The crucial role of the southern midlatitudes in forcing extreme El-Niño - Southern Oscillation events

.J. Stephens & M.H. Lamond†

Agriculture Western Australia, South Perth WA 6151, Australia † Lamond Weather Services, Nedlands WA 6009, Australia

† Lamond Weather Services, Nedlands WA 6009, Australia El Niño/Southern

Oscillation events severely disrupt the global environment (1) with stronger events causing the loss of thousands of lives and billions of dollars of damage (2, 3). Forecasting systems have been developed that rely on equatorial Pacific indicators (4,5), but these all failed to predict significant ocean warming prior to the onset of the severe 1997-98 El Niño (6-8). It has previously been proposed that the re-positioning of low pressure anomalies associated with the Antarctic circumpolar trough plays an important role in El Niño development (9-11). Here we confirm this trough movement, showing that the magnitude of low pressure anomalies over south-eastern Australia (in austral winter/spring) is transferred as a standing wave oscillation to the South Pacific (in the following El Niño autumn/winter). The magnitude of pressure anomalies in the central South Pacific is then strongly related to trade winds anomalies in the equatorial Pacific at the transition stages of extreme events. Stronger variations of the Southern Oscillation therefore lead stronger El Niños, providing the basis for an early warning system to assist disaster prevention and resource management.

Prior to linking El Niño events to the Southern Oscillation in pressure (ENSO)(12), Bjerknes (13) stated that the South Pacific high should play an important role in El Niño development as it directs the easterly trade winds along the equator between South America and the dateline. Normally the high pressure and trades weaken between April-July when the South Pacific surface trough (Circumpolar trough) amplifies and reaches its farthest northward extent (9). Over the year leading into El Nino, the South Pacific trough changes from a weakened to an enhanced state (9), which causes a pressure reversal or ‘Southern Oscillation’ in the midlatitudes (10,11) and other important wind and sea-surface temperature (SST) anomalies (14-17)(summarised in Fig.1). These features were very pronounced in maps showing the development of the strong 1997-98 El Nino (18).

In this study, mean sea-level pressure (MSLP) anomalies were analysed at stations at opposite ends of the pressure sea-saw in Fig.1 to see if the development and strength of El Nino events could be predicted with more lead-time. Monthly data between 1950-99 were normalized by their respective standard deviations and averaged as 3-month running means to minimise the Southern Oscillation ‘signal’ being dominated by monthly noise (14).

In the year before El Nino (Yr (-1)), significant negative pressure anomalies occur over south-eastern Australia (10,11,16) where the circumpolar trough is peaking at higher levels (Fig. 1A). A composite El Niño Prediction Index (EPI) was therefore derived based on the mean normalised pressure anomalies between Alice Springs (a) and Mildura (m) 1400km to the south-east (Fig. 1) i.e.

EPI = () Max Jul (-1) Sep (-1) ((NPA3 Alice Springs) + (NPA3 Mildura))/2

= () Max Jul (-1) Sep (-1) (1)

where () is the sign of the anomaly and Max gives the absolute magnitude of the maximum anomaly between July and September in Yr(-1), NPA3 is normalised 3-month mean MSL pressure anomaly, n is month number, P is monthly MSL pressure, P is long-term monthly mean pressure, S is the standard deviation in monthly pressure.

The relationship between the EPI and the change in central and eastern equatorial SST in the Nino3 region over the following 12-month period is shown in Fig. 2. Over half (52%) of the variance is explained in the change in SST from September-December Yr (-1) to the same interval in the El Nino year (Yr (0)). Five of the six largest increases in SST coincide with the most negative values of EPI in Yr (-1), while the 5 most positive values of the EPI occurred in El Niño type years when the Southern Oscillation Index (SOI) was strongly negative. All these latter years preceded the end of warm events, and with the exception of the extended warm event of 1982/83, moved into La Niña (cold event) conditions. If the EPI had been monitored prior to 1997, a severe El Niño alert could have been issued in early October 1996, 7-8 months prior to other forecasts.

