100 AIR–SEA INTERACTION / Sea Surface Temperature

To take an example, in calculating the flux maps accurate routine source of air–sea flux estimates will
shown in Figure 4 many corrections were applied to be from numerical models of the coupled ocean–
the VOS observations in an attempt to remove biases atmosphere system.
caused by the observing methods. For example, air
temperature measurements were corrected for the
‘heat island’ caused by the ship heating up in sunny,
See also
low-wind conditions. The wind speeds were adjusted Aerosols: Observations and Measurements; Role in Ra-
depending on the anemometer heights on different diative Transfer. Air–Sea Interaction: Freshwater Flux;
ships. Corrections were applied to sea temperatures Sea Surface Temperature; Surface Waves. Boundary
calculated from engine room intake data. Despite Layers: Observational Techniques In Situ; Observational
Techniques–remote; Surface Layer. Buoyancy and
these and other corrections, the global annual mean
Buoyancy Waves: Optical Observations. Climate Var-
flux showed about 30 W m
2 excess heating of the
iability: North Atlantic and Arctic Oscillation. Coupled
ocean. Previous climatologies calculated from ship Ocean–Atmosphere Models. El Niño and the South-
data had shown similar biases and the fluxes had been ern Oscillation: Observation. Reflectance and Albedo,
adjusted to remove the bias, or to make the fluxes Surface. Weather Prediction: Regional Prediction
compatible with estimates of the meridional heat Models.
transport in the ocean. However, comparison of the
unadjusted flux data with accurate data from air–sea
interaction buoys showed good agreement between Further Reading
the two. This suggests that adjusting the fluxes Browning KA and Gurney RJ (eds) (1999) Global Energy
globally is not correct and that regional flux adjust- and Water Cycles. Cambridge: Cambridge University
ments are required; however, the exact form of these Press.
corrections is presently not known. Dobson F, Hasse L and Davis R (eds) (1980) Air–Sea
In the future, computer models are expected to Interaction, Instruments and Methods. New York: Ple-
provide a major advance in flux estimation, Recently, num Press.
coupled numerical models of the ocean and of the Garratt JR (1992) The Atmospheric Boundary Layer.
Cambridge: Cambridge University Press.
atmosphere have been run for many simulated years,
Geernaert GL and Plant WJ (1990) Surface Waves and
during which the modeled climate has not drifted. This
Fluxes, vol. 1, Current Theory. Dordrecht: Kluwer
suggests that the air–sea fluxes calculated by the Academic.
models are in balance with the simulated oceanic and Isemer H-J and Hasse L (1987) The Bunker Climate Atlas of
atmospheric heat transports. However, it does not the North Atlantic Ocean, vol. 2, Air–Sea Interactions.
imply that at present the flux values are realistic. Berlin: Springer-Verlag.
Errors in the short-wave and latent heat fluxes may Josey SA, Kent EC and Taylor PK (1999) The Southampton
compensate one another; indeed, in a typical simula- Oceanography Centre (SOC) Ocean–Atmosphere Heat,
tion the sea surface temperature stabilized to a value Momentum and Freshwater Flux Atlas, SOC Report No.
that was, over large regions of the ocean, a few degrees 6, 30 pp.1figs. (Available from The Library, Southamp-
different from that which is observed. Nevertheless, ton Oceanography Centre, European Way, Southamp-
ton, SO14 3ZH, UK.)
the estimation of flux values using climate or NWP
Kraus EB and Businger JA (1994) Atmosphere–Ocean
models is a rapidly developing field and improvements
Interaction, 2nd edn. New York: Oxford University
will doubtless have occurred by the time this article has Press.
been published. There will be a continued need for in- Stull RB (1988) An Introduction to Boundary Layer
situ and satellite data for assimilation into the models Meteorology. Dordrecht: Kluwer Academic.
and for model development and verification. How- Wells N (1997). The Atmosphere and Ocean: A Physical
ever, it seems very likely that in future the most Introduction, 2nd edn. London: Taylor and Francis.

