Vertical Eddy Transport Research Articles

Mesoscale Ocean–Atmosphere Coupling Effects on the North Pacific Subtropical Mode Water

Abstract Subtropical mode water (STMW) is a thick layer of water mass characterized by homogeneous properties within the main pycnocline, important for oceanic oxygen utilization, carbon sequestration, and climate regulation. North Pacific STMW is formed in the Kuroshio Extension region, where vigorous mesoscale eddies strongly interact with the atmosphere. However, it remains unknown how such mesoscale ocean–atmosphere (MOA) coupling affects the STMW formation. By conducting twin simulations with an eddy-resolving global climate model, we find that approximately 25% more STMW is formed with the MOA coupling than without it. This is attributable to a significant increase in ocean latent heat release primarily driven by higher wind speed over the STMW formation region, which is associated with the southward deflection of storm tracks in response to oceanic mesoscale imprints. Such enhanced surface latent heat loss overwhelms the stronger upper-ocean restratification induced by vertical eddy and turbulent heat transport, leading to the formation of colder and denser STMW in the presence of MOA coupling. Further investigation of a multimodel and multiresolution ensemble of global coupled models reveals that the agreement between the STMW simulation in eddy-present/rich coupled models and observations is superior to that of eddy-free ones, likely due to more realistic representation of MOA coupling. However, the ocean-alone model simulations show significant limitations in improving STMW production, even with refined model resolution. This indicates the importance of incorporating the MOA coupling into Earth system models to alleviate biases in STMW and associated climatic and biogeochemical impacts. Significance Statement North Pacific subtropical mode water (STMW) is a distinct pycnostad within the main thermocline located south of the Kuroshio Extension. As short-term heat and carbon silos, STMW is traditionally thought to be driven by the basin-scale atmospheric forcing. The role of air–sea interactions at mesoscales residing in the Kuroshio Extension region has been overlooked. Here, we demonstrate that the strong thermal feedback of mesoscale sea surface temperature anomalies is not negligible for the STMW formation. This is achieved by accelerating wind and consequently promoting ocean latent heat release. Our results pinpoint the significance of accounting for the role of oceanic mesoscale feedback in improving the simulation of STMW as well as its climatic and biogeochemical impacts in Earth system models.

The Multi‐Scale Response of the Eddy Kinetic Energy and Transport to Strengthened Westerlies in an Idealized Antarctic Circumpolar Current

AbstractThe Southern Ocean's eddy response to changing climate remains unclear, with observations suggesting non‐monotonic changes in eddy kinetic energy (EKE) across scales. Here simulations reappear that smaller‐mesoscale EKE is suppressed while larger‐mesoscale EKE increases with strengthened winds. This change was linked to scale‐wise changes in the kinetic energy cycle, where a sensitive balance between the dominant mesoscale energy sinks—inverse KE cascade, and source—baroclinic energization. Such balance induced a strong (weak) mesoscale suppression in the flat (ridge) channel. Mechanistically, this mesoscale suppression is attributed to stronger zonal jets weakening smaller mesoscale eddies and promoting larger‐scale waves. These EKE multiscale changes lead to multiscale changes in meridional and vertical eddy transport, which can be parameterized using a scale‐dependent diffusivity linked to the EKE spectrum. This multiscale eddy response may have significant implications for understanding and modeling the Southern Ocean eddy activity and transport under a changing climate.

Dependence of Convective Cloud Properties and Their Transport on Cloud Fraction and GCM Resolution Diagnosed from a Cloud-Resolving Model Simulation

The scale-aware convective parameterization for high resolution global climate models must satisfy the requirement that the parameterized subgrid convective transport diminishes as the model resolution increases to convection-resolving resolutions. A major assumption in current scale-aware convection schemes is that the differences between convective cloud properties and their environmental counterparts are independent of cloud fraction. This study examines convective cloud vertical velocity, moist static energy (MSE), moisture, and the vertical eddy transport of MSE and moisture for different averaging subdomain sizes and fractional convective cloudiness using a cloud resolving model simulation of a midlatitude mesoscale convective system. Results show that convective cloud fraction, mass flux, and vertical transport of MSE and moisture increase with decreasing subdomain size. The differences between convective cloud properties in both updrafts and downdrafts and their environment depend on both cloud fraction and the averaging subdomain size. For a given subdomain size, the differences increase with cloud fraction, in contrast to the assumption used in current scale-aware convection parameterization schemes. A consequence of this is that the parameterized convective eddy transport reaches maximum at a higher cloud fraction than believed in previous studies. This has implications on how fast the subgrid convective transport should diminish as GCM resolution increases.