The severity of different El Nino events is affected by SST anomalies peaking at different times, so the EPI was related to two composite measures of El Nino strength based on the Bjerknes ENSO Index (BEI) (17). A measure of El Nino Intensity, BEI3 Intensity, is the maximum 3-month mean value of BEI in a year. A second, BEI3 Power, sums monthly BEI3 values greater than a threshold 2.5 and is a combined measure of El Niño intensity and duration. For the 11 El Niño events defined since 1950 (17), 87% of the variance in intensity was explained by the EPI with a logarithmic relationship (Fig. 3a). For BEI3 power, 82% of the variance was explained with the severe events of 1982-83 and 1997-98 dominating (Fig. 3b). Since a three-fold increase in power occurs as the EPI increases from -1.4 to -1.7, El Niño development appears to be increasingly sensitive to atmospheric forcing from large pressure and associated wind anomalies measured by the EPI.

Once a very low EPI is recorded, the pressures in the central South Pacific begin to become negative over the warm SST anomalies that persist in the more northerly flow (10). This is illustrated in Fig. 43 where pressure data and area averaged SST data over the central South Pacific (160?W-120?W, 20?S-30?S) and the Nino3 region are normalized and averaged for the 11 El Nino events to form a composite time-series. Pressure at Rapa Island (144?W, 28?S) in the central South Pacific becomes increasingly negative as the surrounding SST remains warm through to April (0). Negative surface pressures would encourage the re-positioning of the circumpolar trough (Fig. 1B) which erodes the strength of the South Pacific high and trades (represented by Easter Island) and leads a fall in pressures along the equator (represented by Tahiti). Not surprisingly , positive SST anomalies rise most rapidly between April-July when the South Pacific trough is normally amplified (9).

Fig. 4 also highlights the pressure reversal in the mid-latitudes between winter (-1) and autumn/winter (0). An index of this re-positioning of the surface trough is the difference in MSLP between Rapa Island and south-eastern Australia (10). Equivalent to a mid-latitude Southern Oscillation Index (MLSOI), this is best represented by the difference in normalized MSLP anomalies between Rapa Island and an average of Alice Springs (a) and Mildura (m). The largest 4-month mean of this difference between the months of April-June (Yr (0)) forms a second El Niño Prediction Index (EPI2). i.e.

MLSOIn =

EPI2 = (± ) Max Apr (0) ® Jun (0)

where the same terms are used as in Eq.1.

As the EPI becomes more negative in Yr (-1)(Fig. 4A), the greater will be the reverse difference in pressure between these two regions early in Yr (0) (Fig. 4C). The three lowest values of EPI (in 1996, 1981, 1971) preceded (in order of magnitude) both, the three lowest values of EPI2, and the three most deepened South Pacific troughs between April-July (0). Thus, the maximum amplitude of the trough in the westerlies in the Australian region in Yr (-1) is transferred to the opposite anti-node in the following year through a standing wave oscillation. At the eastern anti-node, very large negative (positive) MSLP anomalies at Rapa Island appear to play a vital role in equatorial warming (cooling) further north in the Nino 3 region (Fig. 5A). Warm and cold events are clearly identified with divergence between the two time-series, with the largest Rapa pressure anomalies leading the largest SST anomalies. The unprecedented rapid rise and fall in SST in the 1997 El Nino and the 1998 La Nina follow the unprecedented enhancement and weakening of the South Pacific trough between April-July in these respective years.

The primary cause of the relationship in Fig. 5A is the influence that MSLP at Rapa has on the trade winds. This can been seen in Fig. 5B, where Rapa MSLP anomalies are plotted against the anomalies in the area-averaged 850 hPa trade wind index for the Western Pacific (135?E-180?W, 5?N-5?S). Pressure anomalies at Rapa appear to lead anomalies in the trades, but more importantly, coincide in anomaly strength at the beginning of warm (1982, 1986-87, 1991, 1997) and cold (1988, 1995, 1998) events. This is typically the case for the critical months between April-July when the South Pacific trough extending more into the tropics (9).

