a Recurrence interval (yr).b Each group’s tally includes cyclones that affected both groups.


3). Of the 4% of cyclones outside the officialseason, the greatest number have occurred inOctober and May, with two cyclones in totalrecorded in each. The 34 cyclones that couldbe unambiguously categorized as hurricanesor major hurricanes on the basis of maximumsustained winds (Table 1) all occurred during

the December–April period. February isagain the most active month for cyclones inthese categories, with just over 38% of all oc-currences (Figure 3). Figure 3 also shows themonthly distribution of the 65 cyclones and20 hurricanes/major hurricanes that have oc-curred since 1970. Although the percentages

Figure 2. Generalized tracks of 104 cyclones in the Cook Islands, 1820–2006. The origin of cyclones tracking fromwest of 175� W longitude is not shown, and many cyclones tracked east or south out of the map area before decayingor undergoing extratropical transition.


of cyclones falling inside and outside the offi-cial tropical cyclone season are almost un-changed from the whole 1820–2006 period(97 and 3, respectively), December–Januaryshow a slightly smaller proportion of cy-clones and hurricanes/major hurricanes, andFebruary–April show a slightly larger pro-portion. February accounts for 33% of allcyclones and 46% of hurricanes/major hur-ricanes since 1970; the latter figure is 8%higher than for the 1820–2006 period ofrecord.

Figure 4 shows the annual number of cy-clones for the 1820–2006 period. It must beemphasized again that the early record is in-complete, and only after the reporting of cy-

clone impacts began in the annual reports ofthe Cook Islands Administration in 1891 andin Pacific Islands Monthly in 1930 does the rec-ord of important cyclones become reasonablycomplete. The record for more minor cy-clones is probably not complete until thebeginning of the satellite era, as discussedearlier. Figure 5 summarizes the frequencyof seasonal cyclone numbers for the 1820–2006 and 1970–2006 periods. The dispropor-tionately large number of cyclone seasonswith one cyclone occurrence in the whole1820–2006 record reflects to an unknownextent the early years when usually only land-falling cyclones could be observed. In con-trast, the lower proportion of seasons with

Figure 3. Monthly distribution of cyclones in the Cook Islands, 1820–2006. The criteria for the hurricane andmajor hurricane categories are shown in Table 1. Short horizontal lines inside or above the bars show 1970–2006percentages.


one or two cyclones and correspondinglyhigher proportion with three or more cy-clones since 1970 probably reflects the impactof satellite observation, when presumably allcyclones have been detected. Since 1970,only nine seasons have experienced no cy-clone activity in the Cook Islands. The maxi-mum number of cyclones in the Cooks’ EEZin one season (six) occurred in the 1980–1981, 1997–1998, and 2004–2005 seasons.

Table 3 shows El Nino–Southern Oscilla-tion (ENSO) episodes since 1870 that haveoccurred during cyclone seasons with ob-served cyclones. El Nino and La Nina eventsare identified using the Nino3.4 sea-surfacetemperature (SST) index, which is the depar-ture in monthly SST (relative to the 1961–1990 mean) averaged over the Nino3.4 re-gion and available from the United KingdomMet Office’s Hadley Centre. An event isidentified if the five-month moving averageof the index exceeds þ0:4�C for El Ninosand �0.4�C for La Ninas for at least six con-secutive months (Trenberth 1997, Interna-tional Research Institute for Climate andSociety 2006). This method is employed by

New Zealand’s National Institute of Waterand Atmospheric Research, among others, toidentify El Nino and La Nina events in theSouthwest Pacific (2007 unpublished reporton ENSO event classification, National Cli-mate Centre, Auckland). Table 4 shows thatthe proportion of total cyclones that have oc-curred during El Nino events has increasedover time, with 56% of cyclones having oc-curred during El Ninos since 1970 comparedwith only 41% for the 1870–1969 period.Table 5 shows that the relationship betweenthe seasonal number of cyclones and ElNino strength is equivocal, with strong ElNinos associated with slightly more cyclonesover the 1870–1969 period but slightly fewercyclones since 1970.

The association between La Nina eventsand cyclone occurrences is also shown inTables 3 and 4. Table 4 shows that the pro-portion of total cyclones that have occurredduring La Nina events has decreased overtime, dropping from 17% over the 1870–1969 period to 9% since 1970. The propor-tion of cyclones that have occurred duringneutral ENSO conditions has also decreased

Figure 4. Number of cyclones per season, 1820–2006.


by 8%. No attempt is made in Table 3 toidentify El Nino or La Nina seasons with nocyclone activity, because before 1970 the cy-clone record should be regarded as incom-plete. After 1970, no El Nino events havecoincided with a lack of cyclone activity inthe Cook Islands, although El Nino condi-tions did develop at the end of the quiet1993–1994 cyclone season.

