Aug 1, 1999 - Oscillation (ENSO) cycle (e.g. Evans and Allan, 1992;. Lander, 1994 ... events (Gray, 1984), but this does not appear to be the case in the North ...
GEOPHYSICAL RESEARCH LETTERS, VOL. 26, NO. 15, PAGES 2251-2254, AUGUST 1, 1999
The relationshipbetweenthe SouthernOscillationIndex andtropicalcyclonetracksin theeasternNorthPacific Rossman P. Irwin III and Robert E. Davis
Department ofEnvironmental Sciences, University ofVirginia, Charlottesville, Virginia
oftheverylimitedrecord of tropical cyclones that Abstract. Pointsof origin and downgradation of tropical resampling Hawaiisupported theGCM simulations (Chuand cyclones intheeastern NorthPacific(ENP)eastof 160øWare hadreached relatedto the SouthernOscillationIndex (SOl) during the Wang, 1997). Althoughthereappearsto be no correlation betweenSOl hurricaneseason. All ENP tropicalstormsand hurricanes
from 1966through1997weregroupedintothreeSOl-based categories: thoseoccurringduringstrongE1 Nifio events (meanSOl< -0.6),strong La Nifiaevents (meanSOl> 0.6), or nearzeroperiods (-0.6< meanSOl< 0.6). DuringE1Nifio storm seasons, ENP tropical cyclones originated approximately 5.7ø(617km)westanddowngraded 7.5ø (780 km) westof the long-term meanlongitudes for the positive SOl group. Near zero groupstormsalso followedmore northerly tracks than the negative SOl group storms. However,no significant differences in stormtrackareevident between thepositiveSOl andnearzerogroups, andthetrack lengthis notsignificantly differentfor anystormgroup.
and stormfrequencyor development, there is a growing consensusthat western Pacific typhoonsoriginate farther eastwardduringan E1 Nifio (negativeSOl) year. Lander (1994) determined that tropicalcyclonesduringEl Nifio eventsoriginated5.0ø eastof the meanoriginpointin the westernNorth Pacific,while positiveSOl stormsoccurred
1. Introduction
centralPacific near Hawaii. The top three years for storm
In several of the world's oceans,the frequency,origin
points,and movementof tropicalcycloneshave been observedto vary with the stageof the E1 Nifio/Southern Oscillation(ENSO) cycle (e.g. Evans and Allan, 1992; Lander,1994; Basherand Zheng, 1995). In the Atlantic basin,the frequency of hurricanes is suppressed by El Nifio events (Gray,1984),butthisdoesnotappear to bethecasein theNorthPacific(WhitneyandHobgood,1997). In theNorth Pacificjustwestof theInternational DateLine,Chan(1985) determined thata strongnegativeSouthern OscillationIndex (SOl) value was correlated with fewer stormsreaching typhoon intensity.However, thetotalnumber of storms in a givenseason wasnotaffected by anE1Nifioevent.As the
3.5ø to the west. Likewise, in the western South Pacific,
tropicalcyclones tendto formnearthe Australian coastline duringLa Nifia (positiveSOl) years,but the genesis region extendsmuchfarthereastduringE1Nifio years(Basherand Zheng, 1995). Again, there was no strongcorrelation betweenthe SOl and stormfrequencyin the westernSouth Pacific(Hastings,1990;EvansandAllan, 1992). Schroeder and Yu (1995) examinedstormgenesisin the
frequency nearHawaiiincluded the El Nifio years1982and 1994, when local storm genesis increasedalong with incursionsfrom the east. However,they correctlynotedthat
thereis noperfectcorrelation between El Nifio eventsandthe coincident storm tracks.
