Weather and (1993) 13: 9-21 9

CLIMATOLOGICAL SEESAWS IN THE SOUTHWEST PACIFIC

John Hayl Jim Salinger2 Blair Fitzharris3 and Reid Basher2

1 Environmental Science, University of Auckland, Auckland, New Zealand

Postal Address: Environmental Science, University of Auckland, Private Bag 92019, Auckland, New Zealand Telephone: 649-373-7599 (Extn 8437) Facsimile: 649-373-7470

2 National Institute of Water and Atmospheric Research, Wellington, New Zealand 3 Geography Department, University of Otago, Dunedin, New Zealand

ABSTRACT

The analyses reported in this paper are based on an unprecedented 80 year record of reconstructed grid-point sea-level pressure data which are used to derive composite pressure anomaly fields for the Southwest Pacific for extreme phases of the Southern Oscillation. These show centres of action to the east and west of New Zealand. However, it is the intervening pivotal region that includes New Zealand which experiences a strong response — anomalous south to southwest flow when the Southern Oscillation Index (SOI) is negative and north to northeasterly flow when the Index is positive. These contrasting responses are reflected in the spatial distribution of rainfall anomalies for New Zealand. A more complex response occurs in another pivotal region in the vicinity of the Southern Cook Islands. The South Pacific Convergence Zone is displaced northward and southward in association with extreme phases of the Southern Oscillation resulting in large rainfall anomalies in the areas which are under its influence only in such conditions. However, the effective finite width of the South Pacific Convergence Zone and its limited north/south displacement mean that there are some locations in the pivotal region which are always influenced by the convergence zone. Consequently these do not exhibit the same high degree of sensitivity to variations in the SOL Such contrasting responses are demonstrated using a recently compiled rainfall archive for the Southwest Pacific.

INTRODUCTION pressure anomalies for two widely separated locations (Sydney and Buenos Aires) were first As Rasmusson and Carpenter (1982) note, described by Hildebrandsson (1897). Lockyer well developed out-of-phase relationships of and Lockyer (1902a, 1902b, 1904) subsequently 10 Climatological Seesaws confirmed these findings, showed that such The present study uses an unprecedented 80 oscillations were almost world-wide and de- year record of reconstructed grid-point sea- scribed them as 'barometric seesaws'. It was level pressure data (Jones, 1991) for the Aus- some 20 years later that Sir Gilbert Walker tralasian region (Fig. 1) to characterize the named one of these seesaws the Southern patterns associated Oscillation (SO) and described the key fea- with extreme Southern Oscillation phases. In tures of its surface pressure, temperature and a spectral analysis of the differences in sea- precipitation variations over the Indian and level pressures for stations known to reflect southern Pacific Oceans (Walker and Bliss, the main characteristics of the SO, Trenberth 1932; 1937). (1976) showed that most of the variance oc- Associated with the SO are 'centres of ac- curred in periodicities between 2 and 10 years. tion' — defined as semi-permanent highs and Thus studies of SO and the related El Nino lows that appear on charts of mean sea-level phenomenon require an extended period of pressure — which principally lie over Indone- record in order to reveal the key features. sia and in the South Pacific anticyclone and The ability to analyse regional pressure are connected by an equatorial east/west cir- patterns extending back to 1911 provides an culation called the opportunity to include a larger number of (Trenberth, 1976). Despite their name, studies significant Southern Oscillation events than such as those by Gordon (1986) and was possible for earlier investigations of re- Rasmusson and Carpenter (1982) have shown gional circulation patterns in the Southwest that some of the more significant climatologi- Pacific (e.g. Rasmusson and Carpenter, 1982; cal variations occur not at these extremities Trenberth, 1976; Kiladis and van Loon, 1988; of the seesaw, but in the intervening pivotal Gordon, 1986; Allan and Pariwono, 1990). Al- region. While the centres of action can also be lan (1988) notes that coherent evidence of identified by their well-developed Southern Oscillation related responses in the with other centres (Trenberth, Australian region can be found only when 1976) the pivotal regions where significant extreme phases of the Southern Oscillation climatological variations occur are Index (SOT) are examined. The present analy- characterized by statistically non-significant sis focuses on such events by examining both correlations (Gordon, 1986). Similar paradoxi- El Nino and La Nina events. cal statistical patterns of sensitivity to pres- The objective of this study is to describe the sure anomalies have also been described by dramatic nature of the circulation and related Fitzharris and Bakkehoi (1986). changes occurring in the pivotal zone of the Physically the relationships are far from seesaw in response to changes in the centres paradoxical — the pressure oscillations, or of action, and hence competing with them for barometric seesaws, are associated with sig- climatological significance. Recently pre- nificant reversals in the atmospheric circula- pared archival data for the South Pacific tion anomalies. For the Pacific region, (Table 1) are used to identify some of the Trenberth and Shea (1987) show pivotal re- climatological consequences of the gions occurring in the vicinity of Rarotonga documented changes in atmospheric for the major centres of action (Indonesia and circulation. tropical Southeast Pacific Ocean) and over New Zealand for one of the several secondary DATA centres. These two areas should therefore ex- perience the most marked reversals in atmo- Southern Oscillation Index (SOI) spheric circulation anomalies as the standing oscillation itself changes sign. The apparent The monthly SOI data (normalised pressure paradox is that, in some situations, locations differences between Tahiti and Darwin) used near the pivot of the seesaw, rather than at here cover the period 1911 to 1989. Mullan the so-called centres of action, have the great- (1989) has described their derivation. Parker's est variability in climatological conditions. (1983) monthly values for the period January Substantial changes in the direction of the 1935 to March 1983 were updated using stand- anomalous flow will manifest as major modi- ard six-hourly pressure differences between fications to precipitation and other atmo- Tahiti and Darwin. These values were normal- spheric patterns. ised against a base period of 1941-80 to form Climatological Seesaws 11

