Ekman Et Al. 1987

[Meri Thilmony]

Thepoleward transport of heat carried out by the ocean currents is a major contributor to interannual, long-term, large-scale, and, quite possibly, global climate changes. The role of winds in the global climate is of crucial importance. Winds drive the ocean gyre circulation and promote and maintain a substantial portion of the global thermal balance.

The Ekman heat flux (EHF) is the transport of heat in the first 100 m in the ocean, induced by the action of the wind. Although the wind-driven transport is restricted to the upper layers in the ocean, it has to be integrated across the entire latitudinal extent of the ocean basin. Therefore, depending on the latitude, the combined effect of strong and persistent wind patterns and high temperatures in the upper layers can produce large wind-driven heat fluxes, sometimes even comparable to the total oceanic heat flux. Kraus and Levitus (1986) using climatological data showed that about 50% of the total heat in the tropic circles of the Atlantic Ocean and almost all of the total heat in the Tropics of the Pacific Ocean is dominated by the Ekman heat flux. Sato and Rossby (2000), using historical hydrographic sections spanning almost 60 years, investigated the seasonal and low-frequency variability of the meridional heat flux at 36N in the Atlantic. They found that the phase of the annual cycle of the total heat flux at the latitude is dominated by the Ekman heat flux.

As pointed out by Montgomery (1974), the estimation of the heat flux should be done for systems whose net mass flux is zero, otherwise it becomes dependent on an arbitrarily defined reference temperature. Because the major contribution of the EHF comes from a relatively shallow layer in the upper ocean, one of the main concerns is how to satisfy the conservation of mass condition. The usual way is to determine the depth in which the lower-layer mass flux compensates the mass flux in the Ekman layer.

The EHF is the only component of the total heat balance that can have its variability directly measured and monitored because of the data availability. On a global scale, the EHF has been previously estimated using wind stress from in situ climatological data (Kraus and Levitus 1986; Levitus 1987; Adamec et al. 1993) and from satellite-derived winds (Ghirardelli et al. 1995). Ekman heat flux was also included in the estimation of the poleward heat flux at specific locations (Hall and Bryden 1982; Bryden et al. 1991; Macdonald and Wunsch 1996; Sato and Rossby 2000).

In the next section we discuss the method and the data used to estimate the temperature in the Ekman layer, the temperature of the compensating flow, and the wind stress decomposition in spectral bands. In section 3 we present the comparison of the four EHFs with emphasis on the annual and interannual variability. The correlation among the wind stresses from all sources are examined at global scale. Also in that section a covariance analysis of the frequency bands for each ocean basin is discussed, followed by the conclusions.

In Eq. (1), θ represents the mean potential temperature of the volume transported by the ocean necessary to compensate the Ekman volume flux in the upper layer. The volume transport of the Ekman layer is balanced by an equal and opposite deep flow so that the mass conservation condition is satisfied at each latitude (Montgomery 1974). Results from a high-resolution model over the Atlantic showed that this return flow should extend to at least 3000 m in the extratropical region (Bning and Herrmann 1994). According to them, the assumption of a barotropic return flow at depth to compensate the ageostrophic Ekman flux at the surface corroborates model results and the theory (Gill and Niiler 1973). For this study, the depth-averaged mean temperature of the oceans was based on climatological monthly mean profiles from 0 to 5000 m from the World Ocean Atlas 1994 (Levitus and Boyer 1994).

The wind stress (τ) is calculated from bulk parameterization and flux equations from the similarity theory (Liu et al. 1979). This calculation takes into account the sea surface temperature, water vapor, and wind. Except for the NCEP data, which have their own set of variables, we used SSM/I water vapor and Reynolds SST to estimate τ for the satellite-based data. The zonal component of the wind stress (τx) is estimated and interpolated to match the same resolution of the temperature maps (1 1 10 days).

Temporal changes in the meridional Ekman heat flux are affected by changes in both the wind stress and temperature fields. Here we investigate the relative contribution of some of the spectral components of τx to the Ekman heat flux. The net fluctuations in the EHF combine wind and temperature signals that are correlated in space and time. The effect of the temperature alone is not considered in this study: First, because we lack high-resolution temperature profiles, since only the SST is measured, and the rest of the profile is climatological. Second, most of the variability observed in the Ekman heat flux is dominated by the variance of the wind stress (Sato et al. 2002). Because of these two reasons, the quantitative assessment of the relative importance of the variability in the temperature of the Ekman layer in the EHF estimates could be the subject of a future study when higher-resolution temperature profiles become available.

