Momentum, Heat, and Vapor Fluxes

PK Taylor, Southampton Oceanography Centre, Southampton, UK

Ó 2003 Elsevier Ltd. All rights reserved.
This article is reproduced from the previous edition, volume 1, pp 93–100, Ó 2003, Elsevier Ltd.

Introduction contrast to land areas, the typical day to night variation in sea
and air temperatures is small, typically less than 1 C. Both the
The maintenance of the Earth’s climate depends on a balance sea and the sky emit and absorb long-wave radiative energy
between the absorption of heat from the Sun and the loss of (wavelength 3–50 mm). Because, under most circumstances, the
heat through radiative cooling to space. For each 100 W of the radiative temperature of the sky is colder than that of the sea, the
Sun’s radiative energy entering the atmosphere nearly 40 W is downward long-wave flux is usually smaller than the upward
absorbed by the ocean – about twice that absorbed in the flux. Hence the net longwave flux acts to cool the sea surface,
atmosphere and three times that absorbed by land surfaces. typically by 30 W m
2 (if cloudy) to 80 W m
2 (clear skies).
Much of this heat is transferred to the atmosphere by the local The turbulent fluxes of sensible and latent heat also typi-
sea to air heat flux, a major component of which is caused by the cally transfer heat from sea to air. The sensible heat flux is the
evaporation of water vapor. Although about 90% of the evap- transfer of heat caused by the difference in temperature
orated water falls back into the sea (see Air Sea Interactions: between the sea and the air. Over much of the ocean this flux
Freshwater Flux), the remainder represents about one-third of cools the sea by perhaps 10–20 W m
2. However, where cold
the precipitation which falls over land. The geographical varia- wintertime continental air flows over warm ocean currents, for
tion of the atmospheric heating drives the weather systems and example, the Gulf Stream region off the eastern seaboard of
their associated winds. The wind transfers momentum to the North America, the sensible heat flux may reach 100 W m
2.
sea, causing waves and the wind-driven currents. Major ocean Conversely, in regions like the summertime North Pacific
currents transport heat poleward and at higher latitudes the sea Ocean, warm winds blowing over a colder ocean may result in
to air heat transfer significantly ameliorates the climate. Thus the a small sensible heat flux into the ocean. Under most weather
heat, water vapor, and momentum fluxes through the ocean conditions the evaporation of water vapor from the sea surface
surface form a crucial component of the Earth’s climate system. results in a water vapor flux from the sea to the air. The latent
Having defined the various fluxes and their order of heat flux is the heat absorbed on vaporization of the water. This
magnitude, this article will review methods of flux measure- heat is released to warm the atmosphere when the vapor
ment and the sources of flux data. The regional and seasonal condenses to form clouds. Usually the latent heat flux is
variation of the fluxes will be summarized. Following a significantly greater than the sensible heat flux, being on
discussion of the accuracy of our present flux estimates, the average 100 W m
2 or more over large areas of the ocean. Over
potential for future improvements will be considered. regions such as the Gulf Stream, latent heat fluxes of several
hundred W m
2 are observed. In foggy conditions, with the air
warmer than the sea, condensation may occur on the sea
Definition of the Fluxes surface, and the vapor flux and latent heat flux are directed from
air to sea. In summertime over the fog-shrouded Grand Banks
The momentum flux is the downward transfer of horizontal off Newfoundland, the mean monthly latent heat transfer is
momentum caused by the drag of the sea surface on the wind. directed into the ocean, but this is a very exceptional case.
The wave-covered sea surface is continually in motion and,
compared with typical land surfaces, appears remarkably
smooth to the air flow. For gale force winds the waves may be Measuring the Fluxes
10 m or more in height, but the momentum flux is no more
than that which would occur over a flat plain. As a result, wind For the radiative fluxes, the standard method is to measure the
speeds over the ocean tend to be greater than those over land. voltage generated by a thermopile exposed to the incident
The total heat transfer through the ocean surface, the net heat radiation. A pyranometer, mounted in gimbals for use on a ship
flux, is a combination of several components. The shortwave or buoy, is used to measure the incoming shortwave radiation
radiative flux (wavelength 0.3–3 mm) is the heat input from the (Figure 1). For better accuracy the direct and diffuse compo-
Sun. Around noon on a sunny day this flux may reach about nents should be determined separately but at present this is
1000 W m
2 but, when averaged over 24 hours, a typical value rarely done over the sea. The reflected short-wave radiation is
is 100–300 W m
2, varying with latitude and season. normally calculated using tabulated values of the albedo for
Depending on the solar elevation and the sea state, about 6% of different solar elevations. Pyrgeometers, used for longwave
the incident radiation is reflected from the sea surface. Most of radiation, are similar to pyranometers but have a coated dome
the remainder is absorbed in the upper few meters of the ocean. to filter out the shortwave radiation. For these the use of gimbals
In calm weather, with winds less that about 3 m s
1, a shallow is less important, but a clear sky view is required and corrections
layer may form during the day in which the sea has been for the dome temperature and short wave leakage are needed.
warmed by a few degrees Celsius (a ‘diurnal thermocline’). Again, only the downward component is normally measured;
However, under stronger winds or at night the absorbed heat the upward component is calculated from the temperature and
becomes mixed down through several tens of meters. Thus, in the emissivity of the sea surface.

