Carlos A. de S. França & Afranio R. de Mesquita
Institute of Oceanography
University of  São Paulo, Brazil

1. Introduction

The interannual signal of the tropical Atlantic Sea Surface Temperature (SST) measurements (Servain, 1998) presents two principal scales of variability: decadal, with a north-south dipole pattern, and an interannual 2 to 5 years east-west equatorially concentrated ``Warm Event'' pattern (Carton et al., 1996; Servain et al., 1998).
The TOPEX/POSEIDON (T/P) altimetry data with its unprecedent accuracy and measurement period allows inferences on the interannual variability of sea level over the open ocean.  More than 5 years of T/P altimetry data are analysed in the tropical Atlantic ocean (15oN-15oS; 060oW-020oE) through Empirical Orthogonal Functions (EOF) of the  Sea Level Anomalies (SLA).

2. Data

SLA for the period October/92 to March/98 were obtained from MGDRs-C T/P altimetry (AVISO, 1996a,b). Standard corrections were applied to the data and summarised in Table 1. Ellipsoidal sea level were resampled to a 30 km alongtrack intervals and time series were formed. Elastic Ocean Tide were removed by harmonic filtering 11 tidal constituents (Table 1) and SLA were estimated relative to October/92 to October/96 (4 years) mean value of the ellipsoidal sea level.  Uncertainty of SLA is 4 to 5 cm for each individual estimate (Fu et al., 1994). Alongtrack SLA  were mapped into a regular  1o ? 1o grid at 10.14 days intervals through Optimal Linear Interpolation (Liebelt, 1967), using decorrelation scales of 150 km for space and 15 days for time. Error estimates for the maps from Optimal Linear Interpolation (OLI) are between 1 to 2 cm in deep ocean, but may reach more than 5 cm over the shelf and coastal borders.
The mapped time sequences were low-pass filtered to remove seasonal and shorter periods variability by a running-mean filter with 37 time intervals length (375 days). Filtered maps were submitted to Empirical Orthogonal Function (EOF) analysis (Preisendorfer, 1988).

3. Results

The two most significant EOF explain altogether about 94% of the SLA variability of the region (70% for EOF-1 and 24% for EOF-2). Spatial structures of these EOFs are presented in the Figure 1. Their correspondent Principal Components (or Time History) are shown in Figure 2. EOF-1 presents some structure but it is mainly positive in the region; only a zonal stripe around  6-7o N presents negative values. The Time History of EOF-1 shows an oscillation with a period of 2 to 4 years and an apparent positive trend.  EOF-2 presents a variability concentrated around the equator with opposing phases between Western and Eastern Atlantic. EOF-2 Time History also shows an oscillation with a 2 to 4 years period, but clearly presents an anomaly starting at the end of 1997 with a shorter time scale (1-2 years).
SLA mean for the region ( 15o N-15o S; 060o W-020o W ) and its low-pass (375 days) filtered values are displayed in the Figure 3. Coherence between the filtered mean SLA for the area and the EOF-1 Time History  is apparent. The trend estimated for the filtered and unfiltered (longer time series) SLA mean in the area is 0.39 ? 0.06 cm/year  and 0.30 ? 0.06 cm/year respectively. Variability with time scales of 100 to 150 days present in the SLA unfiltered mean is an important contributor to the variance of this time series.
Ionosphere NRA - 2 frequencies DORIS
Dry Troposphere ECMWF Model  ECMWF Model
Wet Troposphere TMR TMR
Sea State BM4 BM4
Pole Tide Wahr Wahr
Solid Earth Tide CT&CE CT&CE
Inverted Barometer ECMWF Model ECMWF Model
Elastic Ocean Tide 11 const. Filter 11 const. Filter
Table 1. Geophysical corrections applied to the T/P altimetric sea level data (AVISO, 1996a,b).

CNES ELFE/JGM3 - Centre Nationele d'Etudes Spatiales (CNES) orbit estimation
NRA - NASA (National Aeronautics and Space Administration)
Radar Altimeter based correction
DORIS - Doppler Orbitography Integrated by Satellite based correction
ECMWF - European Centre for Medium Range Weather Forecast model for atmospheric pressure
TMR -  TOPEX Microwave Radiometer based correction
BM4 -  Polynomial model for sea state bias
Wahr - Wahr pole motion model.
CT & CE - Cartwright & Tayler and Cartwright & Edden tidal potential model.
11 const. filter - Mm ,  Mf ,  Q1 ,  O1 ,  P1 ,  K1 ,  N2 ,  ?2 ,  M2 ,  S2 ,  K2 .

Fig.1. Spatial structures for the two most significant EOF of the low pass filtered T/P SLA in the tropical Atlantic ocean for the period October/92 to March/98. (A) EOF-1. (B) EOF-2.}

Fig. 2. Time History of the two most significant EOFs in the region  060o W  -  020o E ,  15o N - 15o S , for the period October/92 to March/98.

