SCIENCE REPORTS

Other TIP pages

TIP homepage

Other TIP 5 pages

TIP 5 Table of contents

Summary

Opening of the meeting

Summary of current conditions in the Tropical Pacific

National reports

Program status reports

Science reports

Recommendations

Acknowledgments

Surface and Subsurface Low-Frequency Variability in the Tropical Atlantic: One Necessity for Implementing the PIRATA Array
J. Servain, ORSTOM/Brest and I. Wainer, IOUSP/Brazil
Though some aspects of the interannual variability in the upper tropical Atlantic Ocean were previously tackled, we do not have any conclusive idea about the time-scale of such variability. Our attention was to focus on the monthly deviations from the seasonal cycle of the vertical displacement of the 20°C isotherm (Z20), which is commonly associated in the tropical regions with the vertical displacement of the thermocline depth. Until now, the TOGA-WOCE XBT data set remains the only candidate to provide such type of information, even if opportunities to construct Z20 long-term series in the tropical Atlantic from XBT probes are exceedingly limited. Our alternative was to look at the Z20 long-term series simulated by an OGCM, after testing primarily the accuracy of the model results vs. the observed data fields, where and when that was possible. In the present experiment, the OGCM OPA7 developed at LODYC is forced during 1985 - 1994 by the outputs (momentum, heat and fresh water fluxes) from the AGCM ARPEGE developed by Meteo-France. Only the oceanic results between 20°N and 20°S can be used in the validation analyses because OPA7 is partially relaxed to the climatology (temperature and salinity) outside the subtropical latitudes.

A first series of tests were made to look at surface variability, especially SST anomalies. The OGCM reproduces pretty well the observed interannual variability of SST in the whole tropical Atlantic, either at shorter time-scales (a few months), or at longer scale-times (many years). That is an important improvement in comparison with what was noted in previous experiments. Furthermore, it was determined that the model is able to reproduce properly low-frequency (decadal time-scale) interhemispheric variability in the surface layer which is evident from SST observations (the "SST dipole mode").

In this investigation a series of observing system simulation experiments (OSSE's) is performed as part of a design study for the proposed PIRATA array of instrumented moorings in the tropical Atlantic. In most OSSE's, a model is used to construct synthetic fields of ocean data, which are then subsampled according to the observing pattern being simulated. The challenge of these experiments is the reconstruction of the reference field from the restricted data set. Such experiments in which the same model is used for construction of the synthetic data set and for its reconstruction for simulated observations are often referred to as "identical twin" experiments. In contrast, here we subsample fields of sea surface height anomalies observed by the TOPEX/Poseidon satellite altimeter, and attempt to reconstruct the full TOPEX/Poseidon height anomaly field through the use of assimilation of data at a restricted number of locations.

A modified Kalman filter is used to assimilate surface height anomaly data into a simple ocean model of the tropical Atlantic. In the past, limitations of computer resources have kept researchers from using the full Kalman filter for assimilation of data into ocean models. Our approach to implementing the Kalman filter eliminates the large memory requirement since the forecast error covariance is calculated and maintained on a grid which is much coarser than that used in the model. This is acceptable since the primary interest is in the large scale characteristics of the error.

With the reduced-space Kalman filter, several data assimilation runs are performed in order to optimize the location of a limited number of moorings for the PIRATA deployment. The OSSE begins by establishing bounds for the suite of experiments. The lower bound is an experiment without assimilation which demonstrates the ocean model errors. Another experiment, in which data are assimilated in an unrealistically dense array configuration, serves as an upper bound to any proposed pilot mooring array. A suite of experiments was then conducted by assimilating altimeter data at three mooring locations along a meridian at 2°N, 2°S, and the Equator, while the longitude was varied sequentially by 5° for each run. Results of these experiments show the biggest impact of the assimilated data occurs when the observations are taken between 15°W and 30°W. For example, assimilation of sea level observations at 20°W shows that a limited number of assimilation sites can improve the temporal signal throughout the equatorial waveguide, however, the amplitude of the sea level signal is poorly reconstructed.

Next, a more theoretical approach is used to determine optimal mooring locations. Optimal points are determined in a more objective fashion using a Monte-Carlo sampling of the model forecast error structure. Using this technique, the optimal mooring locations were found to be along the Equator at 35°W, 20°W, and 10°W. Assimilation at these three locations shows that not only the temporal signal, but also the amplitude is transmitted throughout the equatorial waveguide.