Table 6 shows cyclone occurrences withminor, moderate, and major human impactsin the Cook Islands. ‘‘Minor’’ is defined ashaving resulted in no fatalities or injuries; mi-nor damage to infrastructure or disruptions topower, telephone, and flights; or minor crop

losses; and effects on only one or two islands.‘‘Moderate’’ is defined as having resulted inno fatalities but possibly some minor injuries;or substantial but not widespread damageto infrastructure; or moderate to substantialcrop losses; and effects on a small number ofislands. ‘‘Major’’ is defined as having resultedin one or more fatalities or a substantial num-ber of injuries; or substantial and widespreaddamage to or destruction of infrastructureand widespread crop losses; or a declarationof a state of emergency or request for inter-national assistance. For cyclones where de-tailed information on impacts is unavailable,the damage classifications of Visher (1925),

Figure 5. Frequency of seasonal cyclone totals. Numbers above the bars show the percentage of total seasons (seasonswith no cyclones are excluded from the 1970–2006 calculations).


TABLE 3

El Nino–Southern Oscillation Events since 1870 inTropical Cyclone Seasons with Observed Cyclone

Activity; Events Are Defined by Sea-SurfaceTemperature (SST) Anomalies in the Nino3.4 Region

(See Text)

EventType

CycloneSeason

SST Anomaly inMonth of CycloneOccurrence (Averageif >1 Month) (�C)

El Nino 1876–1877 þ0.491888–1889 þ1.51a

1896–1897 þ1.18a

1902–1903 þ0.53a

1904–1905 þ0.911905–1906 þ0.851913–1914 þ0.841925–1926 þ1.47a

1930–1931 þ1.46a

1940–1941 þ1.46a

1941–1942 þ0.901957–1958 þ1.52a

1963–1964 þ0.831965–1966 þ1.04a

1969–1970 þ0.381972–1973 þ1.40a

1976–1977 þ0.391977–1978 þ0.541982–1983 þ2.16a

1986–1987 þ1.18a

1987–1988 þ0.401991–1992 þ1.66a

1994–1995 þ1.19a

1997–1998 þ2.04a

2002–2003 þ0.80a

2004–2005 þ0.35La Nina 1886–1887 �1.01

1890–1891 �0.841938–1939 �0.791942–1943 �0.991949–1950 �1.481950–1951 �0.791954–1955 �0.591955–1956 �1.161964–1965 �1.051973–1974 �0.831984–1985 �0.821988–1989 �1.331998–1999 �1.092000–2001 �0.58

a Strong El Nino event as defined by the 5-month runningmean of SST anomalies exceeding þ1:00�C for three consecutivemonths.

TABLE 4

Cyclone Occurrences during El Nino, La Nina, andNeutral ENSO Conditions as Defined by SSTAnomalies in the Nino3.4 Region (See Text)

No. of CycloneOccurrences (%)

1870–1969 1970–2006

El Nino events 28 (41) 36 (56)Neutral conditions 29 (42) 22 (34)La Nina events 12 (17) 6 (9)Total number of cyclones 69 64

TABLE 5

Cyclone Occurrences during Weak to Moderate andStrong El Nino Events

No. ofSeasons

TotalCyclones

AverageCyclonesper Season

1870–1969All El Ninos 15 28 1.9Weak to moderate

El Ninosa7 12 1.7

Strong El Ninosa 8 16 2.01970–2006All El Ninos 12 36 3.0Weak to moderate

El Ninosa4 13 3.3

Strong El Ninosa 8 23 2.9

a Five-month running mean of SST anomalies for threeconsecutive months: weak to moderate < þ1:00�C; strong >þ1:00�C.

TABLE 6

Number of Cyclones with Observed or Inferred HumanImpacts

1891–2006

Type ofImpact

No. ofCyclones

No. ofCyclones

AverageCyclonesper Season

RecurrenceInterval (yr)

1820–2006

Minor 38 (33a) 35 (33a) 0.3 (0.6b) 3.3 (1.7b)Moderate 32 30 0.3 3.8Major 21 13 0.1 8.8

a Additional cyclones with inferred minor impacts (see text).b Includes cyclones with inferred minor impacts.