Sea Surface Temperature

W J Emery, University of Colorado, Boulder, CO, USA Introduction
As the controlling variable of heat, momentum, salt,
Copyright 2003 Elsevier Science Ltd. All Rights Reserved. and gas fluxes between the ocean and the atmosphere,
 AIR–SEA INTERACTION / Sea Surface Temperature 101

the sea surface temperature (SST) has always been a position of the Gulf Stream from SST measurements
topic of interest to scientists. It is also the easiest made from the mail packet ships that he rode to and
oceanographic parameter to observe and observations from Europe. Fortunately for Franklin, the Gulf
of some form of SST extend back to the time of the Stream has a very sharp thermal contrast on its
early Greek scientists. In addition to its relationship to western edge where the cold shelf waters coming
ocean–atmosphere fluxes, the SST also relates directly down from the north meet the warm Gulf Stream
to a number of human concerns. For example, the waters advecting to the north east. Together with
success of fisheries and fishermen can be enhanced by a Folger, he published a map (Figure 1) that described
knowledge of the SST pattern. For many years SSTwas the position of the Gulf Stream, advising ship captains
measured by taking a ‘bucket sample’ of the surface not to sail against this current, which was strong
waters and then measuring its temperature. As ships enough to hold a sailing vessel still. This practical
evolved to powered vessels, this practice of SST bucket application of SST made it possible to reduce the
sampling had to be abandoned and the practice of crossing times for ships sailing from the Americas to
using a measurement of the temperature of the cooling the continent.
water coming to the ship’s engines was used. This was As one of the easiest measurements in oceano-
known as ‘ship injection temperatures’ (the SST sensor graphy, SST became widely observed whether as a
was ‘injected’ into the water stream), and these data sample taken from a pier or beach or as a bucket
suffer from many basic problems (engine room heat- sample from a ship on the open ocean. The collection
ing, depth of the water intake, etc.). The advent of of a bucket of water whose temperature was measured
satellite-tracked drifting buoys introduced a platform as an estimate of SST became common practice among
that could measure the SST and then report it in near the sailing ships carrying the world’s commerce. When
real time. Thought to be less noisy and more accurate the ships’ logs became the source of global information
than ship SSTs, the drifting buoy SST data became the on winds and currents, the SST information was also
standard for both the calibration and validation of compiled from the ships’ logs. This information was
satellite infrared estimates of SST. Over the years these routinely published, along with the currents and winds
buoy SSTs have been used to calculate the infrared from the ships’ logs, as part of ‘sailing directions’ put
algorithm coefficients for the computation of the SST. out first by the US Navy and later by the Coast Guard.
The problem is that a buoy cannot measure the The same practice was introduced later in Europe and
temperature of the 10 mm thin ‘skin of the ocean’, became an operational reality for most ocean-going
which is the layer that radiates out into space. Thus, vessels.
satellite infrared measurements of SST are of this skin The collection of a bucket sample from sailing ships
temperature and not of the deeper bulk SST measured that travel at speeds between 5 and 15 knots was easy.
by the ships and buoys. It is the difference between the For powered ships traveling along at 20–30 knots it was
skin and bulk temperature that is directly related to the no longer possible to collect a bucket sample for SST
wind speed and the net air–sea heat flux. In the past the measurement. Instead, the temperature of the cooling
bulk SST has been used in the computation of air–sea water used to cool the ship’s engines was measured as
heat flux terms using empirical ‘bulk’ formulas. These the SST. This temperature was called the ‘injection’
continue to be used, but research efforts are underway temperature because the sensor was ‘injected’ into the
to transform these computations into a satellite-only pipe carrying this cooling water to the heat exchangers.
calculation. These efforts will explicitly involve skin One of the fundamental problems with these injection
and bulk SST, giving a more physical basis to the temperatures was the location of the inlet pipe for this
connections between heat flux and the skin and bulk cooling water, which was positioned far down on the
SSTs. These improvements are critical for the mode- ship’s hull, generally collecting water from about 5 m
ling of climate change, since the SST is the boundary depth – far away from the sea surface. In addition, the
that connects the model atmosphere to the model heat in the engine room was known to heat the cooling
ocean. Any incorrect specification of the global SST water and the associated thermometer, resulting in a
patterns will lead directly to errors in the model warm bias for the SSTs. Other sources of error were the
simulations. reading of the analog gauge, the hand recording of the
data in the ship’s log and the radio broadcast of the
‘SST’ to be included in the SST database.
History of SST Measurements and
Applications Satellite SST
One of the earliest uses of sea surface temperature In spite of all these limitations, the ship SSTs were
(SST) was when Benjamin Franklin mapped the initially used to match and adjust infrared satellite
102 AIR–SEA INTERACTION / Sea Surface Temperature