Encyclopedia of Atmospheric Sciences 2nd Edition, Volume 1 http://dx.doi.org/10.1016/B978-0-12-382225-3.00064-5 129

130 Air Sea Interactions j Momentum, Heat, and Vapor Fluxes

Figure 1 A pyranometer used for measuring short-wave radiation. The

thermopile is covered by two transparent domes. Photograph courtesy
of Southampton Oceanography Centre.

The turbulent fluxes may be measured directly in the near

surface atmosphere using the eddy correlation method. For
example, if upward-moving air in the turbulent eddies is on
average warmer and moister than the downward-moving air,
then there is an upward flux of sensible heat and water vapor
and hence also of latent heat. Similarly, the momentum flux
may be determined from the correlation between horizontal
and vertical wind fluctuations. Accurate eddy correlation
measurements over the ocean are difficult. Since a large range
of eddy sizes may contribute to the flux, fast-response sensors
capable of sampling at 10 Hz or more must be exposed for Figure 2 The sensing head of a three-component ultrasonic anemom-
periods of order 30 minutes for each flux determination. For eter. The wind components are determined from the different times taken
instrumentation mounted on a buoy or ship the six compo- for sound pulses to travel in either direction between the six ceramic
nents of the wave-induced motion must be measured and transducers. Photograph courtesy of Southampton Oceanography Centre.
removed in the signal processing procedure. The distortion
both of the turbulence and of the mean wind by ship, buoy, or
fixed tower must be minimized and, if possible, corrected for. of some quantity x is given by eqn [1], where r is the air density
While three-component ultrasonic anemometers (Figure 2) are and Uz is the wind speed at the measurement height.
relatively robust, the sensors for measuring fluctuations in
Fx ¼ rUz Cxz ðxz
x0 Þ [1]
temperature and humidity have previously been fragile and, in
the marine atmosphere, prone to contamination by salt parti- While appearing intuitively correct (for example, blowing
cles and sea spray. Improved sonic thermometry, and water over a hot drink will cool it faster) these formulas can also be
vapor instruments using microwave refractometry or differen- derived from turbulence theory. The value for the transfer
tial infrared absorption, are relatively recent developments. coefficient, Cxz, characterizes both the surface roughness
Thus eddy correlation measurements are not routinely ob- applicable to x and the relationship between Fx and the vertical
tained over the ocean; rather they are used in air–sea interaction profile of x. The transfer coefficient varies with the atmospheric
experiments to calibrate other flux estimation methods. stability, which itself depends on the momentum, sensible
In the inertial dissipation method, fluctuations of the wind, heat, and water vapor fluxes as well as the measurement height.
temperature, or humidity at a few hertz are measured and Thus, although it may appear simple, eqn must be solved by
related, through turbulence theory, to the fluxes. This method is iteration, initialized using the equivalent neutral value of Cxz at
less sensitive to flow distortion or platform motion but relies some standard height (normally 10 m), Cx10n. Typical neutral
on various assumptions regarding the formation and dissipa- values are shown in Table 1.
tion of turbulent quantities that may not always be valid. It has Many research problems remain. For example: the increase
been implemented on some research ships to increase the range of CD10n at higher wind speeds must depend on the varying sea
of available flux data. state, but can the latter be successfully characterized by the ratio
The bulk (aerodynamic) formulas are the most commonly of the predominant wave speed to the wind speed (the wave
used method of flux estimation. The flux is determined from age), or by factors like the wave height and steepness, or is
the difference between the temperature, humidity or wind at a complete spectral representation of the wave field required?
some measurement height, z, and the value assumed to exist at What are the effects of waves propagating from other regions at
the sea surface – respectively the sea surface temperature, 98% varying angles to the wind (i.e., swell waves)? What is the exact
saturation humidity (to allow for salinity effects), and zero behaviour of CD10n in low wind speed conditions? Since CE10n
wind (or any non-wind-induced water current). Thus the flux Fx and CH10n are not well defined by the available experimental
 Air Sea Interactions j Momentum, Heat, and Vapor Fluxes 131