TABLE 1: LIST OF RAINFALL STATIONS FOR THE Nuku'alofa, Tonga 1944 TROPICAL SOUTHWEST PACIFIC USED IN THE Alofi, Niue 1905 PRESENT STUDY Puka Puka, Cook Islands 1929 Rakahanga, Cook Islands 1941 Station Name Start Year Penrhyn, Cook Islands 1937 Palmerston, Cook Islands 1940 Buteriteri, Kiribati 1945 Aitutaki, Cook Islands 1914 Tarawa, Kiribati 1947 Mauke, Cook Islands 1929 Banaba, Kiribati 1953 Rarotonga, Cook Islands 1927 Beru, Kiribati 1945 Mangaia, Cook Islands 1914 Nanumea, Tuvalu 1946 Nui, Tuvalu 1946 the Troup Index (McBride and Nicholls, 1983). Funafuti, Tuvalu 1941 The SOI time series was extended back to Niulakita, Tuvalu 1947 Nadi, Fiji 1942 1851 (only data back to 1911 were used in the Atafu, Tokelau 1949 present study) using seasonal Index values Nukunono, Tokelau 1944 published by Wright (1975, 1989). Monthly val- Apia, Western Samoa 1890 ues of the Index were calculated from these Keppel, Tonga 1942 Vava'u, Tonga 1947 seasonal seasonal data using regression qua-

120E 140•E 160•E Ito' ISO•VI ILVW I Hawalan is. 20'N r- NORTH EAST TRADES •sx•c3

• Marshall Is

1 irr r1-0 "Tr i rL.-673 tri Fro-g ireg .6-NI Nauru t TERL V—M-073-/-0N Kiribati Lino Is. NewPpuaulneaN • DIVERGENT • , Tuvalu Tokelau

E717i G • sou A L V r G T Scl amr.c'e 'EASTERLIES 0 Vanuatu Fiji ..crs.. •••. ONL 20•S

AUSTRALIA

H H AD'S TRAvELL TROUGHS

DISTURBED

0 I OW km 60'S 1 1 Fig. L Typical circulation features in the Southwest Pacific for summer and (next page) winter. Also shown are the areas analysed using the gridded pressure data and the tropical rainfall data. After Steiner (1980). 12 Climatological Seesaws

120'E 140"E 180* 160'W 140W • 11 Hawalan Is. • 20•14 7— 1 • :s Mariana Is.

• i▪ N• tER F-R-6 Tqz.74-17,,,,4,...... marzhall Is. ... Ct371-1/ERGENCE • • Kiribati Ins Is.