These OSSE results were used to guide the decision for the initial deployment of three equatorial moorings during the first year of PIRATA. In subsequent years, the PIRATA plan calls for additional moorings (approximately 14 in total) to be deployed to enhance the sampling within the equatorial waveguide and off the equator in the vicinity of the dipole SST variability maxima. Additional experiments were performed to demonstrate the efficacy of the full PIRATA deployment and the added value that can be expected from PIRATA observations above and beyond existing XBT observations.

On the Physics of ENSO Displacement of the Pacific Warm Pool
J. Picaut, ORSTOM/Noumea
The Pacific warm pool is peculiar with a relatively deep upper layer of well-mixed warm and low-salinity waters which cover one-third of the equatorial Pacific. Furthermore, this pool is subject to strong interannual migration along the equator, in phase with the Southern Oscillation Index. In the scientific plan for TOGA-COARE, it has been suggested that the Warm Pool may be associated with the zonal convergence of the oceanic circulation in the western tropical Pacific, and that low surface salinity contributes to its insulation from colder and saltier water below. In the present study, with the use of four datasets and three classes of ocean model, we demonstrate the dominance of zonal advection in the zonal migration of the eastern edge of the warm pool. This is evidenced through the discovery of a zonal convergence of water masses and a well-defined salinity front at the eastern edge of the warm pool, which together move along the equator in phase with the Southern Oscillation.

Most data presented in this study was collected during the 1985 - 94 TOGA decade. They consist mainly of surface current fields collected from four equatorial TAO moorings and drifters, as well as currents derived from satellite altimetry measurements (GEOSAT and TOPEX/Poseidon). Three different model outputs were used: the Cane and Patton linear multimode model over the 1961 - 94 period, the Gent and Cane OGCM and the three tropical oceans LODYC high resolution OGCM over the 1982 - 94 period. Following previous studies suggesting the importance of zonal advection in the zonal migration of the warm pool, all surface-current fields were used to compute the zonal displacements of hypothetical drifters transported by the 4°N 4°S averaged current. Such hypothetical drifters, launched on the eastern edge of the warm pool and moved by the observed and simulated surface currents, remain close to this edge all the way to the end of the studied period. This stresses the dominance of zonal advection in the zonal displacement of the eastern edge of the warm pool. The trajectories of different hypothetical drifters launched in each current field a little west or east of the eastern edge of the warm pool converge into a single trajectory after 2 - 3 years. The 4°N 4°S drifter trajectories integrating the eastward and westward surface currents in the equatorial band, this provides evidence for the zonal convergence of water masses towards the eastern edge of the warm pool. Given the presence of rainfall-induced fresh water in the warm pool and saltier water to the east, the zonal convergence of water masses within the equatorial band results in a well-defined salinity front and therefore in a density front, as indicated by salinity observations and LODYC model simulation. Because the salinity front is induced by current convergence, it is situated in a region of very weak or null zonal currents, which together with the corresponding density front restrain the heat exchange between the warm pool and the equatorial region further east. The convergence of surface currents and water masses not only results in the formation of the salinity front but also in the subduction of salty water from the central equatorial Pacific below the fresh water of the western Pacific. This creates the underlying barrier layer in the equatorial band which obstructs entrainment of the colder and saltier water into the surface layer. The warm pool, which is composed of low-density fresh and warm water, floats above the high-density cold and salty water; therefore it can be easily displaced zonally by wind-driven currents, as the momentum is trapped in a shallow mixed layer. All of this explains why the warm pool is somewhat isolated from the remaining equatorial Pacific, and why there is such a good agreement between the ENSO displacements of the eastern edge of the warm pool, of the zonal convergence of water masses, and of the zonal salinity front.

