Background & Need

Background and Need for the Tropical Pacific Observing System 2020 Project

This material is modified from ENSO Research: The Overarching Science Drivers and Requirements for Observations (White Paper #3)

El Niño Southern Oscillation (ENSO) is the largest interannual climate signal, with its coupled ocean-atmosphere core in the equatorial Pacific but producing global effects through atmospheric teleconnections. The unpredicted El Niño event of 1982-83 focused attention on this phenomenon and was the impetus for the original development of the tropical Pacific observing system (TPOS) and tropical atmosphere ocean (TAO- link to array in the mid-1980s, under the Tropical Ocean/Global Atmosphere (TOGA) program.

ENSO has played two roles in stimulating the observational network and its modelling counterpart: it is a forecast problem with obvious practical benefits, and it is a laboratory for the study of tropical air-sea interaction, which informs the study of other phenomena in other ocean basins.

Before the 1960s, no mechanism had been proposed to connect the well-known sea surface temperature (SST) anomalies known as El Niño on the Peruvian coast with winds and precipitation elsewhere in the basin. Bjerknes (1966) described the fundamental coupled interaction between surface winds and SST gradient, but the mechanism of eastern warming continued to be assumed local until Wyrtki (1975) showed that this signal must be remotely forced by wind anomalies in the west, and thus that ENSO was a basin-scale phenomenon. During the 1980s, models and theory (e.g., Anderson and McCreary, 1984) showed the essential role of equatorial Kelvin and Rossby waves in transmitting the signal across the Pacific, and an explosion of work in the late 1980s explained and fleshed out “delayed oscillator” ideas which established ENSO as a coupled oscillation with a distinct and explainable evolution (Zebiak and Cane, 1987; Battisti and Hirst, 1989; Schopf and Burgman, 2006; and Suarez and Schopf, 1988), although its genesis was (and remains) controversial.

At that time, observation of the subsurface ocean in the tropical Pacific was done primarily by expendable bathythermographs (XBTs) deployed from merchant ships, producing at best monthly samples on a few meridional lines barely sufficient to detect the relatively slow Rossby waves (Kessler, 1990). Island sea level gauges added higher temporal resolution in a few places (Wyrtki, 1975 and 1981), but the absence of near-equatorial islands in the more than 7000km between the Line Islands (160°W) and the Galapagos (95°W) made it impossible to observe the Kelvin waves. Nevertheless, recognition of the role of these linear wind-driven waves gave some skill in using simple wave models and early general circulation models (GCMs) to predict the evolution of El Niño events over a lead time of a few months (Zebiak and Cane, 1987). These models showed the strong role of ocean memory – mediated by equatorial wave propagation – in guiding the evolving climate of the tropics (Neelin, 1998).

The early GCMs, however, proved unable to maintain a realistic mean background state or annual cycle, in either the ocean or atmosphere. The intense feedbacks of the tropical climate meant that small errors in modeling either fluid could quickly “run away” and required assimilated observations to control the background properties. The ad hoc parameterizations of processes that mixed and distributed surface fluxes into the interior ocean continued to require subsurface ocean observations to correct. The sense in the community that great progress was at hand if these needs could be met drove the unprecedented development of the TAO array in the late 1980s (Hayes et al., 1991; McPhaden et al., 1998 and 2010). The success of TAO in turn was a proof of concept for the creation of other basin-scale, multinational arrays (e.g., Argo).

ENSO has continued to propel much of the observational, modeling and theoretical progress of the past 30 years, but despite three decades of focused attention of the climate community, ENSO forecasting skill remains stubbornly slow to improve after the initial advances. In fact, while forecast skill for the full set of retrospective ENSO events has increased slightly, forecast skill for recent events has been lower than for the events prior to the turn of the century (Barnston et al., 2012). Coupled GCMs with far better resolution and more developed physical parameterizations than those of the 1980s have seen improvements in ENSO simulations (Wittenberg et al., 2006; Delworth et al., 2012), but have produced only modest progress in forecast skill (Davey et al., 2002; Turner et al., 2005). Strong hints of decadal or longer modulation of ENSO characteristics complicate the prediction problem, as have the emergence of what might be other “modes” of ENSO. These may be related to changes in the background state that alter the relation between upper ocean heat content and SST (McPhaden, 2012), or they may simply emerge at random from the stochastically-forced and/or chaotic ENSO system

(Wittenberg, 2009; Newman et al., 2011ab; Stevenson et al., 2012; Wittenberg et al., 2014).

The overwhelming lesson of the past three decades of ENSO observation is its diversity, the ongoing succession of surprises in the expression of these events. The potential for future surprises in ENSO behavior is high. It appears that the relatively easy issues of wave-carried ocean memory are largely solved, but further advancement will entail diagnosing and explaining the mechanisms of air-sea heat and momentum exchange, in both fluids, which is unlikely to be accomplished by model experiments alone. This situation demands continued attention by the sustained observing system to the physical processes involved in tropical ocean-atmosphere interaction.

As we look to the next decade and beyond, we can’t predict the next surprises but we can expect that surprises will occur, and we must build a robust observing system that will be ready to detect and diagnose them. We take the point of view that the best way to prepare for new surprises is to observe the underlying physical processes, and thereby to teach models to represent these processes.