Solomon is. 01%,juvalti Tokela4 I DIVERGENT '."•!e.\ • • Ade- .IEAST•ERLIES • A%. • SOUTHIEAST.-ADES ••••• 4, cSarta... //P./ • %. Cook Is • Vanuatu \.t %.• "..• • ivle %. • , -, • 20'S Nip .;.34t • .E/i, Q• %. Tahiti NovN. % Caledonia s' Tonga S. S. S. ' o4, S. S.

S. H THAVEIAANG \ ANTIC'YCLO AND TROUGHS \ 40'S H —NV NEW ZEALAND

DISTURBED

O MOO km

60'S 1 tions relating the monthly Tahiti-Darwin pres- Southwest Pacific have long pressure time sure difference to Wright's Index values for series, with some extending back to the mid- the surrounding three seasons. The period dle of the 19th Century. Gridded pressure data 1935-1974 was used to generate the coeffi- were derived following the methods described cients of the regression equations. by Jones (1987) and Jones and Wigley (1988). Ropelewski and Jones (1987) have discussed Their technique uses principal components re- the strengths and weaknesses of this and other gression to develop a series of equations relat- forms of the SO!. ing grid point mean sea level pressure values to contemporaneous pressures observed at in- Gridded Sea Level Pressure Data dividual stations. Pressures were reconstruct- ed for 1911 on, when air pressure was Jones (1991) describes the homogenization measured at 17 stations. The grid provides 65 procedures used to assess the observed sea points in an area between 20 and 60 S latitude level pressure data that form the basis of the and 100 E and 140 W longitude. reconstructed gridded pressure data used in Fitzharris et al. (1992) have used the same the present study. A total of 35 stations in the gridded pressure data set in a study of the Climatological Seesaws 13 long-term relations between atmospheric cir- while the indeterminate composites are based culation patterns in the Southwest Pacific and on 40 years or seasons. In the case of the the behaviour of New Zealand glaciers. They rainfall analyses, which cover the shorter show the grid as well as the average summer period from 1950 to 1989, the sizes of the and winter pressure patterns for the period composites are 10 and 20, respectively. 1911 through 1989. Rasmusson and Carpenter (1982) note that compositing represented an attractive ap- Rainfall Data proach to the study of a specific type of weather event since it aids in the identifica- Over the past three years the N.Z. Meteoro- tion of features which are common to a num- logical Service has been developing a compre- ber of individual cases. hensive archive of climate data for the South- An objective definition of La Nina and El west Pacific, one that would achieve the Nino events somewhat similar to that de- highest practicable quality in terms of accura- scribed above was adopted by Mullan (1989). cy and completeness. The rainfall stations However, he used parametric statistics, three (see Table 1) were selected to give optimum month running means rather than an annual coverage and representation of the Southwest average, and a threshold that identified the Pacific region and to provide long and com- end of the event. plete data time series as discussed by Basher The results presented in the following sec- et al., (1992). They also describe the quality tion are typically expressed in terms of pres- control and other procedures designed to pro- sure or rainfall anomalies. For the former, a vide homogenous time series of precipitation base period of 1911 to 1989 (i.e. the entire data. period of record) was used while the rainfall Table 1 shows that for some locations - anomalies are departures from the average fall observations began last century, but wide- for the period 1950 to 1989. As noted above, spread coverage of the Southwest Pacific did this shorter period was necessitated by there not occur until the late 1940s. Consequently being too few stations reporting rainfall in the the analysis of rainfall data was limited to the study area prior to 1950. period 1950 to 1989. RESULTS AND DISCUSSION METHODS The New Zealand Circulation Seesaw All data were analysed over a climatologi- cal year, running from November through to Fig. 2 shows strongly contrasting pressure October of the following year, made up of a anomaly patterns for the El Nino and La Nina tropical wet season (November-April) and a composites during the tropical dry season and, tropical dry season (May-October). The upper as might be expected, a total lack of coherent and lower quartiles of the annual values of the structure for the indeterminate composite. In SOT for the period 1911 to 1989 were used to terms of pressure, the centres of action are stratify pressure and rainfall data for those well marked, with one to the east and the same years, and the associated wet and dry other to the west of New Zealand. This places seasons, into the following composites: New Zealand in a key pivotal position, as has previously been shown by Gordon (1986) who Designator Criterion for Inclusion in mapped for a similar region the correlation Composite between gridded mean sea-level pressures and the SOL Thus, for May to October, at least, La Nina Actual SOT > Upper Quartile flows over New Zealand are sensitive to the Value of SOI SOT, with anomalous south to southwesterly El Nino Actual SOI < Lower Quartile flows when the SOI is negative and north to Value of SOT northeasterly flows when the index is positive. While analyses of data for the tropical wet Indeterminate Other years or seasons season and for a full year show changes in the specific locations of the centres of action and Each of the La Nina and El Nino compos- hence in the orientation of the anomaly iso- ites for pressure contain 19 years or seasons lines over New Zealand (see Figs. 3 and 4 for 14 Climatological Seesaws examples), the essential features described for beginning of the year before the development the dry season are retained. of the warm (El Nino) or cold (La Nina) event Kiladis and van Loon (1988) have also pre- and showed only the zero line separating posi- sented composite pressure anomaly patterns tive and negative anomalies and the sign of for extremes of the SO. However, direct com- the anomaly when it was outside the middle parison is not possible as they mapped the 40th percentile of the normal t distribution. In anomalies as three month means from the the present study the anomalies are quantified 15 S Lat - c,-o