The Ekman heat flux is estimated globally using the zonal component of the wind stress from four sources. An equatorial band from 4S to 4N is excluded as classical Ekman dynamics do not hold there. The Ekman heat flux estimates are compared to each other in terms of annual mean, annual cycle, and amplitude of the zonally integrated EHF; the interannual variability; and the spatial variability of some of the spectral components.

The general pattern of the annual mean of the meridional Ekman heat flux as a function of latitude is poleward between the tropics in the Atlantic, Pacific, and in the tropical south Indian Oceans for all sources (Fig. 1). The exception is the closed north Indian Ocean, where the transport is equatorward due to the monsoon wind regime. The annual mean was estimated using a complete number of years for each time series to overcome the fact that they do not have the same lengths.

These results are in agreement with previous studies using climatological (Levitus 1987) or satellite (Ghirardelli et al. 1995) data. However, in comparison with them our estimates of the magnitude of the annual mean EHF using any of the four sources of winds are lower. This is mostly because of the difference in the definition of mean Ekman layer temperature and to a much lesser degree is due to the discrepancies in the wind measurements. In previous studies the sea surface temperature is used as representative of the temperature of the Ekman layer. In our study we used a velocity-weighted temperature for the first 100 m of the ocean combining a high-resolution SST field with monthly climatological subsurface temperatures. As a result, the temperature of the Ekman layer that we estimate is significantly lower than the SST.

There are some noticeable differences among the EHF annual mean estimated from the four wind sources. The discrepancies are larger in the Southern Hemisphere (Fig. 1). The largest discrepancies in the annual mean of the zonally integrated meridional EHF occur in the tropical region and increase toward the equator. Two factors contribute to amplify the discrepancies in EHF between the sources toward the equator: the increasing temperatures and decreasing Coriolis parameter.

The annual means of the SSM/I, NCEP, and Quikscat-based EHF agree better with each other in the extratropical regions than in the Tropics. The Quikscat EHF annual mean is significantly larger than the others in the equatorial North Pacific. This difference could be related to intrinsic problems involving the scatterometer wind retrievals. The algorithm that converts the radar cross section into wind vectors is sensitive to local meteorological conditions. The wind stress depends on the stability of the boundary layer and on the moisture flux. Furthermore, scatterometer winds lose reliability under heavy rain and can be influenced by strong surface currents. All of these factors are particularly prone to occur within the equatorial belt. Therefore, the Quikscat-derived EHF may be overestimated for the equatorial North Pacific.

The Ekman heat flux of the Atlantic and the Pacific Oceans have a similar annual cycle; larger annual amplitudes are found in the Pacific. The meridional distribution of the annual cycle is not symmetric between the Northern and Southern Hemispheres (Fig. 2). This is mostly due to the presence of the intertropical convergence zone in the Northern Hemisphere associated with the unequal distribution of land and sea.

For latitudes lower than 30 the heat transported by the Ekman layer is poleward in the Atlantic and Pacific. Near 5N there is a weakening of this poleward flux in the North Atlantic and North Pacific during the autumn months (Fig. 2). These weak areas are caused by a change in the wind pattern because of the northward displacement of the intertropical convergence zone (ITCZ) in the Northern Hemisphere. For latitudes higher than 30 the Ekman flux is mostly equatorward in all oceans. In the north Indian Ocean there is a strong equatorward Ekman heat flux from April to November, consistent with the wind regime characteristic of the summer monsoon. Levitus (1987) pointed out that this wind regime dominates the global zonally averaged Ekman heat transport. However, we see that it is the combination of the Indian monsoon system with the wind patterns of the ITCZs that determine the distribution of the total EHF in the Northern Hemisphere.

Sato et al. (2002) analyzed the changes in the global scale EHF by either suppressing the variability in the temperature or suppressing the variability in the wind stress field. They found out that the changes in the EHF are dominated by the wind field. Therefore, we will focus our analysis on the variability of the wind stress only. The observed wind stress variability ranges from interannual, global scale to eddies of small scale. For the estimation of the annual cycle of the EHF we used the mean and the large-scale filtered components (τ and τl) of Eq. (5). The filtering process is able to precisely isolate most of the annual and interannual signals with a length scale comparable to the basin width (Fig. 3).

3a8082e126