Figure 1 The Poupard version of the Franklin–Folger Gulf Stream chart published in Philadelphia in 1786 with Franklin’s ‘Maritime
Observations’.

measurements of the SST. Later, satellite-tracked ficients from regression with nearly coincident buoy
drifting buoys were equipped with temperature ports SST data. In this approach one not only assumes that
sticking out from the hull to measure the SST as the the satellite and buoys measure the same SST, but also
buoy traveled around the ocean. Most of these buoy that the buoy SSTs have no errors themselves. That is
SST sensors were initially calibrated to 70.11C, but not to say that people assume the buoy SSTs have no
since the buoys are considered expendable there is no error, but the practice of using them to find the
post-deployment calibration and it is not known how algorithm coefficients through linear regression im-
well the buoys retain their calibration. There are about plies that the buoy SSTs are error-free. It appears that
three different hull types used today in these buoys the source of this conflict is the fact the satellite
with slightly different configurations of the hull SST infrared systems and their calibration systems do not
sensors. All of these buoys float in the active near- retain their calibrations over long periods of time and
surface layer and move up and down with the wave that regression to in situ SST is required to overcome
field. As a consequence, the buoy SST represents the these drifts. Since buoy SSTs are the best of the present
temperature between the surface and 1–2 m depth. in situ SSTs, they are used to supply the in situ SST
The best estimate of the accuracy of these data can be observations in spite of the fact that they do not and
made by considering the mean difference and varia- cannot measure the skin SST, as any direct physical
bility of the difference between contemporary SST contact with the skin layer will temporarily disrupt it.
measurements. This value is 0.41C, which can be taken Observations of the effects of breaking waves on the
as an overall error limit for buoy-measured bulk SST. presence of the skin layer have shown that while a
It is important to realize that past practice has been breaking wave does indeed destroy the skin layer, the
to compute the satellite infrared SST algorithm coef- skin forms again after just a few seconds. So, in spite of
 AIR–SEA INTERACTION / Sea Surface Temperature 103

strong winds and breaking waves, the skin SST layer is eliminate obvious outliers. The input buoy and ship
present most of the time and must be considered in the SST data sets were also filtered to remove large errors.
remote sensing of SST. Before various groups and All three of these input data sets were to save space. We
agencies will transform their computations to skin only present four images to represent the 12 month
SST, there must be a source of in situ skin SST to annual cycle. The first is a global map of SST for the
replace the buoy SSTs that are used at present. We will month of January (Figure 2), which represents SST
discuss this point further later in this article. conditions for the Northern Hemisphere winter.
In this map a number of basic features are readily
obvious. The warmest temperatures are in the tropics,
particularly in the Pacific and Indian Oceans. The
SST Climatology warmest temperatures are found in the ‘warm pool’ of
One of the benefits of being relatively easy to observe is the western tropical Pacific. There are cold features
the abundance of data that exists both geographically along the west coasts of South America and South
and over time. This excellent data coverage makes it Africa, which correspond to upwelling events most
possible to compute a climatology of SST conditions, active in the austral summer. Cold temperatures
which is a long-term average map of SST conditions extend a bit farther north in the Southern Ocean
usually over the 12 month period of an annual cycle. than they do in the Northern Hemisphere, which is
Many different climatologies have been computed for primarily a consequence of the open character of the
SST, starting with strictly ship measurements before Southern Ocean compared with the geographically
about 1970 and a mix of satellite, ship, and buoy SSTs restricted waters of the Arctic.
since that time. One must always be careful to The representative Northern Hemisphere spring
determine exactly which periods are covered by such SST map in Figure 3 shows a modest shift in this
a climatology and what data went in to the compu- temperature distribution. The warmest temperatures
tation of the climatology. have increased, particularly in the equatorial Indian
For this article we will use a climatology that Ocean, which has increased from about 281C to about
includes satellite infrared, buoy, and ship SST data. As 301C. The Western Pacific Warm Pool has expanded
part of this analysis, satellite SSTs were filtered to slightly and increased in temperature. There is now a

90° N

60° N

30° N
Latitude

0°

30° S

60° S

90° S
180° W 150° W 120° W 90° W 60° W 30° W 0° 30° E 60° E 90° E 120° E 150° E 180°
Longitude

