Imprint of El Niño/La Niña Cycles on
the Humboldt Current System^*

Gary Shaffer¹, Oscar Pizarro²,
Samuel Hormazabal¹, Silvia Venegas¹

¹Danish Center for Earth System Science, Niels Bohr
Institute for Astronomy, Physics and Geophysics,
University of Copenhagen, Juliane Maries Vej 30, 2100
Copenhagen OE, Denmark, E-mail: gs@dcess.ku.dk
² Programa Regional de Oceanografía Física y Clima,
Universidad de Concepcíon, Casilla 160-C,
Concepción 3, Chile

The El Niño-Southern Oscillation (ENSO) affects climate around the world on interannual time scales. Warm and cold phases of ENSO - El Niño (EN) and La Niña (LN) - are thought to arise primarily from atmosphere -ocean interactions in the tropical Pacific Ocean whereby low frequency, equatorial trapped waves in the ocean play a key role. A main mode of such waves, the eastward-propagating, equatorial Kelvin wave, and poleward-propagating, coastal trapped waves in the ocean along the west coast of the Americas form an efficient oceanic pathway for transmission of tropical Pacific signals to mid-latitudes. In particular, the west coast of South America, devoid of obstructions like the Gulf of California, is in this way at the mercy of the climatic whims of the tropical Pacific. The plot thickens as there also exist several atmospheric pathways by which tropical Pacific signals can be transmitted to the South American coast. It is therefore useful to consider the imprint of EN/LN cycles on the Humboldt Current system in terms of these atmospheric and oceanic pathways.

A statistical analysis of sea surface temperatures (SST) and winds from the tropical and South Pacific regions over the last 50 years yields two EN/LN modes of about 3 year and about 5 year period, modulated by decadal-scale variability. The strong EN events of 1982-83 and 1997-98 coincide with times when these two modes interact constructively (a 15 year separation is a multiple of both 3 and 5 years). As a corollary, the EN event now brewing (March 2002) will likely only develop to be a moderate one. Another result of this analysis is a well-defined SST dipole whereby warm (cold) conditions along the equator correspond to cold (warm) conditions in a band near 30ºS in the central South Pacific. This subtropical anomaly tends to propagate eastward toward the coast of central Chile. In this way a warm anomaly reaches this coast during the year before the onset of an EN event in the tropical Pacific and would appear to be an EN precursor. Warming off central Chile has indeed been observed before EN events (for example in 1996 before the 1997-98 event and in 2001 before the ongoing event). Such "precurser" behavior has been discussed in early and more recent work on interannual variability in the eastern South Pacific.

Disturbances with periods longer than about 150 days, i.e. including EN/LN cycles, can propagate seaward as mid-latitude, Rossby waves off northern and central Chile. Indeed, observations indicate that coastal signals propagate seaward as such waves, strongly modulating the Humboldt Current system. Amplitudes of very long period, equatorial Kelvin waves at the South American west coast can be calculated with a simple, wind-driven model of the equatorial Pacific and used to estimate amplitudes of mid-latitude, Rossby waves. For such long periods, coastal sea level and the poleward, Peru-Chile Undercurrent are found to be well correlated with a 90° phase difference, consistent with Rossby wave dynamics. This may explain why the maximum poleward flow in the Undercurrent off central Chile is observed to occur before EN events. The mid-latitude, Rossby wave associated with the 1982-83 EN event was observed to cross the North Pacific from east to west in about 10 years. In contrast, evidence from the South Pacific points toward more rapid dissipation of EN-related, mid-latitude Rossby waves. During the strong EN events of 1982-83 and 1997-98, sea level rise and warming extended along the Chile coast at least to 35°S. Warming was stronger off northern Chile than off central Chile and surface waters off northern Chile were replaced by saltier water of more equatorward origin.

Disturbances with periods shorter than about 150 days remain trapped near the coast off northern and central Chile. Strong, intraseasonal variability of sea level, coastal currents and SST with periods between about 40 and 70 days has been observed within a narrow coastal band of 50-100 km width along the west coast of South America, at least as far south as 33×S. Much of this variability is due to free, coastal trapped waves, generated when equatorial Kelvin waves impinge upon the South American coast. The Kelvin waves are forced in the equatorial Pacific by zonal wind events of intraseasonal period, associated with the Madden-Julian Oscillation. This equatorial-midlatitude connection is confirmed by good agreement between very simple model simulations forced by satellite winds from the equator and the South American coast and observed fluctuations of currents and sea level off central Chile. However, south of 20°S, some oceanic intraseasonal variability is forced by local winds due to equatorial-midlatitude teleconnections in the atmosphere. Indeed, intraseasonal fluctuations of SST near the central Chile coast are more related to local wind forcing than to the action of the free, coastal trapped waves. Since intraseasonal winds in the central and western equatorial Pacific are strongest during the initial phase of EN events, intraseasonal variability off Chile would then be expected to be strongest during EN events. This is confirmed by observations.

Upper ocean current observations made 150 km from the coast off central Chile, i.e. seaward of the narrow coastal band and within the coastal transition zone, exhibit large variability at periods between 100 and 200 days. Some of this variability can be explained by Rossby wave propagation as described above. However, much of this variability may be associated with eddies formed by local baroclinic instabilities as shown in high-resolution, 3-D ocean model simulations of the region. The observations indicate more variability in the 100-200 day band during LN events than during EN events at this offshore site. Guided by the results of seasonal simulations with the 3-D model, we speculate that this behavior may be related to less available potential energy (to feed into the baroclinic instabilities) during EN events when coastal-offshore temperature differences decrease. Current observations at the offshore site show no significant interannual variability in upper ocean flow over EN/LN cycles. However, deep currents at 2450 m and 3750 m depth did exhibit such variability whereby stronger poleward flow was observed during the 1997-98 EN event.