• New Caledonia „Rarotonga-

Kermadec Is

_Chatham Is ChristchiircH Dunedin

Is Campbell-Is. Macquarie Is ,o•A• 0.2 _06 -0 4 -0 6 -1.0 60 S 8 100 E 120 E 140 E 160 E 180 160 W 140 W Long 15 S Lat ..e • New Caledonia ....Rarotonga

• Norfolk Is 30 S--• Kermadec Is •

• Auckland

Hobart •Chatham Is 45 S- • Christchurch . c, Dunedin

• 'Auckland Is • 'Campbell Is 'Macquarie.ls 0 60 S 100 E 120 E 140 E 1610 E 180 160 W 140 Long Fig. 2a and b: Sea level pressure anomalies (hPa) for May to October for years between 1911 and 1989 where the SOI was (upper) lower than its lower quartile value for the same period; and (lower) between the lower and upper quartiles. Climatological Seesaws 15

155 Lat

Rarotonga ledonia

A •C, 0) •C) Norfolk 30 rma( ec Is,

c,- •9 \.

45 S0 Christchurch/Auckland c"1,Chathan (s) Dun-gin/ Auckland Is

60 S 100 E 120 E 140 E 160 E 180 160 W 140 W Long

Fig. 2c: Sea level pressure anomalies (hPa) for May to October for years between 1911 and 1989 where the SO" was above the upper quartile value for the same period.

15 S Lat

Z•t 0•1- 8•0- Norfolk Is. 03 Q. 30 S / Kermadec-Is o -6‘ 9'0- 17.0- Auckland s Z-0- • 0 •

45 S

'Auckland Is • 1 • Campbell Is Macquarie Is••

SOS 100 E 120 E 140 E 160 E 180 160 W 140 W Long

Fig. 3: Sea level pressure anomalies (hPa) for November to April for the years between 1911 and 1989 where the SO' was above its upper quartile value for the same period. 16 Climatu. 'Seesaws

15 S— Lat 111117,11111111k Fiji 0

•New Caledonia \ Rarotonga-- \ ,c2 0 I \ 30 8— • Norfolk Is / Kermadec Is,' Cb 'b Auckland ,c)\

I. z Hobart ,Chatham Is 45 S- e-4010c/histchurch' Dunedin

z v • •Auckland Is •

• + •Macquarie Is • oy ,I E9 1------,N5-• 'Campbell Is 60 S 100 E 120 E 1408 1608 180 160 W 140 W Long

Fig. 4: Annual (November to October of the following year) sea level pressure anomalies (hPa) for years between 1911 and 1989 where the SOI was above its upper quartile value for the same period. and presented in greater spatial detail so that the locations of the centres of action and the orientation and strength of the pressure anom- aly gradients are more readily seen. The strong anomaly centres over Australia and French Polynesia and the anomalous cir- culation over and north of New Zealand are reversed between El Nino and La Nina events 049 and, as Kiladis and van Loon show, a similar reversal occurs between the year prior to the event and the year of the event itself. Not only do these reversals modulate the strength of the meridional wind over and to the north of New Zealand, but they are also associated with marked equatorward (El Nino and year prior to La Nina) and poleward (La Nina and year prior to El Nino) displacement of the South Pacific Convergence Zone (Kiladis and van Loon, 1988).

