The evolution of the weak El Niño 2004-2005
Tony Barnston, International Research Institute (IRI)
El Niño and La Niña episodes differ from one another, not only in their relative strengths, but also their seasons of onset, maturity, and demise, as well as the locations of their maximum sea surface temperature (SST) anomaly within the tropical Pacific. The peak SST anomaly at the warmest location in the tropical Pacific during an El Niño may exceed 5°C, as was the case in the great 1997-98 episode, or be only near 2°C, as in the 1994-95 or the 2004-05 episodes. The last episode, which just dissipated in early February, was weak enough and "non-standard" enough that many experienced oceanographers and climatologists question whether it should be regarded as an El Niño at all. The answer to this depends upon one’s definition, and the issue of an acceptable definition remains nearly as elusive today as it was 10 years ago. In this short piece we examine some aspects of the recent episode’s evolution, looking in particular for signs that it was, or was not, an El Niño in terms of what most of us have come to look for in one.
El Niño involves both oceanic and atmospheric behaviour in and around the tropical Pacific, and commonly used definitions tend to be based either on an atmospheric manifestation (e.g., the Southern Oscillation Index [SOI] – the difference in sea level pressure between that at Tahiti, French Polynesia, and that at Darwin, Australia) or an oceanic manifestation (e.g., the average SST in an equatorially centred rectangular region such as NINO3 or NINO3.4). There have been attempts to define El Niño using a set of variables from both atmosphere and ocean (e.g., the Multivariate ENSO Index [MEI]). Usually, if an El Niño does not begin to develop by the end of July, it is fairly unlikely to develop during that entire annual cycle, but may do so during April through July of the following year. Occasional exceptions have occurred when an El Niño emerged in August or even September, as for example in 1986. In 2004 the onset of El Niño was late, and most climate forecasting centres had all but written off the possibility of one by the time it appeared during the last half of July. Even as the NINO3.4 weekly SST anomaly first exceeded 0.5°C in mid July, the IRI forecasted only a 40% probability for El Niño later in the year, a 55% probability for neutral conditions, and 5% for La Niña. In August, all forecast centres acknowledged that present SST conditions had warmed to a level which, if sustained over several months, could likely later be called El Niño SST conditions.
Some key factors in the evolution of the 2004-05 El Niño, and the somewhat stronger 2002-03 El Niño, are shown in Figure 1. The three panels show, from left to right, anomalies of zonal wind, ocean heat content, and SST. Time marches downwards from early 2002 to March 2005, capturing both El Niños. Longitude spans from Indonesia (left side) to somewhat offshore of South America (right) in each panel. The 2002-03 El Niño was stronger, as shown by the SST anomalies exceeding 2.5°C near 170°W longitude in October 2002, while the anomalies during the 2004-05 El Niño exceeded only 1.5°C near the Date Line in late October and early November 2004. Perhaps more importantly, the spatial pattern of the SST anomalies was such that the area of SST exceeding 1°C during 2004-05 never expanded eastward of about 140°W except very briefly in late July when the warming initially developed. Related was the fact that the zonal wind anomalies were limited to the central Pacific in much the same way – the trades hardly weakened in the eastern Pacific except for a few very short-lived intervals. Westerly wind anomalies were somewhat more broadly observed during middle and late 2002, including an episode of eastward-expanded anomalies in May, preceding a general increase in oceanic heat content and finally a warming of the SST near and eastward of the Date Line. Even the 2002-03 El Niño is considered to have been focused in the central, as opposed to central-plus-eastern, tropical Pacific.
The zonal wind and heat content anomalies during the 2004-05 El Niño appear as individual pulses of one month (or less) duration, rather than as continuous and broad anomalies as observed in stronger El Niños, such as those of 2002-03 and even more so in still stronger ones. This wave-like feature may be related to the El Niño being supported largely by the westerly wind events associated with the Madden-Julian Oscillation (MJO) rather than by the slower acting physics related to a progressive accumulation of anomalous heat in the near-equatorial water volume west of the Date Line. While the MJO may play a role in initiating many El Niños, in most cases its role is thought to be catalytic rather than basic. In 2004-05, all three of (1) westerly wind anomalies, (2) increases in SST, and (3) anomalous convection were limited to near, and somewhat west of the Date Line, as opposed to expanding further east through positive feedbacks as observed in more typical El Niño events. Throughout most of the second half of 2004, although the NINO3.4 SST index was high enough to qualify as a weak El Niño, the NINO3 SST and the SOI indices did not deviate sufficiently from neutral to categorise the 2004-05 event as an El Niño. Indeed, with the lack of atmosphere-ocean coupling, only a few of the most basic expected climate teleconnections were observed during the last quarter of 2004 and January 2005 – such as below normal rainfall in much of Indonesia and the Philippines, and in part of southeastern Africa and Central America.
In February 2005, anomalous convection finally did appear, and strongly, near and just east of the Date Line – and the SOI dipped to very low levels just for that month. Ironically, the NINO3.4 SST anomaly had returned to neutral levels during February, so that in no month were both the SOI and the NINO3.4 SST above El Niño-indicating thresholds. The anomalous convection during February and early March was sufficient to induce some global climate impacts that had been absent before February: marked dryness in eastern Australia, wetness in many parts of southeast U.S., more severe dryness in much of southeastern Africa, dryness at the start of the rainy season in northeastern Brazil, and heavy rains further east in the central tropical Pacific itself, such as in Nauru and Western Kiribati.
The fact that there were no months during which both ocean and atmosphere exhibited El Niño behaviour raises the question of whether the 2004-05 episode can be called an El Niño, and whether an appropriate definition of El Niño should involve more than an oceanic index alone or an atmospheric index alone – given that the two usually, but not always, behave consistently because of the coupling process mentioned above. The choice of index within either medium alone presents an additional challenge. In 2004-05, the NINO3.4 SST index held within the weak El Niño range for at least six months, but NINO3 did so for only one to two months. This inconsistency between SST indices is due to the fact that the strongest SST anomalies were located slightly west of the Date Line – quite atypical of an El Niño. There is also choice in selection of an atmospheric index, as for example the standard Tahiti minus Darwin SOI versus the equatorial SOI. In February 2005, the standard SOI was very low, but the equatorial SOI was much closer to average.
In conclusion, the categorisation of 2004-05 as a weak El Niño, depends on one’s definition. The choice of definition, in turn, may depend on which aspects of El Niño create climate responses in the country or region in question. Taking this idea even further, to many people in ENSO-impacted regions, El Niño refers historically to their local El Niño-associated climate condition rather than to the physically governing conditions in the tropical Pacific Ocean. Because of these differences in understanding, developing a single definition for El Niño remains a complex challenge.