Events in which strong eastward current exceeding 0.5 m s covers a surface layer above about 100 m were observed at the eastern station (147°E) in May, September, and December 1994. Timings of similar events at the western station (142°E) are different among the events. The event in May occurred at the western station prior to that at the eastern one suggesting eastward propagation of a disturbance, while for the event in December the western station lagged the eastern one. In September, no event was observed at the western station. Surface wind data based on the ECMWF Basic level III were used to examine a forcing mechanism of the events. Enhancements (bursts) of the westerly winds in the western Pacific on the equator occurred all in these events. Areas where the westerly wind burst (WWB) are dominant are found at different longitudes among the events. Thus, it seems that the timing for the occurrence of the surface eastward jet at the equatorial stations is related to the dominant areas of the WWB.

A signal with the period of about two days was often observed in the hourly time series of the zonal and meridional currents at both stations. In the time series of surface wind obtained from the ATLAS buoy at 0°,147°E, no similar signals are detected, which gives no evidence of surface forcing for the oceanic signal. The quasi-two-day signal has a tendency of being active at the surface layer above about 150 m, and little phase difference in the vertical direction. In some periods, the signal is confined at the subsurface layer between 50 and 150 m, and its large vertical phase shift was found at a layer in which time-averaged zonal current has a large vertical shear. It is suggested that the occurrence of the signal may be caused by shear instability of the mean currents.

Observational Study on the New Guinea Coastal Current and Its Undercurrent
Y. Kuroda, JAMSTEC
One-year time series current data from subsurface ADCP moorings were examined to detect the variability of the New Guinea Coastal Current (NGCC) and New Guinea Coastal Undercurrent (NGCUC). These moorings were deployed at 2°S,142°E and 2.5°S,142°E in the core region of these currents from July 1995 to July 1996. Knowing the characteristics of the NGCC and NGCUC are important for understanding mechanisms that maintain the warm pool in the western tropical Pacific because they are major inflows, particularly NGCUC carrying high salinity water from southern hemisphere to northern hemisphere. Seasonal reversal of the NGCC was clearly observed. Flow in the surface layer was dominant to the northwest along the direction of the New Guinea coast at maximum speeds of 80 - 100 cm s-1 down to 100 m depth in the boreal winter; conversely in summer, it flowed southeastward at maximum speeds of 50 - 60 cm s. The time change of zonal volume transport in the upper 50 m layer between 2.5°S and 2°S at 142°E in the NGCC was coincided with the time change of local zonal wind speeds. The NGCUC below the surface layer flowed northwestward all the year. From the annual mean profile at 2.5°S,142°E , the core speed of NGCUC was 50 cm s at 230 m depth with standard deviation of 15 cm s. In the NGCUC core layer of 150 - 250 m depth, 20 - 30 day perturbations were evident all year. The westward volume transport across the 142°E line between 2.5°S and 2°S in the 0 - 200 m layer increased up to 6.5 Sv in boreal summer and decreased to zero in boreal winter.

Large Scale Sea Surface Height Variations of the India Ocean from TOPEX/Poseidon Altimeter
P. Kumar, NIO
TOPEX/Poseidon altimeter data, during November 1992 to August 1995, pertaining to the Indian Ocean region bounded by longitudes 40°E to 100°E and latitudes 10°S to 30°N were extracted from merged geophysical data records (MGDR) available on AVISO CD-ROM. Apart from applying standard corrections, a gradient reduction correction was also applied to minimize the errors in the sea surface height (SSH) estimation due to cross track variations in sampling positions. This was found to be most effective in the region where the Ninety East Ridge is located and the topography presents large variability. Finally, the ten-day snap shot of SSH anomaly was obtained from the gradient reduced collocated data by subtracting the mean.

The time-longitude plots of SSH reveal the signatures of Kelvin wave propagation along the equator: one during mid-February and the other during mid-August, both associated with the collapse of the monsoonal winds. The computed speed is about 1.25 m s. This Kelvin wave hits the eastern boundary in November, as evident by the high SSH in the ten-day snapshots. On hitting the eastern boundary, the high sea level propagates northwards along the eastern boundary and southward along the east coast of India (coastally trapped Kelvin wave). The northward coastal current along the west coast of India is seen developing during November and becomes fully developed by January. The offshore movement of high sea level seen in mid-January, off the southern tip of India is the westward propagation of the downwelling Rossby wave radiated by the coastally trapped Kelvin wave. During this period the circulation in the interior Arabian Sea is zonal and westward and in the Bay of Bengal it is cyclonic. By July, the Somali current is fully developed and the formation of Great Whirl is seen in all the three years. Signatures of upwelling is inferred along the coast of Somalia, Arabia, and the south west coast of India. The upwelling front along the west coast of India, where the SSH shallows markedly, appears to propagate westward.