A deepened trough in the vicinity of Rapa Island must then relate to the trade winds for two reasons highlighted in Fig. 1B. Firstly, the increased equatorward flow on the south-west (rear) side of the trough adds to the westerly momentum change occurring further north along the equator (19,20). Secondly, the westerly wind anomalies to the north of the trough weaken the South Pacific high and the pressure gradient across the equatorial Pacific which reduces the transfer of air (trades) from the south-eastern Pacific to the Indonesian low (21). A favourable environment therefore exists for: 1) westerly wind bursts and Kelvin wave activity to extend further east (22,23), weaker equatorial undercurrent transport in the thermocline (24), and 3) weakened upwelling all the way along the equator from South America to the mid-Pacific (12). These all add to deepen the thermocline and increase SST in the eastern and central Pacific.

Forcing for the decay of El Nino is also explained in Fig. 4. Once very negative pressures occur at Rapa (Apr-Aug(0)), the South Pacific trough directs cold southerly winds into the regions of the SPCZ (Fig 1B) and SST in the vicinity of Rapa quickly turn cold. Pressures at Rapa then rise rapidly leading to an increase in the strength of the South Pacific high (Easter Island pressure) and trade winds. The decay of the 1997-98 El Nino graphically illustrated this forcing of wind anomalies from the high pressure anomalies in the southern midlatitudes between November 1996-April 1998 (18).

Major changes in the midlatitudes do however need to coincide with favourable equatorial processes for an El Niño to develop. The EPI would have predicted moderate El Niño events in 1956, 1979 and 1985 (Fig. 2), but a lack of net convergence on the equator in the Western Pacific in early 1979 prevented mass being released down the equatorial waveguide as a Kelvin surge (25). This also needs investigating for 1956 and 1985. Indicators that can distinguish between weak El Niño years and non-El Niño years also need developing. In 1950, the EPI was positive (Fig. 4) and did not indicate the subsequent weak El Niño in 1951. In this case, negative MSLP anomalies at Darwin in winter (-1) did indicate a warm event (21,26) and these appeared to move down into the SPCZ (Nov(-1)-Feb(0)) and then onto Rapa (Mar-Jul (0)) (Fig. 5a) as part of an enhanced annual cycle of outgoing longwave radiation (27). Since there was no pressure reversal in the mid-latitudes, only Tahiti became significantly negative while Darwin had only slight positive anomalies.

This study puts in question the current understanding that ENSO originates in the tropical Pacific (4) and that it cannot be predicted a year in advance (28). As was quoted by van Loon and Shea (10), “Coming events cast their shadows before” (29). Stronger cases of the Southern Oscillation in pressure lead stronger El Nino events on the equator and this is the key to better forecasts with greater lead-times. Consequently, the ENSO Observing System (5) needs to extend its equatorial focus further south (to 45?S) to include the Australian-South Pacific sector and ENSO modelling should be widened to incorporate mid-latitudinal shifts in trough patterns. Since the magnitude of disastrous droughts, floods and oceanic events are related to the strength of El Niño events, the EPI offers an early warning system that should see an improved capacity to mitigate the effects of these disasters in the 21st century (30). Resource management in many parts of the world could also be improved for these extreme events.