Table 1 Typical values (with estimated uncertainties) for the transfer coefficients. Neither the low wind speed
formula for CD10n nor the wind speed below which it should be applied, are well defined by the available, very scattered,
experimental data. It should be taken simply as an indication that, at low wind speeds, the surface roughness increases
as the wind speed decreases

Flux Transfer coefficient Typical values

Momentum Drag coefficient, CD10n 0.61 (0.05) þ 0.063 (0.005) U10n (U10n > 3ms
1)
0.61 þ 0.57/U10n (U10n < 3ms
1)
Sensible heat Stanton no., CH10n 1.1 (0.2)10
3
Latent heat Dalton number, CE10n 1.2 (0.1)10
3

data, recent implementations of the bulk algorithms have used shown in marine climatological atlases, of which examples are
theoretical models of the ocean surface (known as surface presented below.
renewal theory) to predict these quantities from the momentum Satellite-borne sensors can overcome these sampling prob-
roughness length. lems. ‘Passive’ sensors measure the radiation emitted from the
sea surface and the intervening atmosphere at visible, infrared,
or microwave frequencies; ‘active’ sensors transmit microwave
Sources of Flux Data radiation and measure the returned signal. The problem is to
develop methods of determining the fluxes from the various
Until recent years the only routinely available sources of data satellite data that can be obtained. For example, sea surface
were the weather reports from merchant ships. Organized as temperature has been routinely determined using visible and
part of the World Weather Watch system of the World Meteo- infrared radiometers since about 1980. However, the data must
rological Organization, these voluntary observing ships (VOS) be frequently checked against ship and buoy values to avoid
are asked to transmit coded weather messages at 00.00, 06.00, errors due to changes in atmospheric aerosol content that may
12.00 and 18.00 GMT daily, and to record a more detailed follow volcanic eruptions. Satellite-derived fields of net surface
observation in a weather logbook. The very basic instruments shortwave radiation are available; values for the net surface
used normally include a barometer and a means of measuring longwave radiation are less accurate. The surface wind velocity
air temperature and humidity: wet and dry bulb thermometers can be determined to good accuracy by active scatterometer
mounted in a hand-swung sling psychrometer or in a fixed, sensors by measuring the microwave radiation back-scattered
louvered Stevenson screen. Sea temperature is obtained using from the sea surface. The determination of near-surface air
a thermometer and an insulated bucket, or by reading the temperature and humidity from satellites is hindered by the
temperature of the engine cooling water intake. Depending on relatively coarse vertical resolution of the retrieved data. Thus the
which country recruited the VOS, an anemometer and wind radiation emitted by the near-surface air is dominated by that
vane might be provided, or the ship’s officers might be asked to originating from the sea surface. Statistically based algorithms
estimate the wind velocity from the sea state using a tabulated for determining the near-surface humidity have been success-
Beaufort scale. Because of the problems of adequately siting an fully developed. More recently, neural network techniques have
anemometer and maintaining its calibration, these visual esti- been applied to retrieving both air temperature and humidity;
mates are not necessarily considered inferior to anemometer- however, there is presently no routinely available product. Thus
based values. the satellite flux products for which useful accuracy has been
The bulk formulas are used to calculate the turbulent fluxes demonstrated on a global basis are presently limited to
from the VOS observations. However, in many cases the momentum, short-wave radiation, and latent heat flux.
accuracy is poor. In particular, a large ship can produce Numerical weather prediction (NWP) models (as used in
significant changes in the local temperature and wind flow. The weather forecasting centers) estimate values of the air–sea fluxes
radiative fluxes must be estimated from the observer’s estimate as part of the calculations. Assimilating much of the available
of the cloud amount plus, for short wave, the solar elevation, or data from the World Weather Watch system, including satellite
for long wave, the sea and air temperature and humidity. The data, radiosonde profiles, and surface observations, NWP
unavoidable observational errors and the crude form of the models are potentially the best source of flux data. However,
radiative flux formulas imply that large numbers of reports are there are a number of problems. The vertical resolution of these
needed, and correction schemes must be applied, before models is relatively poor and many of the near-surface processes
satisfactory flux estimates can be obtained. Though there are have to be represented in terms of larger-scale parameters.
presently nearly 7000 VOS, they tend to be concentrated in the Improvements to NWP models are judged on the resulting
main shipping lanes. While coverage over most of the North quality of the weather forecasts, not on the accuracy of the
Atlantic and North Pacific is adequate to provide monthly surface fluxes, which may become worse. Indeed, the continual
mean flux values, elsewhere data is mainly restricted to rela- introduction of model changes results in time discontinuities in
tively narrow, major trade routes. For most of the Southern the output variables. Thus the determination of interannual
Hemisphere the VOS data are only capable of providing useful variations is difficult and, for that reason, centers such as the
values if averaged over several years, and reports from the European Centre for Medium Range Weather Forecasting
Southern Ocean are very few indeed. These problems must be (ECMWF) and the US National Centers for Environmental
borne in mind when studying the flux distribution maps Prediction (NCEP) have recently reanalyzed the past weather
132 Air Sea Interactions j Momentum, Heat, and Vapor Fluxes