0°C 4°C 8°C 12°C 16°C 20°C 24°C 28°C 32°C

Sea surface temperature

Figure 2 SST climatology for January. (Reproduced with permission from Reynolds and Smith, 1995.)
104 AIR–SEA INTERACTION / Sea Surface Temperature

90° N

60° N

30° N
Latitude

0°

30° S

60° S

90° S
180° W 150° W 120° W 90° W 60° W 30° W 0° 30° E 60° E 90° E 120° E 150° E 180°
Longitude

0°C 4°C 8°C 12°C 16°C 20°C 24°C 28°C 32°C

Sea surface temperature

Figure 3 April SST climatology. (Reproduced with permission from Reynolds and Smith, 1995.)

distinct band of warm temperatures in the equatorial to the west. The equatorial warm regions have
Pacific. The upwelling zones off south–western South weakened in magnitude, while they are about the
America and South Africa have decreased in size. The same in geographic coverage as they were in July. The
colder SSTs have remained largely the same. colder waters have not changed much since the last
Turning to the Northern Hemisphere summer, we season (summer).
look at the July SST climatology in Figure 4. In this All of these maps have used a mixture of satellite
case the maximum temperatures in the tropics have infrared measurements, drifting and moored buoy SST
actually reduced slightly, reflecting the decrease in measurements, and ship SST measurements. In this
solar insolation on a global level. The warm band in application the satellite skin SST is adjusted to match
the tropical Atlantic has weakened, as has the equa- coincident drifting buoy SSTs, as introduced earlier.
torial warm band in the tropical Indian Ocean. Even Thus, these maps really represent a ‘pseudo bulk SST’
the warm pool in the western tropical Pacific has due to the overwhelming number of infrared SST
weakened in both magnitude and areal coverage. The observations as compared to the in situ buoy and ship
upwelling zones off western South America and observations. Still, the general seasonal pattern of the
western South Africa have again expanded, and both SST is clearly apparent, and it is not likely that an
show warm tongues that extend westward out from adjustment to skin SST would show any substantially
the northernmost extent of the colder upwelling water. different SST patterns. SST patterns computed from
There is a corresponding cold upwelling region off various satellite SST algorithms all look very similar; it
North America consistent with the seasonal shift to is the absolute temperature value that is different and
northerly upwelling winds off that coast. The same is the skin SST must be considered when addressing
true off north–west Africa. questions such as air–sea heat and gas exchange.
To complete the cycle we look at the Northern
Hemisphere fall (October) map in Figure 5. In this map
all of the west coast upwelling regions have weakened
slightly. In particular, the north–west Africa upwelling
Skin SST
region has disappeared. In the Southern Hemisphere Due to its very high emissivity, the ocean is consi-
the upwelling regions appear to have stretched farther dered to very nearly approximate a ‘blackbody’. This
 AIR–SEA INTERACTION / Sea Surface Temperature 105

90° N

60° N

30° N
Latitude

0°

30° S

60° S

90° S
180° W 150° W 120° W 90° W 60° W 30° W 0° 30° E 60° E 90° E 120° E 150° E 180°
Longitude

0°C 4°C 8°C 12°C 16°C 20°C 24°C 28°C 32°C

Sea surface temperature

Figure 4 Mean July SST. (Reproduced with permission from Reynolds and Smith, 1995.)

90° N

60° N

30° N
Latitude

0°

30° S

60° S

90° S
180° W 150° W 120° W 90° W 60° W 30° W 0° 30° E 60° E 90° E 120° E 150° E 180°
Longitude

0°C 4°C 8°C 12°C 16°C 20°C 24°C 28°C 32°C

Sea surface temperature

Figure 5 Mean October SST. (Reproduced with permission from Reynolds and Smith, 1995.)
106 AIR–SEA INTERACTION / Sea Surface Temperature