Kerr (1976), and Basher et al. (1992) are em-ployed. During the 1820–2006 period 91 cy-clones had an impact in the Cook Islands(Table 6). A further 33 cyclones are inferredto have resulted in minor impacts based onthe proximity of their tracks to inhabited is-lands, because such minor effects frequentlywent unreported. Taking the 1891–2006 pe-riod when much more detailed records ofcyclone impacts are available, an average ofat least 0.3 cyclones per season (recurrenceinterval ¼ 3.3 yr) have had minor impacts,0.6 cyclones per season (recurrence interval¼ 1.7 yr) if those with inferred minor impactsare included (Table 6). On average at least0.3 (recurrence interval ¼ 3.8 yr) and 0.1 (re-currence interval ¼ 8.8 yr) cyclones perseason have created moderate and major im-pacts, respectively (Table 6). Between 1820and 1890, 10 cyclones with moderate or ma-jor impacts were recorded in the SouthernGroup but none in the Northern Group.This discrepancy may reflect the greater pres-ence of Europeans in the Southern Group atthat time, particularly London MissionarySociety priests who made written observa-tions of cyclones and their impacts (see, forexample, Williams 1841, Sunderland and Bu-zacott 1866, Gill 1894). During the 1891–2006 period when more complete records ofcyclone impacts are available for both groups,at least 19 cyclones with moderate impactsand six with major impacts were reported inthe Southern Group, representing 58% ofall cyclones. In the Northern Group at leasttwo cyclones with moderate impacts and twowith major impacts (Cyclone Martin in 1997and Cyclone Percy in 2005) were reported,representing 9% of all cyclones. Nine othercyclones with moderate impacts and fiveothers with major impacts are reported tohave affected both groups during the 1891–2006 period.

Table 7 shows the relationship betweencyclone impacts and ENSO conditions. Since1870 El Nino conditions have prevailed dur-ing 44, 63, and 47% of cyclones producingminor, moderate, and major impacts, respec-tively. Although strong El Ninos were morecommon during cyclones with minor to mod-erate impacts compared with weak to moder-

ate El Ninos, weak to moderate El Ninoswere slightly more common during cycloneswith major impacts. Table 7 also shows that33–40% of cyclone impacts have occurredduring neutral ENSO conditions, and only3–17% have occurred during La Nina events.

Table 8 lists all reported fatalities from cy-clones in the Cook Islands, with the earliestreference being the destruction of almost theentire population of Pukapuka by a stormsurge as described earlier. This record offatalities is probably patchy until the detailedannual reporting by the Cook Islands Admin-istration began in 1891.

discussion

The cyclone track pattern illustrated in Fig-ure 2 reflects the broader geographical distri-bution of cyclones in the Southwest Pacific(Kerr 1976, Revell 1981, Thompson et al.1992). Specifically, the dominant northwestto southeast pattern of cyclone tracks just tothe south of the Southern Group, as well asa secondary pattern of roughly southwesterlymotion originating within the NorthernGroup agree closely with Kerr’s (1976)detailed analysis of cyclones’ directionalmovement in this region. The dominantnorthwest-southeast pattern results from thefact that the SPCZ is often situated over theSouthern Group during the cyclone season(Thompson 1986a,b, Graham et al. 1989,Sturman and McGowan 1999), with cyclones

TABLE 7

Occurrence of Cyclone Impacts during El Nino, LaNina, and Neutral ENSO Conditions, 1870–2006 (SeeTables 3–5 for Criteria Employed for Defining ENSO

Phases)

No. of Cyclone Impacts (%)

Type ofImpact

Weak toModerateEl Nino

StrongEl Nino Neutral La Nina

Minora 12 (17) 19 (27) 27 (39) 12 (17)Moderate 7 (23) 12 (40) 10 (33) 1 (3)Major 4 (27) 3 (20) 6 (40) 2 (13)

a Includes observed and inferred minor impacts (see text).


frequently developing in the zone of cyclonicwind shear that characterizes the SPCZ(Thompson 1986a,b). The lower frequencyof cyclone activity in the Northern Group(Figure 2 and Table 2) results from the usuallocation of the SPCZ to the west and southof this group during the cyclone season.

