Q. J. R. Meteorol. Soc. (2000), 126, pp. 807–821 doi: 10.1256/qj. A self-sustaining climate mode in the tropical Atlantic, 1995-1997: observations and modelling By Itsuki C. Handoh * and Grant R. Bigg University of East Anglia, UK (Received 7 March 1999; revised 4 November 1999) SUMMARY An anomalous climatic event in the tropical Atlantic, starting in autumn 1995 and continuing into the autumn of 1997, is investigated. Using remotely-sensed data sets and a reduced-gravity model, it is shown that this event contained both a warm and a cold phase, each of roughly 12 months duration. The propagating signal within each phase is identified in the sea surface height and temperature signatures as both equatorial and off-equatorial waves in the ocean, with coupling to atmospheric convection. The strength, geographical pattern and timing of the anomalous period is consistent with it being the first observed example of a coupled mode of interaction between the ocean and atmosphere found in two previous coupled ocean-atmosphere models of the tropics. This interpretation means that the two-phased climate event, which we will call the Equatorial Atlantic Oscillation (EAO), was associated with purely internal atmospheric and oceanographic variability within the Atlantic basin and hence was independent of the Pacific climate. It can also be inferred, from long-term monthly sea surface temperature and sea level pressure data sets, that there were potentially several previous EAO events during the past century. KEYWORDS: coupled ocean-atmosphere interaction tropical climate Atlantic remote sensing 1. I NTRODUCTION The anomalous climate of the Atlantic of 1983-84 (Philander 1986) has been in- ferred as being triggered by the 1982/83 El Ni˜ no in the Pacific (Horel et al. 1986; Carton and Huang 1994; Delecluse et al. 1994). The Atlantic ocean and atmosphere then displayed very similar characteristics to those typically seen in the Pacific dur- ing El Ni˜ no, namely, weakening equatorial easterlies coupled with rising sea surface temperatures, moving eastwards. Other such warm events have been observed in the tropical Atlantic (Servain 1984; Lamb et al. 1986; Carton and Huang 1994; Arnault and Cheney 1994), typically lagging El Ni˜ no events by 4-5 months (Enfield and Mayer 1997). However, there are not always clear links to the Pacific climate. This was the case in 1988 (Carton and Huang 1994); the 18 month lag found at other times by Tourre and White (1995) was thought to be due to a passive response to a globally perturbed atmosphere rather than an immediate consequence of Pacific climate variability. Huang et al. (1995) showed direct forcing of the Atlantic ocean by variability in its overlying atmosphere through the 1980s, but as their experiment was with an ocean-only model they could not adequately explore any teleconnections between these local winds and remote forcing from the Pacific. Zebiak (1993) showed in a coupled ocean-atmosphere modelling study that a sequence of warm and cold events with a similar periodicity to the Southern Oscillation could be internally generated within the Atlantic basin, with no necessity for external forcing from the Pacific. The basin was, however, unable to sustain these oscillations beyond a decade or two. Successive warm and cold events have been seen in the observational record (Servain et al. 1982; Lamb et al. 1986; Ser- vain 1991), although this has not always been noted as a distinct feature of the climatic variability. The potential for the Atlantic basin to exhibit a coupled ocean-atmosphere climate oscillation completely independent of the Pacific may thus exist but has not * Corresponding author: School of Environmental Sciences, University of East Anglia, Norwich NR4 7TJ, UK. e-mail: i.handoh@uea.ac.uk; Now at Department of Applied Mathematics, The Hicks Building, Hounsfield Road, University of Sheffield, Sheffield S3 7RH, UK. e-mail: i.handoh@shef.ac.uk. c  Royal Meteorological Society, 2000. 807