Along 5°N, 10°N, and 15°N the time-longitude plots show dominant westward propagating Rossby waves, the speed of which decreases from 21.98 cm s at 5°N to about 7.5 cm s at 15°N. The westward propagation at 10°N in the Bay of Bengal starts during January, while in the Arabian Sea it starts from the eastern boundary (west coast of India) during mid-May. The two-dimensional fast Fourier transform along 15°N shows a dominant westward propagation in the Bay of Bengal, while in the Arabian Sea there are signatures of both westward as well as eastward propagation. The eastward propagation at the western boundary in the Arabian Sea is associated with the offshore advection of the Somali current and upwelling.

Dynamics of the Lakshadweep High and Low
S. Shetye, NIO
A "high" in sea surface topography forms off southwest India, in the vicinity of the Lakshadweep Islands, during the northeast monsoon, and a "low" forms during the southwest monsoon. The high and the low propagate westward, extending across the southern Arabian Sea a few months after formation. Our studies on the dynamics of the high and the low suggest that, unlike eddies often observed in the world oceans, the high and the low do not owe their existence to nonlinearity. They are a consequence of westward propagating Rossby waves radiated by Kelvin waves propagating poleward along the western margin of the Indian subcontinent. The formation of the high/low provides a mechanism for the early onset of upwelling off southwest India, and hence for formation of SST gradient before onset of the southwest monsoon.

Upper-Ocean Circulation in the Indian Ocean, and Possible Feedbacks to the Atmosphere
J. McCreary, NOVA University
What mechanisms of ocean-atmosphere interaction might cause climate variations in the Indian Ocean and elsewhere? More specifically, what processes might cause interannual variability of sea-surface temperature (SST) and surface heat flux (Q)? Three categories of possible mechanisms are: ocean dynamics, mixed-layer thickness, and variability in the Indonesian Throughflow.

Ocean dynamics affect SST and Q most strongly by their influence on upwelling. In contrast to the Pacific, the primary upwelling regions in the Indian Ocean are located in the northern basin; moreover, the meridional circulation cells that carry cool water to the upwelling regions are estimated by models to be 4-5 times weaker than they are in the Pacific (5 Sv as compared to 22 Sv). Thus, mechanisms of ocean-atmosphere interaction thought to be important in ENSO are not likely to be active in the Indian Ocean. On the other hand, due to Ekman suction the thermocline is shallow in a band that extends across the eastern and central Indian Ocean from about 5°S to 15°S. In several Indian Ocean models, the thermocline surfaces in this band thereby cooling SST, but there is little indication in climatological SST data that this surfacing actually occurs. Nevertheless, interannual variability in the strength of Ekman suction, could allow the thermocline to surface there, and so provide a dynamical mechanism for altering SST and Q significantly.

Mixed-layer thickness affects SST and Q through its influence on the ocean's ability to absorb heat. In the southern Indian Ocean, the annual cycle of mixed-layer thickness is large in locations where the Southwest Monsoon gains moisture; hence, interannual variability that affects this cycle may impact monsoon rainfall. In the eastern equatorial Indian Ocean and Bay of Bengal, the mixed-layer thickness is strongly influenced by fresh-water flux, and during the winter, barrier layers with temperature inversions are prominent, just as they are in the western Pacific warm pool. It may be that mixed-layer processes in these regions are involved in the generation of the both ENSO and the biennial oscillation.

The Indonesian Throughflow carries warm, fresh water into the Indian Ocean. Its interannual variability is dominated by the ENSO signal. It is likely that this variability is simply a reaction to ENSO, but one model suggests that Throughflow variability may feedback to affect SST and Q in the eastern Pacific.