  1. Allan, R., Lindesay, J. & Parker, D. El Niño Southern Oscillation and Climatic variability (CSIRO Publ., Melbourne, 1995).
  2. Ptaff, A., Broad, K. & Glantz, M. Who benefits from climate forecasts? Nature, 307, 645-646 (1999).
  3. Kerr, R.A. Big El Niños ride the back of slower climate change, Science, 19, 1108-1109 (1999).
  4. Latif, M. et al. A review of The predictability and prediction of ENSO. J. Geophy. Res., 103, 375-393 (1998).
  5. McPhaden, M.J. The child prodigy of 1997-98. Nature 398, 559-562 (1999).
  6. Barnston, A.G., Glantz, M.H. & He, Y. Predictive skill of statistical and dynamical climate models in SST forecasts during the 1997-98 El Niño episode and the 1998 La Niña onset. Bull. Amer. Meteorol. Soc., 80, 217-243 (1999).
  7. Trenberth, K.E. Development and forecasts of the 1997/98 El Niño. CLIVAR scient. iss., 3, 4-14 (1998).
  8. Anderson, D.L.T. & Davey, M.K. Predicting the El Niño of 1997/98. Wea., 53, 303-310 (1998).
  9. van Loon, H. The Southern Oscillation. Part III, Associations with the trades and with the trough in the westerlies of the South Pacific Ocean. Mon. Wea. Rev., 112, 947-954 (1984).
  10. van Loon, H. & Shea, D.J. The Southern Oscillation. Part IV: The precursors south of 15ºS to the extremes of the oscillation. Mon. Wea. Rev., 113, 2063-2074 (1985).
  11. van Loon, H. & Shea, D.J. The Southern Oscillation. Part VI: Anomalies of sea level pressure on the Southern Hemisphere and of Pacific sea surface temperature during the development of a warm event. Mon. Wea. Rev., 115, 370-379 (1987).
  12. Bjerknes, J. Atmospheric teleconnections from the equatorial Pacific. Mon. Wea. Rev., 97, 163-172 (1969).
  13. Bjerknes, J. A possible response of the atmospheric Hadley Circulation to equatorial anomalies of ocean temperature. Tellus, 97, 820-829 (1966).
  14. Trenberth, K.E. Spatial and temporal variations of the Southern Oscillation. Quart. J. Roy. Meteorol. Soc., 102, 639-653 (1976).
  15. Rasmusson, E.M. & Carpenter, T.H. Variations in tropical sea surface temperature and surface wind fields associated with the Southern Oscillation/El Niño. Mon. Wea. Rev., 110, 354-384 (1982).
  16. Harrison, D.E. & Larkin, N.K. The COADS sea level pressure signal: a near-global El Niño composite and time series view, 1946-1993. J. Clim., 9, 3025-3055 (1996).
  17. Harrison, D.E. & Larkin, N.K. El Niño - Southern Oscillation sea surface temperature and wind anomalies, 1946-1993. Rev. Geophy., 36, 353-399 (1998).
  18. Wang, C. & Weisberg, R.H. The 1997-98 El Nino evolution relative to previous El Nino events. J. Clim., 13, 488-501 (2000).
  19. Harrison, D.E. The appearance of sustained equatorial surface westerlies during the 1982 Pacific warm event. Science, 224, 1099-1102 (1984).
  20. Chen, L. & Wu, R. The role of the Asian/Australian monsoons and the Southern/Northern Oscillation in the ENSO cycle. Theor. Appl. Climatol., 65, 37-47 (2000).
  21. Quinn, W.H., Zopf, D.O., Short, K.S. & Kuo Yang, R.T.W. Historical trends and statistics of the Southern Oscillation, El Nino, and Indonesian Droughts. Fish. Bull.,76, 663-678 (1978).
  22. Wyrtki, K. El Niño-The dynamic response of the equatorial Pacific Ocean to atmospheric forcing. J. Phys. Oceanogr., 5, 572-584 (1975).
  23. Philander, S.G.H. The response of equatorial oceans to a relaxation of the trade winds. J. Phys. Oceanogr., 11, 176-189 (1981).
  24. Yu, X. R. & Mc Phaden, M.J. Dynamical analysis of seasonal and interannual variability in the equatorial Pacific. J. Phys. Oceanogr., 29, 2350-2369 (1999).
  25. Mitchum, G.T. Trade wind fluctuations associated with El Nino-Southern Oscillation events. J. Geophy. Res., 92, 9464-9468 (1987).
  26. Trenberth, K.E. & Shea, D.J. On the evolution of the Southern Oscillation. Mon. Wea. Rev., 112, 3078-3096 (1987).
  27. Meehl, G.A. The annual cycle and interannual variability in the tropical Pacific and Indian Ocean regions. Mon. Wea. Rev., 115, 27-50 (1987).
  28. Wolter, K. & Timlin, M. Measuring the strength of ENSO events: How does 1997/98 rank? Wea., 53, 315-324 (1998)
  29. Campbell, T. Lochiel’s warning. Republished 1968 in The Complete Poetical Works of Thomas Campbell, (ed Robertson, J.L.) 157-160 (Haskell House, 1802).
  30. Stephens, D.J. & Lamond, M.H. Reducing the impact of major droughts in the Indonesian-Australian region through the monitoring of atmospheric pressure anomalies in the preceding year. Proc. Aust. Disaster Conf., 399-404 (1999).