The average seasonal frequencies of cy-clones in the Southern and Northern Groupsduring the 1820–2006 period (0.6 and 0.2, re-spectively [Table 2]) are minimum estimatesbecause the cyclone record on which theyare based is incomplete for the nineteenthand early twentieth centuries. Furthermore,the historically larger population and longerEuropean presence (and thus written records)in the Southern Group, especially Rarotonga,probably exaggerates the difference betweenthe two groups. Thompson (1986a,b) re-ported identical average frequencies for the1940–1984 period, which indicates that therecords employed by him for the calculationsare probably also incomplete; this is rein-forced by the fact that during 1940–1984 herecorded only 11 cyclones for the NorthernGroup, but de Scally et al. (2006) recorded atleast 22 cyclones. The average seasonal fre-quencies of cyclones in the two groups over

the 1970–2006 period (1.6 and 0.7, respec-tively [Table 2]) are in good agreement withpreviously published estimates of 1.4 and 0.8for the same period (Salinger et al. 2004).Given the more or less complete coverage ofcyclone activity by weather satellites duringthis period, these figures are probably an ac-curate reflection of the true frequencies ofcyclone occurrence (Thompson 1986b). Themuch higher frequency of cyclones in theSouthern Group is related to the dominantnorthwest-southeast pattern of tracks just tothe south of this group (Figure 2), as dis-cussed earlier. The proximity of Palmerstonto this track pattern is underscored by thefact that this island experienced more cy-clones during the 1820–2006 period than allof the Northern Group combined, and anequal number since 1970 (Table 2).

The vast majority of historical cyclones(96–97%) have occurred during the officialNovember to April Southern Hemisphere cy-clone season, reflecting the pattern elsewherein the Australia–Southwest Pacific region(Henderson-Sellers et al. 1998). All hurri-canes and major hurricanes have occurredduring a narrower December–April period(Figure 3). However, complacency or a lack

TABLE 8

Cyclones with Reported Fatalities

Cyclone Island No. of Fatalities

ca. 1590 Pukapuka Est. 1,000–2,000December 1831 Rarotonga ‘‘Hundreds’’ from ensuing starvation,

with almost 5,000 more by 1847 fromdisease due to food deprivation

December 1842 Mangaia 1 (on ship)March 1866 Mangaia ‘‘Many deaths’’ from ensuing starvationMarch–April 1869 Mangaia Est. 11 (on ship)ca. 1872 Mitiaro 2December 1883 Rarotonga 7 (on three ships)November 1890 Aitutaki 14 (on ship and two canoes)January 1905 Between Ma‘uke

and ‘Atiu1 (on ship)

January 1914 Rakahanga 1March 1926 Palmerston 1February 1935 Rarotonga 1 plus a ‘‘high number’’ of tuberculosis

fatalities by 1937 due to food deprivationMarch 1943 Rarotonga? 1Tropical Cyclone Peni, February 1990 Unknown 1Tropical Cyclone Martin, November 1997 Manihiki 19


of preparedness outside the official cycloneseason can have serious consequences, asdemonstrated in Manihiki by Cyclone Martinin late October–early November 1997, when19 people perished, 200 were injured and 700more were affected, and 90% of propertieswere destroyed or damaged by a 10-m stormsurge and waves. The human toll in this di-saster was attributed partly to the lack oftimely warnings to the inhabitants (Fiji Mete-orological Service 1999, Syme-Buchanan1997). The monthly distribution of cyclones,including hurricanes and major hurricanes,appears to have shifted since 1970 (Figure3), with slightly less activity in December–January and slightly more in February–Aprilcompared with the complete 1820–2006 pe-riod. Although this shift is small overall, an8% increase in hurricanes/major hurricanesin February is noteworthy.

Figures 4 and 5 show an apparent increasein the number of seasons with three or morecyclones since the mid-1970s, but this is al-most certainly attributable to the beginningof satellite monitoring of cyclones. Otherstudies have found no discernible trend inthe frequency of cyclones in the SouthwestPacific since 1970 (Henderson-Sellers et al.1998, Emanuel 2005). What is not apparentin Figures 4 and 5 is the fact that all six cy-clones known to have reached major hurri-cane status (Table 1) in the Cook Islandssince 1970 have occurred in the last four sea-sons: Cyclone Dovi in 2002–2003, CycloneHeta in 2003–2004 (which tracked just tothe west of the Cooks’ EEZ), and CyclonesMeena, Nancy, Olaf, and Percy in 2004–2005 (it is uncertain if Cyclone Ima in 1985–1986 became a major hurricane while insidethe EEZ). On the Saffir-Simpson scale of in-tensity (National Oceanic and AtmosphericAdministration 2007), Cyclones Dovi, Meena,and Nancy achieved a category four, and Cy-clones Heta, Olaf, and Percy achieved a cate-gory five, the highest possible. This incidenceof major hurricanes in recent years may re-flect the fact that between 1975–1989 and1990–2004, the number of category four andfive hurricanes in several ocean basins in-creased by 120%, and in the Southwest Pa-cific the proportion of these making up the

total number of cyclones increased from 12to 28% (Webster et al. 2005). Emanuel(2005) also detected a marked increase in thedestructiveness of cyclones in the North Pa-cific since the mid-1970s. These increases,however, cannot be attributed to globalwarming without a longer record and a betterunderstanding of the role of cyclones in thegeneral circulation of the atmosphere andocean (Webster et al. 2005). Nor can theeffect of improved satellite monitoring andanalysis in the last few years be discounted asa contributing factor (Diamond 2006).