The interannual variability is reproduced well by the model as seen by comparisons to data such as NMC Reanalysis and TOPEX/Poseidon. The SST anomalies in the Arabian Sea and the Bay of Bengal are less than a degree C and are mainly determined by radiation anomalies associated with the cloudiness and convection anomalies. The thermocline close to the surface in the western part of STIO mentioned above results in the largest interannual anomalies in SST (~1øC) and sea-level (~10 cm). The thermocline in the eastern Indian Ocean at the same latitudes is deep and typically has anomalies of opposite sign in both SST and sea-level. A perturbation to the pressure center in the Indonesian region on a quasi-biennial time-scale (associated with the SO) leads to basin scale wind anomalies between 5°S and 15°S which appear to be modulations of the annual cycle. These wind-anomalies produce anomalies of opposite sign in SST and sea-level in the western and eastern STIO in this latitude-band due to the sloping thermocline. Analysis of heat budgets shows that upwelling anomalies in the west (where the thermocline is close to the surface) and advection anomalies in the east determine the SST anomalies. Thus a quasi-biennial oscillation involving the thermocline tilt exists in the STIO. This oscillation may play an important role in the Indian monsoon and also the heat balance of the Indo-Pacific warm pool and is being investigated further.

Uncertainties of ERS-1 Surface Winds Over the Arabian Sea
D. Halpern, JPL
Spatial variations of the east-west and north-south components of surface wind stress are critical in studies of ocean circulation and biological-physical interactions because surface wind stress curl produces a vertical velocity (We) in the upper ocean at the bottom of the Ekman layer. The wind forced We acts as a "pump," causing water to upwell and sink. The ERS-1 scatterometer provides reasonable coverage and unique direct measurements of vector of winds. However, empirical schemes must be used to generate wind velocity from ERS-1 radar measurements. Three schemes (ESA, IFREMER, and JPL) are evaluated relative to high-quality moored-buoy wind observations recorded in the central Arabian Sea, where high surface waves and high atmospheric water content during the southeast monsoon adversely affect the estimation of satellite-derived winds. The sensitivities of Weand other wind-driven characteristics of ocean circulation (such as Ekman transport and Sverdrup transport) to different ERS-1 surface wind velocity data products are discussed.

Atmospheric Forcing and Upper Ocean Response to Monsoonal Forcing in the Arabian Sea
R. Weller, WHOI
The atmospheric forcing and variability of the upper ocean observed for one year at a site in the Arabian Sea are described. Meteorological observations and the air-sea fluxes of momentum, heat, and freshwater computed from them are compared to climatologies and to the output of numerical weather prediction models. Temperature, salinity, and velocity variability in the upper 300 m of the ocean are described, and the ability of a one-dimensional ocean model forced with the observed fluxes to predict the observed variability is investigated.

From October 1994 to October 1995, a surface mooring was maintained at 15°N, 61°E, some 500 km off the coast of Oman. The mooring line carried oceanographic instruments from Lamont- Doherty Earth Observatory (Dr. John Marra) and the University of California, Santa Barbara (Dr. Tom Dickey) as well as those deployed by the Woods Hole Oceanographic Institution. The mooring was recovered and redeployed with fresh instrumentation in April 1995.

The winds of the northeast monsoon were moderate, typically 6 m s, dropped to 3 to 4 m s in the intermonsoon, and then increased to monthly averages of between 9 and 13 m s in the southwest monsoon, when they were particularly steady in direction and peaked in July when the highest daily average reached 15.7 m s. The air during the southwest monsoon was more humid than during the winter, and in July and August sea surface temperatures were cooler than air temperatures. As a result, during the southwest monsoon sensible heat flux was positive (ocean heat gain) and latent heat loss reduced, so that in contrast with the northeast monsoon (when monthly net heat fluxes were for two months an average of -45 W m), the ocean typically gained heat (50 to 100 W m monthly averages). The observed summer heat gain contrasts with the Hastenrath climatology (Hastenrath and Lamb, 1979 a,b) that suggests significant heat loss to the atmosphere during both the northeast and southwest monsoons but is close to the present COADS climatology (da Silva et al., 1994) that also shows heat gain by the ocean through the southwest monsoon. Observed meteorological variability and air-sea fluxes were also compared with those fields from numerical weather prediction models. Net heat fluxes from NCEP were typically 100 W m less (and thus negative during the southwest monsoon) than observed, due largely to greater latent heat loss and less net shortwave radiation than observed. ECMWF net heat fluxes were also lower than observed (though not negative during the southwest monsoon) but the source of the difference varied. In the northeast monsoon, ECMWF had less than observed shortwave radiation and larger negative sensible heat loss; in the southwest monsoon, ECMWF, the net shortwave was in good agreement, and the difference came from the model's more negative latent heat flux.