Consequences of the New Zealand Seesaw Given the strength of this seesaw action one would expect it to be revealed in the climato- logical record. Such is indeed the case. Corre- lations between rainfall and the SOT, as developed by Salinger (1981), show a dipolar relationship (Fig. 5). Areas in the north and Fig. 5: Correlation between annual rainfall and the South- northeast of the North Island of New Zealand ern Oscillation Index (from Salinger, 1981). Climatological Seesaws 17 and in the north of the South Island are posi- also noted that the preceding changes were tively correlated, reflecting the consequences most pronounced over the New Zealand region of increased northerly and northeasterly flow confirming that area as a key pivotal region and enhanced orographic uplift when the SOI in the Southwest Pacific. is positive. Rainfall data from stations in the Similarly, New Zealand temperature anom- south of the South Island correlate negatively alies are significantly correlated with the SOI, with the SOI, relating to rain shadow effects explaining about 16% of the variance in mean with a positive SOI and to enhanced south- annual temperatures (Salinger, 1981 and westerly flow when the SOI is negative. 1983). Therefore, the SOT is a contributing These interpretations are consistent with factor to New Zealand's overall climate vari- the findings of Khatep et al. (1974) who inves- ability and to the contrasting responses within tigated a tropospheric meridional transect the country (Gordon, 1985; 1986). from Campbell Island to Fiji. They found that during 1960-73 there were only small changes The Tropical Southwest Pacific Seesaw in both the magnitude of the water vapour flux and total column precipitable water be- The mean annual rainfall for the tropical tween wet and dry periods. However, they Southwest Pacific (Fig. 6) exhibits consider- showed substantial changes in the flux direc- able spatial variation. However, there is a tion, especially in summer. Wetter periods broad zonal maximum south of the equator were associated with enhanced lower tropo- associated with the South Pacific Conver- spheric northeast flow while for dry periods gence Zone (SPCZ). The irregular seasonal westerly transfer was more dominant. They north and south migration of this zone follows

N Lot

Butaritari, 3081 • • Tarawa • 1907

Banaba• ;. • Beru 1791 ' 1255

Nanumea 2881 • Nut • 3269 •s• • Funafuti Atatu, P 3498 2752 • Nukunonu 26021" loos 2707 • Rakahanga • Niulakita • Pukapuka 3456 2743

,n Apia - .2949 •

Keppel 2264 Nandi Palmerston. 1912 • Vava'u Niue. Altutakk 2266• 2102 1934 20 S • ' Mauke Rarotonga. 1827 .*Nukualofa 1970 • Mangala 1802 1971

S Lot 160' E Long 170°C igO'E 170 %At 160'W 150"W Long Fig. 6: Mean annual rainfall for selected stations in the tropical Southwest Pacific. Based on the period 1950 to 1989. 18 Climatological Seesaws

-50

Butaritari,/ /' :Tarawa

Banaba • 0.• Ben/

1 DRIER - — ---- — ------Nanumea • ----- _

• Nukunonu — — — , Penrhyn ' • • Niulakita • Pukapuka Rakahanga

n

+25 ,-, cf?, • ' Kennel Nandi Palmerston ava 1;•• Niue. Altutaki* • Mauke • Rarotonga '•.. • Nuku'alofa • Mangala

WETTER '-+25 30°S

Lot 160°E Lon 1VN Lot

+50 Butaritari

,(Tarawa WETTER Banaba • c''.•• • io ; Beal

Nanumea

+25 +251------• Funafuti Maki • ---- — ------0., • Nukunonu Rakaha• nga ,Penrhyn • Nkilakita • Pukapuka

so Apia

it • • Keppel

Nandi •C23 Vava'u PlIznerston , 1 Niue. Aitutaki o 2/Ps

• Nuku'alofa \ Rarotonga • • MMa naguaki ea DRIER s -25 3CPS

L'' taPE Long litrE 180°E 170 W 160.W 150°W Long Fig 7: Mean annual rainfall anomalies (%) for years between 1950 and 1989 where the SOI was (upper) above its upper quartile value for the same period; (lower) below the corresponding lower quartile value; and (next page) between the lower and upper quartiles. Climatological Seesaws 19