Although some studies have failed to finda statistically significant association betweenthe ENSO phenomenon and Southwest Pa-cific cyclone activity (Ramage and Hori 1981,Hastings 1990), many have demonstrated asubstantial eastward shift in Pacific cycloneactivity during El Nino conditions as a re-sult of changes in sea-surface temperatureand tropospheric winds (Revell and Goulter1986, Hastings 1990, Evans and Allan 1992,Basher and Zheng 1995, Terry et al. 2001,Elsner and Liu 2003). This eastward shifthas the effect of extending cyclone activity ina wide band between 10� S and 25� S to as fareast as the Cook Islands and French Polyne-sia, with the greatest incidence around thedate line and east-southeast of Fiji (Basherand Zheng 1995). In contrast, during LaNina conditions little cyclone activity extendseast of the date line (Basher and Zheng1995). For both the Southern and NorthernGroups of the Cook Islands, El Nino condi-tions are therefore accompanied by an in-creased probability of cyclones (Thompson1986a,b). Salinger et al. (2004), for example,estimated increases in cyclone probability of36–50% for the Southern Group and 13–25% for the Northern Group during evenweak El Nino conditions, relative to the aver-age probability for all years. This pattern isreflected in the much greater number of cy-clones in our database that have occurredduring El Nino conditions compared withneutral ENSO and La Nina conditions (Ta-bles 3 and 4). Six of the eight seasons withfour or more cyclones each have occurredduring El Nino events; the exceptions are1943–1944 with four cyclones and 1980–


1981 with six cyclones, both during neutralENSO conditions. Also of interest is the15% increase in cyclone occurrences duringEl Nino conditions and corresponding 16%decrease during neutral and La Nina condi-tions since 1970 compared with the 1870–1969 period (Table 4). This at least partly re-flects the greater frequency of El Nino eventssince about 1990 (McGuire et al. 2002). It issurprising that the average seasonal numberof cyclones is affected little by the strengthof the El Nino events; in fact since 1970 theaverage is higher for weak to moderate ElNinos than for strong ones (Table 5). Thiswas dramatically illustrated by the modest ElNino conditions of 2004–2005 when a recordsix cyclones affected the Cook Islands. Theresults in Table 4 also support the observa-tion by Thompson (1986b) and Henderson-Sellers et al. (1998) that cyclone-free seasonstend to coincide with La Nina conditions;only a small proportion of cyclone occur-rences in the Cooks are associated with LaNina events, with this proportion having de-creased from 17% in the 1870–1969 periodto 9% since 1970. In contrast Hastings (1990)found that in five cyclone seasons with strongEl Nino conditions and three with strong LaNina conditions, 36% of cyclones occurredduring the latter. This may reflect the greaterprobability of cyclones off the east coast ofAustralia during La Nina conditions (Mc-Guire et al. 2002).

The record of cyclone impacts since 1891shows that, on average, the Cook Islands havesuffered minor impacts about every 2 yr,moderate impacts about every 4 yr, and majorimpacts about every 9 yr (Table 6). Thesefrequencies should be viewed as minima be-cause it is not certain whether all cycloneswith impacts are captured in the record. Thestatistics show a major difference between theSouthern and Northern Groups in terms ofimportant (moderate to major) impacts: dur-ing the 1891–2006 period, 58% of cycloneswith important impacts have struck the seveninhabited islands of the Southern Group, andonly 9% of this kind of cyclone have struckthe five inhabited islands of the NorthernGroup. In 33% of cases the cyclones resulted

in substantial impacts in both groups. There-fore, notwithstanding some possible biastoward reported impacts in the SouthernGroup, it appears that this group is at morethan twice the risk of significant cyclone im-pacts than the Northern Group. Given thedefinitions for moderate and major impactsemployed in this paper, this difference inrisk levels can be ascribed mostly to the sub-stantially higher probability of cyclone occur-rence in the Southern Group (Figure 2 andTable 2). In the Northern Group, PenrhynAtoll is generally regarded as being outsidethe cyclone belt, but our database shows itto have been affected by three cyclones withimportant impacts and four to eight cycloneswith minor impacts.