over several decades. The surface fluxes from these reanalyses are many books on marine meteorology. The major features of
receiving much study. Those presently available appear less the wind stress variability derived from ship observations
accurate than fluxes derived from VOS data in regions where from the period 1980 to 1993 are summarized here, using
there are many VOS reports; in sparsely sampled regions the plots for January and July to illustrate the seasonal variation
model fluxes are more accurate. Particular weaknesses for the (Figure 3).
models are in the shortwave radiation and latent heat fluxes. In Northern Hemisphere winter (Figure 3(a)) large wind
New reanalyses are underway and efforts are being made to stresses due to strong mid-latitude westerly winds occur in the
improve the flux estimates; eventually these reanalyses will North Atlantic and the North Pacific west of Japan. To the
provide the best source of flux data for many purposes. south, the extratropical high-pressure zones result in low wind
stress values, and south of these is the belt of north-east trade
winds. The very light winds of the Intertropical Convergence
The Regional and Seasonal Variation Zone (ITCZ) lie close to the Equator in both oceans. In the
of the Momentum Flux summertime Southern Hemisphere the south-east trade wind
belt is less pronounced. The extratropical high-pressure regions
The main features of the wind regimes over the global oceans are extensive but, despite it being summer, high winds and
have long been recognized and descriptions are available in significant wind stress exist in the mid-latitude Southern

Figure 3 Monthly vector mean wind stress (N m
2) for (a) January and (b) July calculated from voluntary observing ship weather reports for the
period 1980 to 1993. Adapted from Josey SA, Kent EC, and Taylor PK (1998). The Southampton Oceanography Centre (SOC) Ocean–Atmosphere Heat,
Momentum and Freshwater Flux Atlas, SOC Report No. 6.
 Air Sea Interactions j Momentum, Heat, and Vapor Fluxes 133