01 Butaritari, :Tarawa

r. ;. •Beru

Nanumea • .1•4 • • +4 Nui • • Funafuti Atatu• \ 4 Penrhyn

• Niu‘lat

+:4 +4 _Apia. -4, s, • Keppel Nandi' Palmerston. •)\ • Vava•tf4 Niur;! Altutakk t • • Mauke -41 ; Rarotonga. \ • Nuku'alofa s‘•Mangala / +'4 0 (/ 1+4

170.E 11110.0 170 W 160.W 150.W Long the annual cycle of the solar position (Fig. 1), the study area. Comparisons show that in El but is lagged by a few months due to the Nino years it is considerably wetter in the thermal inertia of the oceans. The SPCZ is north, and drier in the south. The opposite is stronger during the southern summer. Islands the case for La Nina years. This analysis lying beneath the passing convergence zones identifies a climatological consequence of the (whether they be the SPCZ or the Intertropical seesaw that is characterised by the SOT, but in Convergence Zone (ITCZ) to the north) experi- this case the core of the pivotal region that ence wet seasons followed by dry seasons. runs diagonally from near Western Samoa to Variations in the position and strength of the the Cook Islands is relatively insensitive to the zones from year to year mean that the natural SOL The core comes under the influence of the interannual variability in South Pacific rain- SPCZ irrespective of the state of the SOL fall is relatively high, especially for those However, in La Nina years the SPCZ is dis- locations at the periphery of the convergence placed southward, thereby giving positive pre- zones. Nicholls and Wong (1989) have shown cipitation anomalies south of the core. The that the Southern Oscillation has a strong reverse is true for El Nino years. Thus greater influence on the behaviour and location of the sensitivity to the SOT is exhibited outside the convergence zones. core of the pivotal region. This marked influence is shown in Fig. 7 Such changes are consistent with variations which presents composite rainfall anomalies in the pressure patterns shown in Figs. 2 to 4. for the La Nina, El Nino and indeterminate There is an increase in northerly flow through SOT categories between 1950 and 1989. Again the New Caledonia-Rarotonga area when the the average anomalies for the non-extreme SOT is strongly positive, associated with a years (Fig. 7) are weak and show little spatial southward displacement of the SPCZ. When coherence. However, for the ten most positive the SOT is negative an increase in southerly and ten most negative SOI years the anoma- flow is consistent with a northward movement lies are spatially coherent and well developed of the SPCZ. These anomalous north/south in the northern and southern extremities of shifts in the SPCZ contrast with the conclusion 20 Climatological Seesaws of Streten (1975) that the SPCZ is displaced northeasterly flows during La Nina years, and eastward (westward) during ENSO (anti- increased southwesterly flows in El Nino ENSO) years and are more in keeping with the years. In the case of the tropical Southwest conclusions of Kiladis and van Loon (1988). Pacific seesaw the finite width of the SPCZ They show that in the year of a warm (El results in a somewhat different response. Lo- Nino) event, weakened trade winds and higher cations in the core of the pivotal region re- than normal sea-surface temperatures occur main under the influence of the SPCZ along the equatorial Pacific together with regardless of the circulation anomalies and heavy that reach south to about 15 S hence exhibit little interannual variability in near the dateline, associated with the equato- precipitation. However, locations to the north rial shift of the ITCZ and the SPCZ. Thus