AMIP II Diagnostic Subproject 35

Seasonal-to-decadal variability in the tropical Atlantic Ocean

Project coordinators:
James Carton, Sumant Nigam and Jiande Wang

Department of Meteorology, University of Maryland, College Park, MD 20742


Data Requirements


Recent studies have presented a confusing array of interpretations of the existence and causes of climate variability in the tropical Atlantic sector (e.g. Hastenrath and Druyan, 1993; Enfield and Mayer, 1997; Chang et al., 1997; Tourre, et al., 1998; Mehta, 1998; Rajagopalan et al., 1998; Xie and Tanimoto, 1998; Delworth and Mehta, 1998).  We have examined these issues ourselves through a diagnostic study of historical atmospheric and oceanic reanalysis data sets as well as COADS (Ruiz-Barradas, Carton, and Nigam, 1999) covering the period 1958-1993. In this study we carry out a combined rotated principal component analysis of atmospheric and oceanic variables in order to identify covarying phenomena.  We conclude that in addition to externally forced variability, associated with the tropical Pacific and North Atlantic, there are indeed at least two modes of variability intrinsic to the tropical Atlantic.  We refer to these as the Atlantic Nino and the interhemispheric modes.

The underlying causes of these two modes seem to be different and lead us to the following scenarios.  In the Atlantic Nino mode, a relaxation of the equatorial trade winds is associated with a deepening of the equatorial thermocline.  In the boreal summer months of June through August the presence of a deep thermocline leads to unusually warm surface temperatures.  In the interhemispheric mode surface temperature anomalies in the Northern Hemisphere lead to cyclonic wind anomalies, enhanced cross-equatorial winds, and a general relaxation of the Northeast trade winds.  The relaxation of the trade winds reduces latent heat loss, leading to an amplification of the original anomaly.  This effect is most prominent during boreal spring when the Intertropical Convergence Zone is at its most southern position.

These scenarios are consistent with aspects of previous studies, but are still quite incomplete.  We would like to understand the sensitivity of these modes to a variety of parameters such as atmospheric heating distributions, etc.  Finally, we want to know more about the reasonableness of approximations that might be used in constructing simplified models.


This is a proposal to carry out a statistical examination of surface winds and heating in the tropical Atlantic sector of the available AMIPII GCMs to examine the causes of interannual to decadal variability.  Improved understanding of how the tropical atmosphere responds to SST variations in the tropical Atlantic has become a critical issue for the CLIVAR-Atlantic program.  Much of this understanding can only come from modeling studies.  Here we propose to examine the available AMIPII runs to see the extent to which their variability is consistent with the historical record, and to learn what we can about the physics of climate variability in the tropical Atlantic.  This effort will represent part of a long-term effort by the
PIs to address climate variability in the tropical Atlantic.


  • Methodology Our methodology will follow that of Ruiz-Barradas, Carton, and Nigam (1999).  We will carry out a rotated principal component analysis of the monthly anomaly fields looking for covariability among atmospheric and oceanic variables where the anomalies are defined relative to their monthly climatology.
  • We will begin with a three variable analysis using all twelve months of SST and surface wind components.  We will also explore a five variable analysis using 500mb diabatic heating and ocean heat content (available separately).
  • We will examine the projection of the modes of low frequency variability on 800mb diabatic heating, and surface heat flux components.
  • We will examine the seasonal dependence of these relationships by carrying out separate analyses in boreal spring (March-May) and summer (June-August) when we can expect different responses based on our previous work.
  • We will use our previous work as a baseline calculation and look at the projection of our principal component time series determined from
  • the reanalysis data sets on each of the model runs.
  • Finally, we will specifically examine 'extreme events' as determined from the reanalysis data sets.
Validation As indicated in the background section, we have already carried out a similar analysis of the NCAR and ECMWF reanalysis data sets, and of the COADS surface observation data set.  We will determine the extent to which we see similar behavior, spatially and temporally, in the AMIPII runs.  But, we will also be interested to explore the sensitivity of the models by looking for differences among them.  We will attempt to interpret the differences based on what is known about the differences in the physics and dynamics of the models.

As indicated above, we would also like to address some basic physics questions, among them: The patterns of diabatic heating.  The relationship between SST anomalies in the tropical Atlantic and diabatic heating anomalies.  The relationship between heating anomalies over the surrounding continental areas and over the oceanic sector.  The relationship between pressure gradient forces and winds

Data requirements


This study will require the following six monthly averaged fields for the full period, January, 1979- December, 1995 for the available AMIPII runs (a total of 19 as of July, 1999).  The variables represent a subset of those examine by Ruiz-Barradas et al. (1999) and Chung, Carton and Nigam (1999, manuscript in preparation) and are chosen to reflect the
driving function for the ocean (winds, surface heating) and measures of the tropical boundary layer (surface air pressure) and midtroposphere (500mb heating).

  • 10m winds (zonal and meridional)
  • surface air pressure
  • temperature tendency due to total diabatic heating, 500mb
  • surface latent heat flux
  • surface incident short wave radiation

We only require this information for the Atlantic sector (90W-10E, 30S-60N).  However, it may be more convenient to obtain the global data sets.


Chang, P., L. Ji and H. Li, 1997: A decadal climate variation in the tropical Atlantic Ocean from thermodynamic air-sea interactions. Nature, 385, 516-518.
Delworth, T. L. and V. M. Mehta, 1998: Simulated interannual to decadal variability in the tropical and sub-tropical North Atlantic. Geophys. Res. Lett., 25, 2825-2828.
Enfield, D. B. and D. A. Mayer, 1997: Tropical Atlantic SST variability and its relation to El Nono-Southern Oscillation. J. Geophys. Res., 102, 929-945.
Hastenrath, S. and L. Druyan, 1993: Circulation anomaly mechanisms in the tropical Atlantic sector during the Northeast Brazil rainy season.
Results from the GISS general circulation model. J. Geophys. Res., 98, 14917-14923.
Mehta, V. M., 1998: Variability of the tropical ocean surface temperatures at decadalmultidecadal time sacles, part I: the Atlantic ocean. J. Climate, 9,  1750-1771.
Rajagopalan, B., Y. Kushnir and Y. M. Tourre, 1998: Observed midlatitude and tropical Atlantic climate variability. Geophys. Res. Lett., in press.
Ruiz-Barradas, A., J.A. Carton, and S. Nigam, 1999: Structure of interannual-to-decadal climate variability in the tropical Atlantic sector, J. Clim., accepted.
Tourre, Y. M., B. Rajagopalan and Y. Kushnir, 1998: Dominant patterns of climate variability in the Atlantic ocean region during the last 136 years. J. Climate, in press.