Fig. 3. Mean value and low pass filtered mean value of the T/P SLA in the tropical Atlantic ocean area for the period October/92 to March/98.}

4. Discussion

The anomaly occurring at the end of 1997 observed in the second EOF of the T/P SLA in the tropical Atlantic ocean seems to fit to a “Warm Event” (Carton & Huang, 1994), similar to “El Niño” phenomena occurring in the tropical Pacific ocean.  During boreal winter 1996-1997 easterly zonal winds were anomalously strong leading to an increase in the sea level (and heat content) in the Western sector of the equatorial Atlantic.  In the boreal spring-summer 1997 zonal easterly winds relaxed and sea level have readjusted transferring heat zonally towards the eastern section making  next fall-winter 1997 season anomalously warm in the Gulf of Guinea with weak easterly winds, characterising an Atlantic “El Niño”. Differently from others warm events reported (Arnault & Cheney, 1994), the present one seems to occur simultaneously to an “El Niño” event in the Pacific (e.g. NOAA, 1998), running against the hypothesis that the equatorial Atlantic response is delayed by one year relative to Pacific events. Although the 5.5 years period is not long enough to evaluate the  decadal variability, some indication of it should be present. A 10 years dipole pattern, however, is not observed in the T/P SLA in the present analysis. In the whole tropical Atlantic region considered an oscillating (2 to 4 years) positive trend is observed instead (Fig. 3).The trend estimated in the SLA for the region, however,  have to be seen with caution. It can be linked to variability of time scales not resolved by the current T/P data and even linked to drifts in the altimeter measurement system (Nerem et al., 1997).

5. Conclusions

The analysis of the interannual variability of SLA from T/P altimetry for the period October/1992 to March/1998 in the tropical Atlantic ocean ( 15o N-15o S; 060o W-020o E ) through EOF analysis has revealed an occurrence of an “Warm Event” in the Gulf of Guinea in 1997, simultaneously to a Pacific warm event “El Niño”. The North-South dipole pattern with decadal scales signature does not seem to be present in the current analysis of SLA.A sea level trend of 0.30 ? 0.06 cm/y was estimated for the considered region for the period October/1992 to March/1998 from T/P altimetry.

6. Acknowledgements

We are thankful to the French space agency CNES (Centre National d'Etudes Spatiales) through the AVISO (Arquivage, Validation et  Interprétation des données des Satellites Oceanographiques) for making available for analysis the TOPEX/POSEIDON data.

7. References

Arnault, S. &  Cheney, R. E.1994. Tropical Atlantic sea level variability from Geosat (1985-1989). J. Geophys. Res., 99(C9):18207-18223.

AVISO. 1996a. User Handbook for Merged TOPEX/POSEIDON Products. AVI-NT-02-101-CN, Edition 3.0 .

AVISO. 1996b. User Handbook: TOPEX/POSEIDON Update Product. AVI-NT-02-102-CN, Edition 1.0.).
Carton, J. A. &  Huang. B. (1994). Warm events in the tropical Atlantic. J. Phys. Oceanogr., 24: 888-903.

Carton, J. A.; Cao, X.; Giese, B. S. &  Silva, A. da 1996. Decadal and interannual SST variability in the Tropical Atlantic Ocean. J. Phys. Oceanogr.,  26: 1165-1175.

Fu, L.-L.; Christensen, E. J.; Yamarone, C.;  Lefebvre, M.; Menard, Y.; Dorrer, M. & Escudier, P.. 1994. TOPEX/POSEIDON Mission Overview. J. Geophys. Res. 99(C12): 24369-24381.

Liebelt, P. B. 1967. An introduction to optimal estimation. Addison Wesley, Reading, MA. 273p.

Nerem, R. S.; Haines, B. J.; Hendricks, J.; Minster; J. F.; Mitchum, G. T. & White, W. B. 1997. Improved determination of global mean sea level variations using  TOPEX/POSEIDON altimeter data. J. Geophys. Res. , 24(11): 1331-1334.

NOAA, 1998. National Oceanic and Atmophere Administration Laboratory for Satellite Altimetry. http://ibis.grdl.noaa.gov/SAT.

Preisendorfer, R. W. 1988. Principal Component Analysis in Meteorology and Oceanography. Developments in Atmospheric Science, 17. Curtis D. Mobley Editor.Elsevier, New York. 425p.

Servain, J.  1998. SST and PWS 1964-1997 over Tropical Atlantic. Research-Quality Product (1964-97). internet: http://www.coaps.fsu.edu/WOCE/SAC/atlantic/.

Servain, J.;  Wainer, I. & Dessier, A. 1998. Évidence d'une liaison entre les deux principaux modes devariabilité climatique interannuelle de l'Atlantique tropical. C.R.Acad.Sci.Paris, Sciences de la terre et desplanètes, 327: 1-8.