In thispaper,we analyzethegenesis regions andtracksof tropicalcyclones in theeastern NorthPacific(ENP)(eastof 160øW)which heretoforehave not been quantified. Most ENP tropicalcyclonesare generated betweenthe Mexican westerncoastand ClippertonIsland(at 110ø W), but storms have developedwest of 140øWin the ENP. The eastern South Pacific does not exhibit regular tropical cyclone
development andthereforeit is not considered in thisstudy. Previous work on western Pacific storms has noted a shift in
SOl reaches its minimum and the warm-phaseWalker the genesis regiontowardthe centralPacificduringEl Nifio circulationis established, sinkingair in thatregionsuppresses events.If theresponse of ENP tropicalcyclones to theENSO
cyclonedevelopment, according to Chan(1985)andothers. cycleis similarto thosein the westernPacificbasin,we This conclusion was partiallysupported by Lander(1994) hypothesize thattheywouldoriginate andtracksignificantly who, with a larger data set, also found no statistically westwardof the meanpointof origin.
significant correlation in thenorthwest Pacificbetween the SOl and the annual storm totals. Lander (1994), however, found no correlationbetween SOl and the frequency of
2. Data
intense(Category 3 or greaterstormson the Saffir-Simpson scale)typhoons.Basedon generalcirculation model(GCM) Data for this analysiswere providedin the EastPacific simulations, Wu andLau (1992) noteda reductionin storm Tracks File of January 1998 by the National Hurricane frequency in theirPacificnear-coast regionsduringan E1 Center,Miami, Florida (National HurricaneCenter, 1998). Nifio event. However,asthis reductionwas associated with a The data include6-hourly center locations,in tenthsof shiftof stormgenesis intothecentralPacific,thetotalnumber degrees,of the 520 ENP (eastof 160øW)hurricanes and of stormswas not necessarilyaffectedby the event. A tropical stormsoccurringbetweenJanuary,1966, and December,1997. Datafrom 1949through1965areexcluded, asthefile maybe incomplete priorto theadventof consistent satellite monitoring (Mayfield and Rappaport, 1998).
Copyright .1999bytheAmerican Geophysical Union.
Additionally,those stormsoriginatingwest of Hawaii (160øW)wereexcluded dueto possible influences thereof
Papernumber1999GL900533. 0094-8276/99/1999GL900533505.00
2251
2252
IRWIN AND DAVIS: NORTHEAST
PACIFIC TROPICAL CYCLONE TRACKS
western Pacific circulation patterns. The origin and downgradation pointswerethe first and lastreportedpositions DOWNGRADATIONIsles a. of eachstormas it acquiredand losttropicaldepressionstatus .a..0...... .+.....i....r.•.,...•...•....• ....... :: :: (M.•lco); (25 knot maximum sustainedwind velocity). Tropical :.. : . depressions are includedas we are interestedin the complete pathtakenby the storms,but we have only considered those ................. •............................................ •...................... :i• ..................... -;a.•,-•.•iasa depressions thatreachedat leasttropicalstormstatus. SOl data were provided by the Climate Impacts and Getapago•. Islands tin•Ismnds:i :: : "(Kiribai)l•0 140 130 120 ApplicationsGroupof the Queensland Departmentof Natural Resources,Indooroopilly,Australia(Brook, 1998). The SOl Figurel a. Themeanpointsof originanddowngradation for is calculatedas ((T- D)/S), where T and D are, respectively, all easternNorth Pacific(eastof 160øW)tropicalcyclonesin valuesof the sealevel pressureanomaly(departurefrom the the SOl positive(+), negative(-), andnearzero(0) groups. long-termmean)at Tahiti andDarwin, Australia,and S is the standarddeviationof the value (T - D) for all monthsfor the years 1951to 1980 inclusive. 3. Methods
Previouswork on ENSO/tropicalcyclonerelationshipshas producedseveralclassifications of stormseasons, basedeither on seasurfacetemperature(SST) or SOl in differenttimes of the year. The classificationschemedependson the location of the studyareaandthe thresholds established by the various researchers (e.g. Chan, 1985; Basherand Zheng, 1995; Chu and Wang, 1997). Our SOI-basedcategoriesare roughly Figure lb. EasternNorth Pacific Storm Tracks in 1997, when the meanMay-October SOl was-1.70. consistent with the Northern Hemisphere sea-surface temperature divisions used by several researchers(see Trenberth,1997), as El Nifio andthe SouthernOscillationare causallylinked. To dividethe tropicalcyclonesinto groups, the storm seasons from 1966-1997