Table 7 shows that approximately half ofcyclone impacts have occurred during ElNino events, roughly in proportion to thefrequency of cyclones occurring during ElNinos (Table 4). Strong El Ninos are moreclosely associated with minor to moderate cy-clone impacts, which is strange given the lackof any notable relationship between El Ninostrength and cyclone frequency (Table 5).Major cyclone impacts have occurred morefrequently during weak to moderate El Ninosthan during strong ones (Table 7). Only avery small proportion of cyclone impacts areassociated with La Nina events (Table 7).

Table 8 shows that although CycloneMartin in 1997 was the deadliest cyclone inthe Cook Islands in over a century, calamitiessuch as this pale into relative insignificancecompared with historic death tolls such as re-corded in Pukapuka in ca. 1590, Rarotonga in1831, and Mangaia in 1866. Many of thefatalities in these historic disasters resultedfrom famine when food crops were destroyedand, after the arrival of Europeans early inthe nineteenth century, also from illness dueto decreased resistance to introduced diseasesfrom hunger. With the notable exception ofCyclone Martin, few fatalities have occurredin modern cyclones due to improved warningsystems, use of evacuation and cyclone shel-ters, and provision of relief supplies by seaand air. This trend toward reduced mortalitymirrors that in developed countries such as


the United States and Australia (Henderson-Sellers et al. 1998, Smith 2004). On the otherhand, the monetary value of damage fromcyclones in the Cook Islands appears to beescalating, mirroring another trend apparentalong tropical coasts undergoing rapid de-velopment (Henderson-Sellers et al. 1998,Smith 2004). For example, damage estimatesfor recent major cyclones (unadjusted for in-flation) include US $20–40 million in Cy-clone Sally (1986–1987), US $8 million inCyclone Martin (1997), and over US $25 mil-lion in Cyclones Meena, Nancy, Olaf, andPercy (2005). Such losses can have a substan-tial impact on the Cook Islands’ small econ-omy: the damage from the 2005 cyclones, forexample, represents at least 14% of the coun-try’s gross domestic product at the time.

With respect to future cyclones, Raro-tonga appears to be most at risk of costlydamage given the increase in tourism infra-structure along the shoreline, which is ex-posed to storm surges and high waves.Rarotonga also serves as the transportation,administrative, and commercial hub of theCook Islands, with the result that impactson these sectors there will have repercussionsthroughout the country. Despite its muchsmaller population, the Northern Group isprobably most at risk of loss of life: as Cy-clone Martin in Manihiki, the ca. 1590cyclone in Pukapuka, the February 1942 cy-clone in Suwarrow, and others have demon-strated, a storm surge can raise sea levelsufficiently to cause waves to wash completelyover the low-lying atolls. Because evacuationof these remote atolls during the approachof a cyclone is not an option, reducing the in-habitants’ vulnerability hinges on properlysituated and constructed cyclone shelters andtimely warnings. Palmerston in the SouthernGroup is probably the atoll most likely to bestruck (Figure 2 and Table 2); fortunately itspopulation numbers less than 100. The outerislands of both groups are also substantiallymore vulnerable in terms of human liveli-hoods, because they are much slower to re-cover from a disaster in comparison withRarotonga. This was illustrated by the slow,and to some extent still ongoing, recovery in

Pukapuka and Nassau after Cyclone Percy in2005.

acknowledgments

The assistance of numerous agencies andindividuals in the creation of the historicaldatabase (de Scally et al. 2006), and the ongo-ing assistance and support of Arona Ngari,Director of the Cook Islands MeteorologicalService, are gratefully acknowledged. Jim Sal-inger from New Zealand’s National Instituteof Water and Atmospheric Research suppliedvaluable insights and data for defining ElNino–Southern Oscillation phases. The com-ments of two anonymous reviewers on anearly version of this paper were very helpful.The work in this paper is dedicated to thememory of Patricia Numa of Upper Tupapa,Rarotonga, who provided the main motiva-tion for this project.

Literature Cited

Asian Development Bank. 1991. Disaster mit-igation in Asia and the Pacific. Asian De-velopment Bank, Manila.

Basher, R., B. Collen, B. Fitzharris, J. Hay, B.Mullan, and J. Salinger. 1992. Preliminarystudies of South Pacific climate change:Present climate change and its impacts,data resources, and scenario possibilities.New Zealand Meteorological Service andNew Zealand Climate Change Program,Wellington.