long-wave heat emission is directly proportional to the T T

0
skin SST, which is the only SST that interacts with the
overlying atmosphere. Having a thickness of between
5 and 10 mm, the skin of the ocean can easily be 1 mm
destroyed by breaking ocean waves. When this hap- 1a 1b
pens the skin reforms within 3–6 s, which means that
the skin of the ocean is most generally present. This 1 cm
skin layer is the molecular boundary between the DAY
turbulent atmosphere and the turbulent ocean. It is 1 dm
this skin layer that transfers heat, gases, and momen-
tum between the atmosphere and the ocean. The
temperature of this ultra–thin layer can only be 1m
measured radiometrically, since any contact with the
skin layer will disturb it. Thus, this layer cannot be
10 m
measured directly by drifting or moored buoys or by
any ship. Equipped with infrared radiometers, ships
could measure this skin layer temperature radiomet-
rically without the attenuating atmosphere in between ∆T < 0 ∆T > 0
z
the ship radiometer and the sea surface. Such meas-
urements could provide validation and possibly cali-
Figure 6 Temperature profile in the shallow upper layer of the
bration information for satellite-based infrared
ocean.
radiometric measurements of skin SST.
An important question is: ‘What is the precise
from the traditional bulk formulas using routine
definition of bulk SST’? Drifting buoys measure a
meteorological measurements also made from the
temperature somewhere between 0.5 and 1.5 m below
ship. At the top is a series from a cruise on the FS
the sea surface. This is a result of the buoy’s interaction
Meteor in the North Atlantic. Here, interestingly
with the waves at the sea surface. As described earlier,
ship cooling water intake ports may be located Observed night time ∆T from F/S meteor
anywhere from 2 to 5 m beneath the waterline. So is
the bulk SST located at 2–5 m beneath the surface? If 0.60
we consider the temperature profile in the shallow 0.50
upper layer of the ocean (Figure 6), we can see that the 0.40
∆T (K)

skin SST is always slightly colder than the temperature 0.30

just below it. Note the logarithmic abscissa in this plot. 0.20
At night this temperature slightly below the cooling 0.10
skin layer is isothermal down to a few meters. In this
0.00
case the bulk temperature could be measured at any of 0
40 00 10
the depths down to 5 m. During the day, however, solar He 3 200 6
8
insolation can heat up a shallow layer sufficiently so at f
lux 00 2
4 _1
Wind (m s )
(W 1 _
0

that the temperature of the ‘cool skin’ is a little higher m 1

)
than the isothermal layer below the warm diurnal
layer. Often referred to as ‘warm skin SST’, this 0.60 Observed night time ∆T from CEPEX
condition can only exist during daytime under rela- 0.50
tively clear sky conditions when the shallow surface 0.40
∆T (K)

layer is heating. 0.30

The relationship between the skin and the bulk 0.20
temperatures depends on two forcing factors: the wind 0.10
and the net air–sea heat flux. A good example of this is
0.00
shown in Figure 7, which is taken from two different 0
40 00 10
oceanographic research cruises. Here we have plotted He 3 0 6
8
at f 20 4
the difference between the skin SST and the bulk lux 00 2
_1
Wind (m s )
(W 1 _
0

temperature (taken as a temperature between 2 and m 1

)
5 m) as a function of wind speed and net air–sea heat
flux. The wind speed was observed from the research Figure 7 Skin and bulk temperatures as a function of wind speed
vessel while the net air–sea heat flux was calculated and net air–sea heat flux.
 AIR–SEA INTERACTION / Sea Surface Temperature 107

enough, the DT between the skin SST and the bulk sea surface. It is likely that these effects are smaller in
temperature both rises and falls with increasing wind magnitude than those driven by the wind and net heat
speed. At low heat flux levels DT decreases as wind flux. Much more sophisticated models will be required
speed increases. This is consistent with traditional to include these phenomena in modeling DT.
wisdom that increased mixing due to wind stirring will
homogenize the upper layer, eventually making skin
and bulk temperatures the same. At higher heat flux In situ Measurement of Skin SST
values, however, DT actually increases as the wind Since infrared satellite sensors are only able to measure
speed increases. In the lower figure a data set was radiation from the skin of the ocean, the challenge is to
collected from the tropical Pacific. A much more provide in situ measurements of the skin SST that can
limited change in DT is found for this tropical sample, be used to calibrate/validate the satellite radiances in
but there is still a decrease in DT with increasing wind. terms of temperature. As mentioned earlier, ships can
These effects can be modeled with a semiempirical be equipped with radiometers to measure directly the
formula that includes the effects of wind and net heat skin SST without the atmosphere attenuating the
flux on the DT difference between skin SST and bulk infrared signal. In principle these same radiometers
temperature: could be installed on moored buoys to continuously
(
 measure skin SST. The problem with both of these
QN nZ0 1=2 installations is that the radiometer optics must be
DT ¼ Cshear n3
rw cp k1=2 u protected from sea spray, which is difficult to do in an
"
1=2 autonomous installation. At least on the ship the
nrw cp
þ Cconv radiometer can be examined each day and cleaned off
agQN
to maintain a clear optical path.