and south of this pivotal zone experience high compared to all isohyets are interannual variability in precipitation consis- shifted equatorward while drier than normal tent with the anomalous positioning of the conditions prevail in the sub-tropical South SPCZ as the atmospheric circulation of the Pacific. In contrast, in the year of a cold (La wider Pacific changes with the ENSO Nina) event stronger than normal trades and a phenomenon. poleward shift in the ITCZ and SPCZ the The results demonstrate some of the clima- precipitation anomaly is also displaced south- tological consequences for the South Pacific of ward away from the equator. Kiladis and van well-developed teleconnections that occur Loon also show a pivotal zone (located around within the tropical atmosphere and between 10 S at the dateline) with little sensitivity in the tropics and mid latitudes. An understand- precipitation variability to cold and warm ing of linkages in the present atmospheric phases of the SOL The reversal in the relation- circulation, as provided by the present study ship between precipitation anomalies and SOI and those such as Fraedrich (1990), Hamilton as one goes north or south of the pivotal line is (1988) and Fraedrich and Muller (1992), is confirmed by the pattern of seasonal correla- desirable before considering the climatologi- tion described by Mullan, 1991. He shows that cal consequences of human induced changes negative correlations occur north of about 10 such as those resulting from greenhouse S while the correlations are positive south of warming and ozone depletion. that latitude. The patterns described here for the compos- ACKNOWLEDGEMENTS ited annual precipitation are repeated for The authors express their thanks to Dr both the wet and dry seasons. In all cases the Brett Mullan and Dr Neil Gordon of the New pivotal zone remains close to that shown in Zealand Meteorological Service (for providing Fig. 7 and the substantial seesaw in precipita- the monthly SOI time series for the period tion anomalies occurs to the north and south. 1851 through 1989), to Dr Phil Jones of the Climatic Research Unit, University of East Anglia (for making available the gridded sea- SUMMARY AND CONCLUSIONS level pressure data) and to Ms B. Colien of the New Zealand Meteorological Service for her This study has investigated circulation and assistance with accessing the rainfall data. rainfall interactions, focusing on the pivotal This work was supported by PRCS 90/83 from regions located between the primary atmo- the New Zealand Foundation for Research, spheric centres of action (Indonesia and Science and Technology. French Polynesia) and one of the secondary centres (southeast of New Zealand). It was REFERENCES noted that, paradoxically, the regions between the centres of action would experience greater Allan, R.J. 1988. El Nino Southern Oscillation influences climate variability due to marked reversals in in the Australasian region, Progress in Physical Geo- graphy, 12(3), 313-348. the circulation anomalies in such pivotal Allan, R.J. and Pariwono, J.I. 1990. Ocean-atmosphere regions. interactions in low-latitude Australasia, International This is indeed the case for the New Zealand Journal of Climatology, 10(2), 145-178. circulation seesaw, where the topography en- Basher, RE., Colien, B., Fitzharris, RR, Hay, IE., Mullan, B. and Salinger, J. 1992. Preliminary Studies hances the spatial anomalies in precipitation for South Pacific Climate Change. New Zealand Mete- associated with strengthened northerly and orological Service, Wellington, 7Opp. Climatological Seesaws 21