Basher, R., and X. Zheng. 1995. Tropical cy-clones in the Southwest Pacific: Spatialpatterns and relationships to Southern Os-cillation and sea surface temperature. J.Clim. 8:1249–1260.

Bath, M., and J. Deguara. 2007. Australiasevere weather: Tropical cyclones. Aus-tralia Severe Weather Association, St.Ives (New South Wales), http://www.australiasevereweather.com/cyclones/index.html.

Beaglehole, E., and P. Beaglehole. 1938. Eth-nology of Pukapuka. Bernice P. BishopMus. Bull. 150.


de Scally, F. A., G. V. Wood, L. K. Maguire,M.-A. Fournier-Beck, and D. Silcocks.2006. A history of tropical cyclones andtheir impacts in the Cook Islands. CookIslands Meteorological Service, Nikao(Rarotonga).

Diamond, H. J. 2006. Review of recent trop-ical cyclone climatological research. IslandClim. Update 72, http://www.niwa.cri.nz/ncc/icu/2006-09/article.

Elsner, J. B., and K.-B. Liu. 2003. Examin-ing the ENSO–typhoon hypothesis. Clim.Res. 25:43–54.

Emanuel, K. 2005. Increasing destructivenessof tropical cyclones over the past 30 years.Nature (Lond.) 436:686–688.

Evans, J. L., and R. J. Allan. 1992. El Nino/Southern Oscillation modification to thestructure of the monsoon and tropical cy-clone activity in the Australian region. Int.J. Climatol. 12:611–623.

Fiji Meteorological Service. 1999. Tropicalcyclone Martin, 31st October–5th Novem-ber 1997. Tropical Cyclone Report 97/2.

Gill, W. W. 1894 (reprinted 1995). Fromdarkness to light in Polynesia. Institute ofPacific Studies, University of the SouthPacific, Suva.

Graham, N. E., T. P. Barnett, R. M. Chervin,M. E. Schlesinger, and U. Schlese. 1989.Comparisons of GCM and observed sur-face wind fields over the tropical Indianand Pacific Oceans. J. Atmos. Sci. 46:760–788.

Granger, K. 1999. An information infrastruc-ture for disaster management in Pacificisland countries. Australian GeologicalSurvey Organization, Canberra, Record1999/35.

Hastings, P. A. 1990. Southern Oscillationinfluences on tropical cyclone activity inthe Australian/South-West Pacific region.Int. J. Climatol. 10:291–298.

Henderson-Sellers, A., H. Zhang, G. Berz, K.Emanuel, W. Gray, C. Landsea, G. Hol-land, J. Lighthill, S.-L. Shieh, P. Webster,and K. McGuffie. 1998. Tropical cyclonesand global climate change: A post–IPCCassessment. Bull. Am. Meteorol. Soc.79:19–38.

Herold, C., F. Mouton, O. Norbeck, and P.

Peduzzi. 2006. GNV-199-Preview globalcyclone tracks datasets. Division of EarlyWarning and Assessment/Global and Re-gional Integrated Data, United NationsEnvironment Program, Geneva, http://www.grid.unep.ch/activities/earlywarning/preview/data/trop_cycl/trop_cycl.php.

Holland, G. J. 1983. Tropical cyclones in theAustralian/Southwest Pacific region. At-mospheric Science Department, ColoradoState University, Atmospheric Science Pa-per No. 363.

Hutchings, J. W. 1953. Tropical cyclonesin the South west Pacific. N. Z. Geogr.9:37–57.

International Research Institute for Climateand Society. 2006. Defining ENSO. TheEarth Institute, Columbia University,http://iri.columbia.edu/climate/ENSO/background/pastevent.html.

Joint Typhoon Warning Center. 2005. Trop-ical cyclone best track data site. U.S. NavalPacific Meteorology and OceanographyCenter, Pearl Harbor, https://metoc.npmoc.navy.mil/jtwc/best_tracks/index.html.

Kerr, I. S. 1976. Tropical storms and hurri-canes in the Southwest Pacific, November1939 to April 1969. New Zealand Meteo-rological Service, Wellington, Miscellane-ous Publication 148.

Maunder, J. 1995. An historical overview re-garding the intensity, tracks and frequencyof tropical cyclones in the South Pacificduring the last 100 years, and an analysisof any change in these factors. World Me-teorological Organization, Geneva, Tech-nical Document WMO/TD-No. 692/Report No. TCP-37.

McGuire, B., I. Mason, and C. Kilburn. 2002.Natural hazards and environmentalchange. Arnold, London.