were classified based on
the mean monthly SOI value duringthe May-October period when 97.5% of the stormsoccurred(Table 1). As only 13 out of 520 storms(2.5%) occurredin Novemberor later,the mean seasonalSOl doesnot incorporate valuesafterOctober. This procedure allowsusto distinguish betweenthe effectsof large positiveor negativeSOl anomalies,while those of smaller magnitudeor shortdurationare groupedin the "near zero" category. For a storm seasonto be designatedas "positive
Table 1. Stormseasons (heredefinedasMay-Octoberof eachyear) includedin eachcategory,with the meanSOl duringthatseason.N, 0, andP designatenegative,near zero,andpositiveSOl seasons, respectively.The categories aredelimitedby SOl valuesof + 0.6. Year
Mean
Group
Year
SO1
1966 1967 1968 1969 1970 1971 1972 1973 1974 1975 1976 1977 1978 1979 1980 1981
-0.12 0.23 0.45 -0.65 0.56 1.02 -1.43 0.91 0.85 1.65 -0.49 -1.20 0.39 0.28 -0.20 0.58
Mean
Group
SO1
0 0 0 N
0 P
N P P P 0 N
0 0 0 0
1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997
-1.75 0.18 -0.12 -0.07 0.00 -1.40 1.08 0.64 0.10 -1.02 -0.54 -1.10 -1.39 -0.05 0.61 -1.70
N 0 0 0 0 N
P P 0 N
0 N N
0 P
N
•
Elida
•
F½ -",•.•_.•.
-10
ina t
'•"C
.
ß
I $0
140
130
120
I I 0
I O0
90
Figure I c. EasternNorthPacificStormTracksin 1996,when the meanMay-October SOl was +0.61. SOl" (La Nifia) or "negative SOl" (El Nifio), the mean sixmonth SOI value must be greater than +0.6 in magnitude. The near zero seasonscontained individual monthly SOI values with extremesranging from -2.7 to +2.2, but these were short-livedfluctuations.The SOI data segregateditself neatlyaccordingto this procedure.For example,the weakly negativeSOI seasons1976 and 1992 were includedin the near zero category,while the strongeranomaliesof 1987 and 1997 were included in the negative SOl category(Table 1). There were nine strongEl Nifio years,16 nearzero years,and sevenstrongLa Nifia yearsduringthe 32-yearperiod. Of the 520 total storms, 138 were classified in the negative SOI category,276 were in the nearzerocategory,and 106 were in the positiveSOI category. The mean latitudes and longitudesof tropical cyclone origin and downgradationwere calculatedfor the stormsin each SOI group. The mean northerly (increasein latitude over the lifetime of the storm) and westerly (increase in longitude)movementof eachstormwas then computedfrom the six-hour positions. Two-sample unpairedt-tests were used to compare the three group means for significant differencesin positionof origin, positionof downgradation, andtrack length.
IRWIN AND DAVIS: NORTHEAST
PACIFIC TROPICAL
Finally, where significantdifferencesexisted in latitude andlongitudeof origin,we countedthe stormsin eachgroup that occurredat the edgesof the primaryregionof tropical cyclonegenesis,illustratedby the graybox in Figure l a. We tallied thosestormsthat originatedeast of 100øW, west of 120øW, north of 16øN, or southof 10øN. We also countedthe
number of tropical cyclonesthat tracked west of 150øW, nearer to the Hawaiian Islands,regardlessof the point of origin. To determine whether the SOl groups showed
significant differences inthepreference ofthese regions, a X2test for consistencyin a K x 2 table was used(Kanji, 1993). The samples(K) were the three SOl-basedgroups,and the stormsin eachgroupwere dividedinto the classes"in area"
CYCLONE
TRACKS
2253
Table 2. Mean locationof originanddowngradation and the meannortherlyandwesterlymovementsin degrees subtendedfor easternNorth Pacifictropicalcyclonesin the positive(La Nifia), nearzero,andnegative(El Nifio) SOI categories.The originanddowngradation pointsare the first andlastrecordedpositionsof the tropical depression, wherethe maximumsustained windswere 25 knots. The null hypothesisof similaritybetween categoriesis rejectedwith P values