Ocean. The north-east monsoon dominates the wind patterns about 20–30 W m
2, being dominated by the variations in
in the Indian Ocean and the South China Sea (where it is latent heat flux. In winter the typical variability increases to
particularly strong). In the latter regions the ITCZ is a diffuse about 30–40 Wm
2, although in particular areas (such as over
area south of the Equator with relatively strong south-east trade the Gulf Stream) variations of up to 100 W m
2 may occur.
winds in the eastern Indian Ocean. The major spatial pattern of interannual variability in the
In Northern Hemisphere summer (Figure 3(b)) the wind North Atlantic is known as the North Atlantic Oscillation
stresses in the mid-latitude westerlies are very much decreased. (NAO). This represents a measure of the degree to which
Both the north-east and the south-east trade wind zones are mobile depressions, or alternatively near-stationary high-
evident respectively to the north and south of the ITCZ, which pressure systems, occur in the mid-latitude westerly zone.
mainly lies north of the Equator. The south-east trades are
particularly strong in the Indian Ocean and feed air across the
Equator into a very strong south-westerly monsoon flow in the Discussion: Accuracy of Flux Estimates
Arabian Sea. These ship data indicate very strong winds in and Future Trends
the Southern Ocean south-west of Australia. Such winds are
also evident in satellite scatterometer data, which suggest that We have seen that, although the individual flux components
the winds in the Pacific sector of the Southern Ocean, while are of the order of hundreds of W m
2, the net heat flux and its
still strong, are somewhat less than those in the Indian Ocean interannual variability over much of the world ocean is of the
sector. In contrast, the ship data appear to show light winds. order of tens of W m
2. Furthermore, it can be shown that
The reason is that in wintertime there are practically no VOS a flux of 10 W m
2 over one year would, if stored in the top 500
observations in the far South Pacific. The analysis technique m of the ocean, heat that entire layer by about 0.15  C.
used to fill in the data gaps has, for want of other information, Temperature changes on a decadal time scale are at most a few
spread the light winds of the extratropical high-pressure region tenths of a degree, so the global mean budget must balance to
farther south than is realistic; this is a good example of the better than a few W m
2. For these various reasons there is
care needed in interpreting the flux maps available in many a need to measure the flux components, which vary on many
atlases. time and space scales, to an accuracy of a few W m
2. Given the
available data sources and methods of determining the fluxes
described above, it is not surprising that this accuracy at present
The Regional and Seasonal Variation cannot be achieved.
of the Heat Fluxes To take an example, in calculating the flux maps shown in
Figure 4 many corrections were applied to the VOS observa-
The global distribution of the mean annual net heat flux is tions in an attempt to remove biases caused by the observing
shown in Figure 4(a). Averaged over the year, the ocean is methods. For example, air temperature measurements were
heated in equatorial regions and loses heat in higher latitudes, corrected for the ‘heat island’ caused by the ship heating up in
particularly in the North Atlantic. However, this mean distri- sunny, low-wind conditions. The wind speeds were adjusted
bution is somewhat misleading as the plots for January depending on the anemometer heights on different ships.
(Figure 4(b)) and July (Figure 4(c)) illustrate. The ocean loses Corrections were applied to sea temperatures calculated from
heat over most of the extratropical winter hemisphere and engine room intake data. Despite these and other corrections,
gains heat in the extratropical summer hemisphere. It is only the global annual mean flux showed about 30 W m
2 excess
because the tropical oceans are heated throughout the year, and heating of the ocean. Previous climatologies calculated from
atmospheric moisture from trade wind zones converges in the ship data had shown similar biases and the fluxes had been
ITCZ, that the tropics are so important with regard to driving adjusted to remove the bias, or to make the fluxes compatible
the atmospheric circulation. The major regions of ocean cool- with estimates of the meridional heat transport in the ocean.
ing occur in winter over the Gulf Stream and the Kuroshio However, comparison of the unadjusted flux data with accurate
currents. However, in summer the long period of day-light in data from air–sea interaction buoys showed good agreement
these mid-latitude regions results in mean short-wave radiation between the two. This suggests that adjusting the fluxes globally
values similar to or larger than those observed in equatorial is not correct and that regional flux adjustments are required;
regions. Thus the mean monthly short-wave flux is greater than however, the exact form of these corrections is presently not
the cooling induced by the combined latent heat and net long- known.
wave fluxes. At a more typical mid-latitude site the ocean cools In the future, computer models are expected to provide
in winter and warms in summer, in each case by around 100 W a major advance in flux estimation, Recently, coupled
m
2. The annual mean flux is small, of order 10 Wm
2, but numerical models of the ocean and of the atmosphere have
cannot be neglected because of the very large ocean areas been run for many simulated years, during which the
involved. modeled climate has not drifted. This suggests that the air–
Considering now the interannual variation of the surface sea fluxes calculated by the models are in balance with the
fluxes, the major large-scale feature over the global ocean is simulated oceanic and atmospheric heat transports.
the El Niño Southern Oscillation system in the equatorial However, it does not imply that at present the flux values are
Pacific Ocean. In the eastern equatorial Pacific the change in realistic. Errors in the short-wave and latent heat fluxes may
the net heat flux under El Niño conditions is around 40 W compensate one another; indeed, in a typical simulation the
m
2. For extratropical and mid-latitude regions the interan- sea surface temperature stabilized to a value that was, over
nual variability of the summertime net heat flux is typically large regions of the ocean, a few degrees different from that
134 Air Sea Interactions j Momentum, Heat, and Vapor Fluxes