Summertime Transport Pathways and Dynamics From Northern India and Tibetan Plateau to the Lower Stratosphere: Insights From Idealized Tracer Experiments

AbstractWe use the idealized tracer experiments and investigate the summertime transport from the surface region of northern India and Tibetan Plateau to the lower stratosphere. It is found that the transport, compared to other surrounding regions, has an overall younger modal age in the northern lower stratosphere away from the tropopause. Analysis of the tracer budget reveals that the tracer is transported to the tropical lower stratosphere rapidly in the first 5 days due to vertical eddy transport and afterward in a month or two advection associated with the Brewer‐Dobson circulation. Meanwhile the tracer is also transported to the northern extratropical lower stratosphere in the first 3 months due to horizontal eddy mixing. The results highlight the uniqueness of the northern India region in the summertime transport to the lower stratosphere and implications for the transport of short‐lived chemical species in the destruction of stratospheric ozone.

Vertical diffusion transport of atomic oxygen in the mesopause region consistent with chemical losses and continuity: Global mean and inter-annual variability

This study describes the vertical diffusion velocity and eddy transport coefficient means derived from the climatological measurements of mesospheric [O] by SABER. Fifty years ago, Colegrove et al. (1967) described a process to transition between the two regions of the atmosphere using transport and continuity equations along with a 1-D composition profile. The two parts of this study are calculations of the downward O diffusion velocities and eddy diffusion coefficients using O densities derived from SABER for: 1) the global mean (±55°) and 2) the inter-annual variation of global and equatorial annual mean O densities. The derived transport diffusion velocities range from ∼1 to 4 cm s−1 at 95-83 km, and kzz varies between ∼9.0–5.0 × 105 cm2 s−1 over the same altitude range. The values of kzz derived here in the 80–100 km region are larger at 98 km than those calculated by Salinas et al. (2016) using SABER CO2 measurements and ∼0.5 times less than Qian et al. (2009, 2013 and TIEGCM model) which were derived from studies of thermospheric O/N2 measurements. Consideration of the kzz altitude profiles in the mesosphere, the mean values for both Salinas et al. (2016) and this study is ∼7. E+05 cm2s-1. The standard deviation of the inter-annual variability in kzz (and downward transport velocities) from globally averaged O profiles is <5%. The deviation is slightly larger (7.4%) at the equator, where a QBO trend is noted at 88 km, although not sufficiently large to be detected in the time history of the mesospheric O column density. The global average column density of O in the mesosphere (80–100 km) had a 0.30% change per sunspot. We suggest that model studies may need to consider higher turbulence (and kzz) in the 100–110 km region in order to satisfy both mesospheric and thermospheric composition.

Odd oxygen and its atmospheric lifetime

The concept of “odd oxygen” is considered and its atmospheric lifetime is evaluated. It is demonstrated that the modern interpretation of this concept precludes obtaining reliable data on the atmospheric lifetime of odd oxygen, which is of fundamental importance in ozonosphere chemistry. An algorithm for correctly assessing the atmospheric lifetime of odd oxygen is presented and relevant estimates for the end of the XX century are made. It is also shown that, if the lifetime of odd oxygen is comparable to or greater than the time of vertical eddy transport, the former should be replaced by a combined lifetime, taking into account the action of both photochemical and dynamic factors.

A Unified Representation of Deep Moist Convection in Numerical Modeling of the Atmosphere. Part II

Abstract In Part I of this paper, a generalized modeling framework for representing deep moist convection was presented. The framework, called unified parameterization, effectively unifies the parameterizations in general circulation models (GCMs) and cloud-resolving models (CRMs) and thus is applicable to any horizontal resolution between those typically used in those models. The key parameter in the unification is the fractional convective cloudiness σ, which is the fractional area covered by convective updrafts in the grid cell. The central issue of Part I is to formulate the σ dependence of vertical eddy transports of thermodynamic variables and to determine σ for each realization of grid-scale processes. The present paper completes the formulation through further analysis of the simulated data. The analyzed fields include the vertical structure of the σ dependence of vertical and horizontal eddy transports of moist static energy and horizontal momentum and that of cloud microphysical sources. For the momentum transport, the analysis results clearly show the limits of the traditional approach of parameterization based on an effectively one-dimensional model. For cloud microphysical conversions, it is shown that those taking place primarily inside and outside the updrafts are roughly proportional to σ and 1 − σ, respectively.