 # )1=2 Another requirement of these radiometers is that
nZ0 1=2

Cshear n3 e
Rfcr =Rf0 ½1 they are very well calibrated. The best approach to this
u

In eqn [1], QN is the net heat flux, rw is the density of

water, cp is the specific heat capacity of sea water at a
constant pressure, k is the thermal diffusivity of water,
Cshear is the proportionality constant for shear-driven
time scale, Cconv is the proportionality constant
for convective driven time scale, n is the kinematic
viscosity, z0 is the momentum roughness length, un is
the friction velocity of water, g is the acceleration due
to gravity, a is the thermal expansion coefficient, Rf0 is
the surface Richardson number, and Rfcr is the critical
Richardson number.
This equation connects the skin SST with the
temperature just below it, but does not account for
the further transfer of heat downward or upward
within the water column. To do this a ‘mixed-layer’
model must be added to the skin–bulk parametriza-
tion. Such a model has been employed to show the
development of the full upper layer temperature
profile. Using this model combination it was possible
to trace the temperature history of the upper layer for a
couple of consecutive days of field measurements.
There is excellent agreement between the observed
bulk–skin ðDTÞ temperature difference observed and
that modeled using the wind speed and net air–sea heat
flux. Thus, it appears that this combination of models
can go a long way to simulating the temperature
behavior of the upper layer of the ocean. It should be
noted that these models do not account for surface
wave effects, or the effects of turbidity and foam on the Figure 8 Radiometer requirements.
108 AIR–SEA INTERACTION / Sea Surface Temperature

Stddev = 8.45275
Mean = 18.7291
Num points = 1132447

Figure 9 Global merchant ship routes for 1996.

requirement is to equip the radiometers with two meters. It is clear that the need for continuing sampling
blackbodies with one at ambient temperature and the means that the ships must be merchant ships traveling
other heated a few degrees above ambient. To correct long routes on a regular basis. A plot of the present
for reflected infrared sky radiation, these radiometers ship coverage for a year (Figure 9) reveals that there
must look up and down to view the sky, the ocean, and are plenty of ships to choose from in the Northern
both blackbodies each scan. The easiest way to Hemisphere, but in the Southern Hemisphere the
implement these requirements is to use a rotating available selection is much more restrictive. Ship
mirror to channel the radiation to the detector, as seen tracks that make the long transits to Australia and
in Figure 8. In this setup the sensor is a low–cost New Zealand are good candidates, as are those routes
thermal infrared ‘thermometer’ with a supplemental from Asia to southern Chile. Only experience will
rotating mirror to collect radiation from the sky reveal just how many ships over which routes will be
through the top hole, the sea surface through the required to supply the in situ skin SSTs needed to
bottom hole, and both of the blackbodies. In this case calibrate/validate the satellite infrared radiances.
no special efforts have been taken to protect the mirror
from sea spray. The addition of a cover that would
move into place in cases of high seas and/or rainfall has
Summary and Conclusions
been considered. Other protection schemes use a SST is and has been one of the most measured variables
moving plastic wrap ‘window’ that would carry the in the ocean, and as such it has received much scientific
sea spray or rainfall away from the radiometer view. attention. Using a combination of ship SSTs, moored
One suggestion was to use ‘windshield wipers’ from and drifting buoy SSTs, and satellite infrared SSTs, we
automobile headlights. It will be a challenge to see have characterized the global and seasonal patterns of
which of these methods is successful in protecting the SST. Realizing that the satellite infrared SSTs can only
radiometer from sea spray and rainfall. represent the SST of the 10 mm thick skin layer of the
It is difficult to predict just how many ships need to ocean, future efforts should be aimed at separating the
be operating to provide the global coverage needed to satellite skin SSTs from the 1–5 m deep bulk SST
regularly calibrate and validate the satellite radio- measured by the buoys and ships. The skin layer is
 AIR–SEA INTERACTION / Storm Surges 109