National Oceanic and Atmospheric Adminis-tration. 2007. Saffir-Simpson hurricanescale, http://www.aoml.noaa.gov/general/lib/laescae.html.

Ramage, C. S., and A. M. Hori. 1981. Mete-orological aspects of El Nino. Mon.Weather Rev. 109:1872–1835.

Revell, C. G. 1981. Tropical cyclones in theSouthwest Pacific, November 1969 to


April 1979. New Zealand MeteorologicalService, Wellington, Miscellaneous Publi-cation 170.

Revell, C. G., and S. W. Goulter. 1986.South Pacific tropical cyclones and theSouthern Oscillation. Mon. Weather Rev.114:1138–1145.

Salinger, J., J. Renwick, and S. Burgess. 2004.Tropical cyclone guidance. Island Clim.Update 50, http://www.niwa.cri.nz/ncc/icu/2004-11/article.

Shaw, S. L. 1981. A history of tropical cy-clones in the central North Pacific andthe Hawaiian Islands, 1832–1979. Na-tional Oceanic and Atmospheric Adminis-tration, Silver Spring, Maryland, ReportS/T 81-168.

Smith, K. 2004. Environmental hazards: As-sessing risk and reducing disaster, 4th ed.Routledge, London.

Spennemann, D. H. R. 2004. Typhoons inMicronesia: The history of tropical cy-clones and their effects until 1914. Di-vision of Historic Preservation, Saipan,Commonwealth of the Northern MarianaIslands.

Sturman, A. P., and H. A. McGowan. 1999.Climate. Pages 3–18 in M. Rapaport, ed.The Pacific islands: Environment and soci-ety. Bess Press, Honolulu.

Sunderland, J. P., and A. Buzacott, eds. 1866.Mission life in the islands of the Pacific:Being a narrative of the life and laboursof the Rev. A. Buzacott, missionary ofRarotonga, for some time co-worker withthe Rev. John Williams, martyr of Erro-manga. John Snow, London.

Syme-Buchanan, F. 1997. Islands devastatedin the wake of Cyclone Martin. Pac. Is-lands Mon. December:11–12.

Terry, J. P., R. Raj, and R. A. Kostaschuk.2001. Links between the Southern Oscilla-tion Index and hydrological hazards on atropical Pacific island. Pac. Sci. 55:275–283.

Thompson, C. S. 1986a. The climate andweather of the Northern Cook Islands.New Zealand Meteorological Service,

Wellington, Miscellaneous Publication188 (3).

Thompson, C. S. 1986b. The climate andweather of the Southern Cook Islands.New Zealand Meteorological Service, Well-ington, Miscellaneous Publication 188 (2).

Thompson, C. S., S. Ready, and X. Zheng.1992. Tropical cyclones in the SouthwestPacific: November 1979 to May 1989. Na-tional Institute of Water and AtmosphericResearch/New Zealand MeteorologicalService, Wellington.

Tobin, G. A., and B. E. Montz. 1997. Naturalhazards: Explanation and integration.Guilford Press, New York.

Trenberth, K. E. 1997. The definition of ElNino. Bull. Am. Meteorol. Soc. 78:2771–2777.

U.S. Navy. 2004. Global tropical cyclone at-las v. 1.0. Fleet Numerical Meteorologyand Oceanography Detachment, Asheville,North Carolina, http://navy.ncdc.noaa.gov/products/gtcca/gtccamain.html.

Visher, S. S. 1925. Tropical cyclones of thePacific. Bernice P. Bishop Mus. Bull. 20.

Visher, S. S., and D. Hodge. 1925. Australianhurricanes and related storms, with an ap-pendix on hurricanes in the South Pacific.Commonwealth Bureau of Meteorology,Melbourne, Bulletin No. 16.

Webster, P. J., G. J. Holland, J. A. Curry, andH.-R. Chang. 2005. Changes in tropicalcyclone number, duration, and intensity ina warming environment. Science (Wash-ington, D.C.) 309:1844–1846.

Williams, J. 1841. A narrative of missionaryenterprises in the South Sea islands; withremarks upon the natural history of theislands, origin, languages, traditions, andusages of the inhabitants. John Snow,London.

Zillman, J. W. 1999. Opportunities and chal-lenges for meteorological and hydrologicalwarning systems. Pages 9–14 in Proceed-ings of the Australian Disaster Confer-ence, 1–3 November 1999, Canberra.Emergency Management Australia, Dick-son, Australian Capital Territory.


historical tropical cyclone activity and impacts in the ... · cyclones have occurred during el...

Documents