Comparison of lower stratospheric tropical mean vertical velocities

We have analyzed 13 years (1993–2005) of tropical stratospheric water vapor data from the Halogen Occultation Experiment and over 3 years of data (October 2004 through November 2007) from the Aura Microwave Limb Sounder. By correlating the phase lag of the water vapor “tape recorder” signal between levels we estimate the time mean vertical velocity. Our estimated vertical velocity compares well with calculations from the Goddard Earth Observing System (GEOS) chemistry‐climate model (CCM) and from the GEOS data assimilation system. Between 18 and 26 km both the GEOS CCM simulations and water vapor observations agree that the vertical velocity is below 0.04 cm/s, with a minimum near 20 km of 0.03 cm/s. Vertical velocities deduced from water vapor observations are higher than those from the GEOS CCM in the region 16–18 km (0.04 cm/s) and above 26–30 km (up to 0.07 cm/s). These estimates are close to earlier estimates from a shorter water vapor record and radiative transfer models. No evidence is found for velocities as high as 0.15 cm/s as was recently estimated from aircraft CO2 measurements in the upper troposphere/lower stratosphere. Further diagnosis of the aircraft CO2 data and model simulations of CO2 show that while the CO2 data give an apparent upward transport velocity of ∼0.06 cm/s, about half of this is due to vertical and horizontal eddy transport. Accounting for the eddy terms gives a CO2‐based estimate of the vertical velocity of ∼0.03 cm/s, in much closer agreement with that estimated from water vapor.

Numerical Estimations of Horizontal Advection inside Canopies

Abstract Local advection of scalar quantities such as heat, moisture, or carbon dioxide occurs not only above inhomogeneous surfaces but also within roughness elements on these surfaces. A two-dimensional advection–diffusion equation is applied to examine the fractionation of scalar exchange into horizontal advection within a canopy and vertical turbulent eddy transport at the canopy top. Simulations were executed with combinations of various wind speeds, eddy diffusivities, canopy heights, and source strengths. The results show that the vertical turbulent fluxes at the canopy top increase along the fetch and approach a limit at some downstream distance. The horizontal advection in the canopy is maximum at the edge of canopy and decreases exponentially along the fetch. All cases have the same features, except the absolute quantities depend on the environmental conditions. When the horizontal axis is normalized by using the dimensionless variable xK/uh2, horizontal diffusion is negligible, and the upwind concentration profile is constant, the curves of horizontal advection and vertical flux collapse into single, unique lines, respectively (x is the horizontal distance from the canopy edge, K is the eddy diffusivity, u is the wind speed, and h is the canopy height). The ratios of horizontal advection to the vertical turbulent flux also collapse into one universal curve when plotted against the dimensionless variable xK/uh2, irrespective of source strength. The ratio R shows a power-law relation to the dimensionless distance [R = a(xK/uh2)−b, where a and b are constant].

Decomposition of dissolved organic matter from the continental margin

Decomposition of dissolved organic carbon, nitrogen and phosphorus (DOC, DON, DOP) was measured for surface and bottom waters of the middle Atlantic bight (MAB) and deep slope water adjacent to the MAB on two occasions in March and August 1996. We used standard bottle incubation techniques to measure the decrease in dissolved organic matter (DOM) concentrations over a 180-day interval. Generally DOM concentrations in the MAB were elevated (125mM DOC, 10.2mM DON and 0.30mM DOP) relative to the surface ocean and deep slope water (46.7mM DOC, 2.76mM DON, 0.03mM DOP). On average the C:N:P ratio of shelf DOM (431:36:1) was substantially higher than the Redfield ratio, but not nearly as high for that of deep slope water (2700:215:1). Decomposition time course data were fit to a three-pool (very labile, labile, and recalcitrant pools) multi-G model using a Marquardt fitting routine. The threepool model was superior to a simple exponential decay model assuming a single pool of DOM. We observed no significant changes in concentration of DOM in deep-water samples, attesting to the old age of this material, its recalcitrant nature, and the cleanliness of our technique for measuring decomposition. There were major differences in the relative amount of very labile, labile and recalcitrant fractions of shelf-water DOC, DON and DOP as a result of preferential remineralization of P over N and N over C. Averaged over stations, the decomposable portion of the bulk DOC, DON and DOP pools increased from 30% to 40% to 81% for C, N and P. There was a wide range in decay coefficients for the very labile and labile DOM pools: average decay coefficient for the very labile pool was 0.219 d � 1 , and 0.018 d � 1 for the labile pool. Average half-lives calculated from the decay coefficients were 4, 12 and 8 days for the very labile DOC, DON and DOP pools, and 54, 113 and 90 days for the labile DOC, DON and DOP pools. On the basis of pool turnover times relative to shelf-water residence time (B100 days) we conclude that autochthonous algal production is the source of the very labile DOM pools. Its rate of production is sufficient to sustain estimated rates of bacteria C demand in continental margins. Our results for the MAB indicate that while substantial amounts of DOM are remineralized in the same time frame as shelf-water residence time, there is substantial DOM remaining that is depleted in N and P relative to C. Strong concentration gradients in DOM occur between shelf and ocean waters and between surface and deeper waters. Coupled with appropriate vertical and horizontal advective and eddy diffusive transports, DOM export from the MAB and other shelf systems may be a significant component of ocean C dynamics. r 2002 Elsevier Science Ltd. All rights reserved.