simply the molecular layer that interfaces between a Emery WJ and Baldwin D, Schluessel P and Reynolds RE
turbulent ocean and a turbulent atmosphere. The (2001) Accuracy of in situ sea surface temperatures used
temperature difference between the skin and the bulk to calibrate infrared satellite measurements. Journal of
temperatures is directly related to the wind speed and Geophysical Research 2387–2405.
net air–sea heat flux. Thus, an improved understand- McClain EP, Pichel WG, Walton CC, Ahmad Z and Sutton J
(1983) Multi-channel improvements to satellite-derived
ing of this relationship can help us to better resolve the
global sea surface temperatures. Advances in Space
net heat and momentum fluxes between the ocean and
Research 2: 43–47.
the atmosphere. In part, this understanding and the McClain EP, Pichel WG and Walton CC (1985) Comparative
shift to the computation of skin SST from the satellite performance of AVHRR-based multichannel sea surface
infrared data depends on the creation of a network of temperatures. Journal of Geophysical Research 90:
ship-of-opportunity based skin SST radiometers col- 11587–11601.
lecting global and continuous samples of skin SST Reynolds RW and Smith TM (1994) Improved global sea
‘ground truth’ data (i.e., without an intervening surface temperature analysis using optimum interpola-
atmosphere). tion. Journal of Climate 7: 929–948.
Reynolds RWand Smith TM (1995) A high resolution global
sea surface temperature climatology. Journal of Climate
See also 8: 1571–1583.
Air–Sea Interaction: Freshwater Flux; Gas Exchange; Richardson PL (1979) The Benjamin Franklin and
Momentum, Heat and Vapor Fluxes; Storm Surges; Sur- Timothy Folger charts of the Gulf Stream. In: Sears M
face Waves. Global Change: Surface Temperature and Merriman D (eds) Oceanography the Past,
Trends. Observation Platforms: Buoys. Observations pp. 703–717. New York, Heidelberg, Berlin: Springer-
for Chemistry (Remote Sensing): IR/FIR. Satellite Verlag.
Remote Sensing: Temperature Soundings. Walton CC, Pichel WG, Sapper JF and May DA (1998) The
development and operational application of nonlinear
algorithms for the measurement of sea surface tempera-
Further Reading
tures with the NOAA polar-orbiting environmental
Emery WJ and Yu Y (1997) Satellite sea surface temperature satellites. Journal of Geophysical Research 103: 27999–
patterns and absolute values. International Journal of 28012.
Remote Sensing 18: 323–334. Wick GA (1995) Evaluation of the Variability and Predict-
Emery WJ, Baldwin DJ, Schluessel P and Reynolds RE ability of the Bulk–Skin Sea Surface Temperature
(2001) Accuracy of in situ sea surface temperatures used Difference with Application to Satellite-Measured
to calibrate infrared satellite measurements. Journal of Sea Surface Temperature. PhD thesis, University of
Geophysical Research 105: 2387–2405. Colorado, Boulder, CO.

Storm Surges
R A Flather, Proudman Oceanographic, Prenton, UK from a time series of sea levels recorded by a tide gauge
Copyright 2003 Elsevier Science Ltd. All Rights Reserved. using
surge residual ¼ ðobserved sea levelÞ

ðpredicted tide levelÞ ½1
Introduction and Definitions
Storm surges are changes in water level generated by producing a time series of surge elevations. Figure 1
atmospheric forcing, specifically by the drag of the shows an example.
wind on the sea surface and by variations in the surface Sometimes, the term ‘storm surge’ is used for the sea
atmospheric pressure associated with storms. Storm level (including the tidal component) during a storm
surges last for periods ranging from a few hours to event. It is important to be clear about the usage of the
2 or 3 days and have spatial scales that are large com- term and its significance to avoid confusion. Storms
pared with the water depth. They can raise or lower also generate surface wind waves that have periods of
the water level in extreme cases by several meters; order seconds and wavelengths, away from the coast,
a raising of level is a positive surge, a lowering a comparable to or less than the water depth.
negative one. Storm surges are superimposed on Positive storm surges combined with high tides and
the normal astronomical tides generated by variations wind waves can cause coastal floods which in terms of
in the gravitational attraction of the Moon and the loss of life and damage are probably the most
Sun. The storm surge component can be derived destructive natural hazards of geophysical origin.