Fitzharris, B.B. and Bakkehoi, S. 1986. A synoptic clima- McBride, J.L. and Nicholls, N. 1983. Seasonal relation- tology of major avalanche winters in Norway, Journal ships between Australian rainfall and the Southern of Climatology, 6, 431-446. Oscillation, Monthly Weather Review, 111, 1998-2004. Fitzharris, B.B., Hay, J.E. and Jones, P.D. 1992. Behav- Mullan, A.B. 1989. Influence of Southern Oscillation on iour of New Zealand glaciers and atmospheric circula- New Zealand Weather. Paper presented at the West- tion changes over the past 130 years, The Holocene, ern Pacific International Meeting and Workshop on 2(2), 97-106. TOGA-COARE, Noumea, New Caledonia, May 24-30, Fraedrich, K., 1990. European Grosswetter during warm 1989, Ilpp. and cold extremes of the El Nino/Southern Oscilla- Mullan, A.B. 1991. Atmospheric circulation processes and tion, International Journal of Climatology, 10, 21-32. features in the South-West Pacific, South Pacific Envi- Fraedrich, K., and Muller K. 1992. Climate anomalies in ronments: Interactions with Weather and Climate Europe associated with ENSO extremes, International (J.E. Hay, ed.), Environmental Science Occasional Journal of Climatology, 12(1), 25-31. Publication No. 6, Environmental Science, Auckland Gordon, N.D. 1985. The Southern Oscillation: a New University, 15-23. Zealand perspective, Journal of the Royal Society of Nicholls, N. and Wong, K.K. 1989. Dependence of rainfall New Zealand, 15(2), 137-155. variability on mean rainfall, latitude and the Southern Gordon, N.D. 1986. The Southern Oscillation and New Oscillation, J. Climate. Zealand weather, Monthly Weather Review, 114, 371- Parker, D.E. 1983. Documentation of a Southern Oscilla- 387. tion index, Meteorological Magazine, 112, 184-188. Hamilton, K. 1988. A detailed examination of the extra- Rasmusson, E. and Carpenter, T. 1982. Variations in tropical response to tropical El Nino/Southern Oscil- tropical sea-surface temperature and surface wind lation events, Journal of Climatology, 8, 67-86. fields associated with the Southern Oscillation/El Hildebrandsson, H.H. 1897. Quelques reserches sur les Nino, Monthly Weather Review, 110, 354-384. entres d'action de l'atmosphere, K. Svenska Vetens.- Ropelewski, C.F. and P.D. Jones, 1987. An extension of Akad. Handl., 29,33pp. the Tahiti-Darwin Southern Oscillation index. Monthly Jones, P.D. 1987. The early twentieth century Arctic high Weather Review, 115(9), 2161-2165. — fact or fiction?, Climate Dynamics, 1, 63-75. Salinger, M.J. 1981. New Zealand Climate: The Instru- Jones, P.D. 1991. Southern hemisphere sea level pressure mental Record. Unpublished Ph.D. thesis, Victoria data: an analysis and reconstructions back to 1951 and University of Wellington. 1911, International Journal of Climatology, 11(6), 585- Salinger, M.J. 1983. The New Zealand temperature series, 608. Climate Monitor, 9, 112-118. Jones, P.D. and Wigley, T.M.L. 1988. Antarctic gridded Steiner, J.T. 1980. The Climate of the Southwest Pacific sea level pressure data: an analysis and reconstruction Region — a Review for Pilots. New Zealand Meteoro- back to 1957, Journal of Climate, 1, 1199-1220. logical Service Miscellaneous Publication 166, 35pp. Khatep, Fitzharris, B.B. and Bardsley, W.E. 1974. Streten, N.A. 1975. Satellite derived inferences to some Water vapor transfer over the Southwest Pacific: characteristics of the South Pacific atmospheric circu- Mean patterns and variations during wet and dry lation associated with the Nino event of 1972-73, periods, Monthly Weather Review, 112, 1960-1975. Monthly Weather Review, 103, 989-95. Kiladis, G. and van Loon, H. 1988. The Southern Oscilla- Trenberth, K.E. 1976. Spatial and temporal variations of tion. Part VII: Meteorological anomalies over the the Southern Oscillation, Quarterly Journal of the Indian and Pacific sectors associated with the ex- Royal Meteorological Society, 102, 639-653. tremes of the oscillation, Monthly Weather Review, Trenberth, K.E. and Shea, D.J. 1987. On the evolution of 116, 120-136. the Southern Oscillation, Monthly Weather Review, Lockyer, N. and Lockyer, W.J.S. 1902a. On some phenom- 115, 3078-3906. ena which suggest a short period of solar and meteoro- Walker, G.T. and Bliss, E.W. 1932. World weather V, logical changes, Proceedings of the Royal Society Mem. Roy Meteorological Society, 4, 53-84. London, 70, 500. Walker, G.T. and Bliss, E.W. 1937. World weather VI, Lockyer, N. and Lockyer, W.J.S. 190211 On the similarity Mem. Royal Meteorological Society, 4, 119-139. of the short-period pressure variation over large ar- Wright, RR 1975. An Index of the Southern Oscillation. eas, Proceedings of the Royal Society of London, 71, Climatic Research Unit Report, CRU RP4, University 134-135. of East Anglia, Norwich, England, 22pp. Lockyer, N. and Lockyer, W.J.S. 1904. The behaviour of Wright, P.B. 1989. Homogenized long-period Southern the short-period pressure variation over the earth's Oscillation indices, International Journal of Climatol- surface, Proceedings of the Royal Society London, 73, ogy, 9, 33-54. 457-470.