Predominant semi-annual oscillation of the upper mesospheric airglow intensities and temperatures in the equatorial region

The upper mesospheric and lower thermospheric airglow emissions, OI 557.7 nm, NaD 589.3 mn and the OH (9,4) band and its rotational temperature have been measured using a ground-based multichannel airglow photometer located at Fortaleza (3.9°S, 38.4°W) since 1986. The observed emission intensities show predominantly semi-annual oscillations with maxima at the equinoxes and minima during the solstices. The amplitudes of the oscillations are larger than those observed from the low latitude station, Cachoeira Paulista (22.7°S, 45.0°W). The OH rotational temperature, which represents a gas kinetic atmospheric temperature at around 85–95 km, also shows a strong semi-annual oscillation, 18 K peak to peak, with an. average value around 10 K higher than that observed from Cachoeira Paulista. These results do not agree with model atmospheres presently available. It is suggested that the differences result from the effects of seasonal variations in vertical eddy transport and/or meridional circulation.

Heating, Moisture, and Water Budgets of Tropical and Midlatitude Squall Lines: Comparisons and Sensitivity to Longwave Radiation

A 2D time-dependent and nonhydrostatic numerical cloud model is presently used to estimate the heating, moisture, and water budgets in the convective and stratiform regions for both a tropical and a midlatitude squall line. The model encompasses a parameterized, three-class ice phase microphysical scheme and longwave radiative transfer process. It is noted that the convective region plays an important role in the generation of stratiform rainfall for both cases. While a midlevel minimum in the moisture profile for the tropical case is due to vertical eddy transport in the convective region, the contribution to the heating budget by the cloud-scale fluxes is minor; by contrast, the vertical eddy heat-flux is relatively important for the midlatitude case due to the stronger vertical velocities present in the convective cells.

Feedback of Deep Moist Convection to Its Near Environment as Diagnosed from Three-Dimensional Numerical Model Output. Results from an Early Experiment

This paper presents preliminary results from an investigation into the feedback between convective storms and their new surroundings, using output from a three-dimensional anelastic cloud-scale model. Convective feedback budgets for heat, moisture and horizontal momentum are diagnosed from horizontally Reynolds-averaged governing equations, analogous to the theory of Anthes. There is limited horizontal scale separation between the active convection area and the averaging area, which at 30 km on a side is comparable to one grid cell of a typical mesoscale numerical weather prediction model. The simulation is run with an idealized midlatitude severe thunderstorm sounding. The resulting storm displays several supercell features. These include a vigorous erect large-diameter updraft that splits at lower levels, a vaulted weak echo region in the lower part of the main (right flank) updraft core, and a midlevel mesovortex couplet with cyclonic vorticity in the main updraft. The vertical profiles of the various budget terms show several findings of potential relevance to cumulus parameterization. The vertical eddy transport (flux divergence) is highly important to each budget; it significantly raises the height of the maximum apparent heat source and lowers the height of the maximum apparent moisture sink, and acts to reduce the net tropospheric vertical wind shear. At the same time, the horizontal eddy momentum transport and the mean horizontal pressure gradient force both act strongly to sharpen the tropopause-level jet, so that the net shear is little changed. The horizontal eddy transport is much less important to the heat budget, and remains negligible for moisture. Mean storage contributes significantly to the apparent source for each budget under consideration. Other terms derived by Anthes but ignored in existing cumulus parameterizations (the resultant of eddy storage, transport of mean fields by the eddy wind, and transport of eddy fields by the mean wind) become quite appreciable. These terms partially oppose the condensational heating and drying, and the processes that sharpen the tropopause-level jet.