The mixed layer depth (defined as the depth that was 0.1°C cooler than the surface temperature) was shallow initially, in October 1994, with a depth of 10 to 20 m. It deepened to over 100 m in early January 1995 and cooled 1°C during the northeast monsoon. From early January it shoaled steadily and warmed until late March 1995. Then it continued to warm and remained shallow through early June, when sea surface temperature was in excess of 30°C. With the onset of the southwest monsoon, the mixed layer cooled and deepened to reach 80 m by early July 1995, warmed and shoaled to 30 to 40 m through early August, and continued to warm and shoaled to 20 m through the end of the deployment in October 1995. Some of this variability can be replicated by a one-dimensional mixed layer model (Price et al., 1986) forced with the observed fluxes and initialized with a CTD profile taken in October 1994. The predicted mixed layer deepens and cools with both the northeast and southwest monsoons. However, the predicted mixed layer depths are too shallow during both monsoons; and the model mixed layer's temperature and salinity increase beyond those observed during the southwest monsoon.

Other processes that contribute to the local heat and salt budgets are being investigated. Estimates of the magnitude of Ekman pumping based on the ECMWF winds suggest that local pumping does not play a major role in balancing surface heating and evaporation. Two effects of horizontal advection, change in the vertical structure of the temperature and salinity fields and heat and salt transport within the mixed layer appear to be more important. Large velocities, in excess of 70 cm s-1 in the mixed layer, were observed that were related to mesoscale variability in the ocean, particularly in October through December, 1994. Associated with these events is a change in water properties, with cooler temperature found below the mixed layer and intensification of the pycnocline. Experiments in which the one-dimensional model is re-initialized often throughout the year suggest that the temporal change in upper ocean stability associated with the mesoscale flows modulates the mixed layer response to local forcing. In addition, and particularly during the latter half of the southwest monsoon, the eastward flow in the upper advects in cooler water upwelled along the coast of Arabia.

References:

da Silva, A. M., C.C. Young, and S. Levitus, 1994: Atlas of surface marine data 1994. Volume 1: Algorithms and procedures. NOAA Atlas NESDIS 6, U.S. Dept. of Commerce, Washington, D. C., 83 pp.

Hastenrath, S. and P. Lamb, 1979a: Climatic Atlas of the Indian Ocean. Part 1: Surface circulation and climate. The University of Wisconsin Press, Madison, 116 pp.

Hastenrath, S. and P.J. Lamb, 1979b: Climatic Atlas of the Indian Ocean. Part 2: The oceanic heat budget. University of Wisconsin Press, Madison, 110 pages.

Price, J.F., R.A. Weller, and R. Pinkel, 1986: Diurnal cycling: Observations and models of upper ocean response to diurnal heating, cooling, and wind mixing. Journal of Geophysical Research, 91(C7), 8411 - 8427.

The sea surface temperature in the Bay of Bengal continues to rise before the occurrence of the first transition and drops quickly after. Before the transition, the atmospheric conditions in the Bay of Bengal are characterized by low surface winds, low cloud cover, and small optical thickness. Such condition favors the rising of the sea surface temperature through increasing shortwave radiation and low latent heat flux. During and after the transition, strong low-level winds and convection result in the drop of sea surface temperature by decreasing shortwave radiation and increasing latent heat flux. The ocean seems to contribute to the abrupt change of the atmospheric circulation during the transition. Before the transition, the warm sea surface in the Bay of Bengal, that is located to the north of the convective region, helps to destabilize the lower troposphere and leads to the northward shift of the convection. The large latent heat fluxes that the ocean supplies to the atmosphere during and after the transition help to sustain the convection and the large-scale circulation.

It is proposed that the interaction between large-scale atmospheric circulation, convection, and lower-boundary forcing leads to the abrupt change of the atmospheric circulation during the first transition. The ocean-atmosphere interaction occurring in the Bay of Bengal, that contributes to the lower-boundary forcing, could be one of the important factor.

TAO Project Office
NOAA | Pacific Marine Environmental Laboratory
7600 Sand Point Way NE
Seattle, WA 98115
atlasrt@noaa.gov
Credits | Disclaimer | Privacy Policy