Acknowledgements. We thank I. Smith (CSIRO) for providing South Pacific SST data. A. Finet and (Meteo-France), J. Salinger (N.I.W.A), D. Shea (NCAR) all provided MSLP data for the South Pacific. H. van Loon (NCAR) offered invaluable advice and his original work was instrumental in this analyis.

Correspondence and requests for materials should be addressed to the author (e-mail: dstephens@agric.wa.gov.au).

 

Fig. 1. Schematic diagram showing major weather systems and anomalies in the Southern Hemisphere in winter/spring in (A) the year before El Niño (Yr (-1)) and (B) during El Niño (Yr (0)). The Walker Circulation (white) and circumpolar troughs (red curve) sit above surface anomalies measured at key stations represented by crosses (Alice Springs (A), Darwin (D), Easter Island (E), Mildura (M), Noumea (N), Rapa Island (R), Tahiti (T), Willis Island (W)). The South Pacific Convergence Zone (SPCZ) is represented by parallel grey lines. Warm (red) and cold (green) SST anomalies are indicated, as are anomalous wind anomalies (black arrows). High and low MSLP are indicated by “H” and “L” and anomalous pressures in midlatitudes are dashed.

Fig. 2. Scatter diagram and linear line of best fit showing the change in average (Sept-Dec) Nino3 SST from Yr (-1) to Yr (0) versus the EPI in Yr (-1). Yr (0) El Niño years (solid circles), years El Nino end (solid diamonds) and other years (solid triangles). Fig. 3. Scatter diagrams showing the EPI (in Yr (-1)) versus: (A) El Niño Intensity (solid triangles, logarithmic line of best fit), and (B) El Niño Power (solid diamonds, power function line of best fit). (C) The relationship between the EPI2 early in Yr (0) and El Niño intensity (open circles, linear line of best fit).

Fig. 3. Scatter diagrams showing the EPI (in Yr (-1)) versus: (A) El Niño Intensity (solid triangles, logarithmic line of best fit), and (B) El Niño Power (solid diamonds, power function line of best fit). (C) The relationship between the EPI2 early in Yr (0) and El Niño intensity (open circles, linear line of best fit).

Fig. 4. Composite time-series of averaged normalized SST anomalies in Nino3 region (red, open diamonds) and the central South Pacific ( brown, open squares) and MSLP anomalies for the 11 most recent El Niño events at Rapa Island (green, solid circles), Easter Island (black, open circles), Tahiti (purple, solid triangles), south-eastern Australia (average of Alice-Springs and Mildura; blue, open triangles). All pressure data have been smoothed by a 3-month running mean.

Fig. 5. (A) Plot of normalized Rapa Island MSLP anomalies since 1951 (black line, left (y) axis) versus actual Nino3 SST anomalies (red and blue colour fill, right (y) axis). (B) Plot of normalized Nino3 SST anomalies (red curve) versus normalized Rapa Island MSLP anomalies (blue curve) and normalized 850 hPa West Pacific trade winds (purple curve) (1979-1999). All curves have been smoothed using a 5-month running mean.

Fig. 2

Fig. 3

Figure 1a

Figure 1b

Figure 2

 

Figure 3

 

 

 

 

Figure 4

 

Figure 5

 

Figure 6

 

Figure 7a

 

Figure 7b

Public Data: About Us | Services | Subscribe | El Nino | Links | Contact Us
Members Only: 2005 Outlook | Monthly Update | Weekly Update | Commentary | Links

© 2000 - 2005 Lamond Weather Services. All Rights Reserved. Designed by PerthWeb.