Quantitative and Theoretical Analyses of an Experimental Helical Vortex

Abstract This paper extends the earlier preliminary results of a laboratory simulation of a mesocyclonic circulation with an embedded tornadic vortex, produced by a vertically veering but nonrotating inflow with high helicity. Higher quality velocity measurements are performed which allow more detailed quantitative analyses, and are interpreted with the aid of a secondary flow model. The vortex is produced by vertical eddy transport of angular momentum generated by a helical secondary flow. The positive angular momentum on the right side of the inflow remains at low levels until it is stretched in the vortex core, while the negative angular momentum on the left side is lifted and carried out of the circulation.

A growth‐sedimentation model of polar mesospheric clouds: Comparison with SME measurements

A numerical model for the brightness of polar mesospheric clouds (PMC) is described, and is compared with measurements from the Solar mesosphere Explorer (SME) satellite. These clouds occur during the summer months at polar latitudes where temperatures are known to fall below 140 K. We calculate the optical properties of a cloud by simulating the growth and sedimentation of ice particles at the cold supersaturated mesopause. Time‐dependent trajectories of ice particles are calculated from their origin at the temperature minimum region to their demise at the cloud base through evaporation. We consider the effects of the removal of atmospheric water vapor by the growing particles and its restoration by ice evaporation at the cloud base. This “freeze‐drying” effect is crucial in limiting the maximum size of the particles and therefore the maximum brightness of the cloud. Assuming spherical particles, Mie‐scattering calculations of the directional albedo of a cloud are performed using a range of possible values for the atmospheric variables (water vapor mixing ratio, temperature, upward wind speed, atmospheric pressure, and eddy diffusion coefficient). We find that for a nominal atmospheric case the model predicts a moderately weak cloud at 265 nm, the wavelength of the SME measurements. An extreme model (cold and moist with high vertical wind and eddy transport) is needed to account for the brightest cloud observed. We estimate an upper limit for the water vapor to be of the order of 5 parts per million by volume. Higher values would imply the existence of clouds which exceed in brightness every cloud observed by the satellite over the time period 1981–1986. SME observations of greater cloud height in the northern hemisphere, despite their greater brightness, possibly imply an excess (by a factor of 2) of northern hemisphere water vapor. This holds if the other atmospheric variables (and cloud particle numbers) are the same in both north and south. Dependence of model cloud brightness on atmospheric pressure (∼P4.4), water vapor mixing ratio (∼w2.8), thickness of the cloud saturation region (∼d4) and advective wind speed (∼υ) are determined. These scalings are shown to result from a calculated proportionality of the cloud brightness to R6 (R is the maximum particle radius), and from simple considerations of ice layer growth, particle sedimentation, and the mass budget of water. Mie scattering calculations for a wavelength of 550 nm show that SME and OGO‐6 data on PMC brightnesses are consistent.

The O/O 2 ratio and eddy mixing at heights from 85 km to the turbopause

The O/O 2 ratio at heights from 85 km to the turbopause is represented parametrically in terms of eddy diffusion coefficient K, solar dissociating flux and ambient mesospheric conditions appropriate to a given date and latitude. The relation obtained is consistent with results of detailed calculations undertaken by various authors for the concentration ofphotochemically produced O under conditions of vertical eddy transport and chemical removal in an otherwise static atmosphere.

The vertical distribution of tropospheric ammonia

A one‐dimensional tropospheric photochemical model is used to simulate measured profiles of NH3 obtained with the Infrared Heterodyne Radiometer. The relative roles of homogeneous loss, heterogeneous loss, and vertical eddy transport are discussed in terms of selecting parameters which best fit the measurements. The best fit was obtained for a vertical eddy diffusion coefficient of 2 × 105 cm−2 sec−1 or greater (corresponding to a characteristic vertical transport time in excess of about 35 days), and a characteristic heterogeneous loss time in excess of 10 days. The characteristic homogeneous chemical loss time was found to be about 40 days at the surface and decreased to about l80 days at 10 km, and not very sensitive to model chemical perturbations. Increased ground level concentrations of NH3 to about 10 ppb, compared to background surface concentrations of about 1 ppb, were measured several weeks after application of ammonium nitrate fertilizer. This suggests that the volatilization of ammonium nitrate fertilizer is rapid, and an important source of NH3. Because of the characteristic times for the loss mechanisms, synoptic time scale phenomena may play an important role in determining the tropospheric distribution of NH3 concentrations.