Figure 4 Variation of the net heat flux over the ocean; positive values indicate heat entering the ocean: (a) annual mean; (b) January monthly mean;
(c) July monthly mean. Adapted from Josey SA, Kent EC, and Taylor PK (1998). The Southampton Oceanography Centre (SOC) Ocean–Atmosphere
Heat, Momentum and Freshwater Flux Atlas, SOC Report No. 6.

which is observed. Nevertheless, the estimation of flux most accurate routine source of air–sea flux estimates will be
values using climate or NWP models is a rapidly developing from numerical models of the coupled ocean–atmosphere
field and improvements will doubtless have occurred by system.
the time this article has been published. There will be
a continued need for in-situ and satellite data for assimila-
tion into the models and for model development and veri- See also: Aerosols: Observations and Measurements; Role in
fication. However, it seems very likely that in future the Radiative Transfer. Air Sea Interactions: Freshwater Flux; Sea
 Air Sea Interactions j Momentum, Heat, and Vapor Fluxes 135

Garratt, J.R., 1992. The Atmospheric Boundary Layer. Cambridge University Press,
Surface Temperature; Surface Waves. Boundary Layer
Cambridge.
(Atmospheric) and Air Pollution: Observational Techniques In Geernaert, G.L., Plant, W.J., 1990. Surface Waves and Fluxes. In: Current Theory,
Situ; Observational Techniques: Remote; Surface Layer. vol. 1. Kluwer, Academic, Dordrecht.
Climate and Climate Change: Climate Variability: North Atlantic Isemer, H.-J., Hasse, L., 1987. The Bunker Climate Atlas of the North Atlantic Ocean.
and Arctic Oscillation. Gravity Waves: Buoyancy and Buoyancy In: Air–Sea Interactions, vol. 2. Springer-Verlag, Berlin.
Josey SA, Kent EC and Taylor PK (1999) The Southampton Oceanography Centre
Waves: Optical Observations. Numerical Models: Regional (SOC) Ocean–Atmosphere Heat, Momentum and Freshwater Flux Atlas, SOC
Prediction Models. Tropical Cyclones and Hurricanes: Report No. 6, p. 30 1figs. (Available from The Library, Southampton Oceanography
Hurricanes: Observation. Centre, European Way, Southampton, SO14 3ZH, UK.)
Kraus, E.B., Businger, J.A., 1994. Atmosphere–Ocean Interaction, second ed. Oxford
University Press, New York.
Stull, R.B., 1988. An Introduction to Boundary Layer Meteorology. Kluwer Academic,
Further Reading Dordrecht.
Wells, N., 1997. The Atmosphere and Ocean: A Physical Introduction, second ed.
Browning, K.A., Gurney, R.J. (Eds.), 1999. Global Energy and Water Cycles. Cambridge Taylor and Francis, London.
University Press, Cambridge.
Dobson, F., Hasse, L., Davis, R. (Eds.), 1980. Air–Sea Interaction, Instruments and
Methods. Plenum Press, New York.