Three-dimensional time-dependent simulations of hydro-thermal behaviour of Biscayne Bay

The purpose of the paper is to evaluate the performance of rigid-lid formulations in predicting velocity and temperature fields in domains where the applicability of rigid-lid formulations is not immediately obvious. A test was conducted to determine whether local depth scaled vertical exchange coefficients yield better simulations compared to constant vertical coefficient formulations. Extensive infrared imagery was used to evaluate the model's ability to predict temperature yields in Biscayne Bay, Fla. Two cases, one with constant vertical eddy transport coefficient, and the other with vertical transport coefficient scaled with local depth are discussed.

The evolution and variability of atmospheric ozone over geological time

The rise of atmospheric O 3 as a function of the evolution of O 2 has been investigated using a one-dimensional steady-state photochemical model based on the chemistry and photochemistry of O x (O 3, O, O( 1D)), N 2O, NO x (NO, NO 2, HNO 3), H 2O, and HO x (H, OH, HO 2, H 2O 2) including the effect of vertical eddy transport on the species distribution. The total O 3 column density was found to maximize for an O 2 level of 10 −1 present atmospheric level (PAL) and exceeded the present total O 3 column by about 40%. For that level of O 2, surface and tropospheric O 3 densities exceeded those of the present atmosphere by about an order of magnitude. Surface and tropospheric OH densities of the paleoatmosphere exceeded those of the present atmosphere by orders of magnitude. We also found that in the O 2-deficient paleoatmosphere, N 2O (even at present atmospheric levels) produces much less NO x than it does in the present atmosphere.

On the vertical eddy transports in the northern atmosphere 2. Vertical eddy momentum transport for summer and winter

Complete hemispheric budgets of relative angular momentum are presented for the solstice seasons. Domain is the zonal mean northern atmosphere, divided into 4 × 4 equal‐area/equal mass annuli (Δ sin ϕ = 0.25, ΔP = 250 mbar) plus four annuli in the southern tropical belt, making 20 annuli in total. Emphasis is on the vertical momentum transport due to eddies on all scales. This component is determined as being residual from the budget equation. Observed transports, used as specified input, include net horizontal momentum transport across latitude circles, vertical cell transport of Ω momentum and relative momentum across pressure surfaces, and vertical momentum transport across the earth's surface due to mountain torque (observed) plus stress torque (parameterized). The input data are not adjusted, although the transports, being from different sources, do not make balanced budgets. Rather, the residual transports are distributed in the vertical such that the annulus imbalances are minimized. The imbalances represent the net reliability of the input data plus the error of miscellaneous torques which are not explicitly specified (storage, mass drift, and variation of annulus area with height). Principal results are: (1) The vertical eddy momentum transport (WE) is of the order of 10 Hadley units (1 HU = 1018J). In the tropics it is upward in summer and downward in winter, presumably because of the location of the Intertropical Convergence Zone and because of the Asiatic monsoon. In the extratropics it is upward in both seasons. (2) The vertical cell transport of Ω momentum (PSI) is represented by the mass flux stream function. It is also of the order of, and opposite to, WE. The principal momentum balance is between WE and PSI; the horizontal transports are significantly smaller (except across 30°N in winter). (3) The hemispheric rms imbalance, averaged over all annuli and seasons, is 0.4 HU. This is well below the order of WE and is considered as the statistical error of the vertical eddy transports. (4) Possible systematic errors were investigated with a sensitivity test by independently changing the scales of PSI and of the stress torque (TL). The rms imbalance reaches a minimum for 78%of the observed PSI and for 90% of the TL. This seems to indicate that both the observed meridional wind and the surface wind stress as used in this study are overestimates. When the reduced fields are used as input variables, WE is also reduced by about a quarter; however, the patterns of WE remain essentially unaltered. Thus our vertical eddy momentum transport estimates deserve credence with an error margin of 25%. (5) WE is highly variable throughout the year, mostly because of the variability of PSI. Although WE is, on the seasonal time scale, large and clearly a significant link in the momentum budget, its annual values are small and much less significant. The reason is that opposite processes work together to form WE. (6) WE combines the net effect of all eddy processes, from the molecular up to the synoptic scale. It is not possible with the present data to split WE into its components. Various mechanisms contributing to WE are speculatively discussed. They include the hot tower mechanism, which seems to be predominant in the tropics and over the monsoon areas as well as in extratropical convective zones, and baroclinic plus gravity wave processes in the extratropics.