sea surface. Therefore, there is no geostrophic
movement of water associated with the ocean’s
response. If not corrected for such static response,
sea-level data would be very difficult to analyse
for studying the variability caused by ocean circulation. The variability of atmospheric pressure
has a global rms magnitude of about 7 mbar with
a high degree of spatial variability. The resulting
inverted barometer (IB) sea-level variations can
be as large as 15 cm (rms) in the Southern Ocean
where the atmospheric pressure variability is
the largest. The ‘inverted barometer’ correction is
thus a very important issue in analysing altimeter
data.
How well can the ocean’s response be represented by the IB approximation? This subject has
recently been reviewed by Wunsch and Stammer
(1997). Their conclusion is that an IB response is
expected at all frequencies and wavenumbers
except those falling on certain dispersion curves,
where dynamic response occurs. The dynamic
response assumes various forms of gravity and
Rossby waves. Resonance is possible depending on
the reflecting properties of the ocean bottom and
specific geometry of the ocean basin. Analysis of
the T/P altimeter data has indeed shown a near IB
response in most of the non-tropical regions. Fu
and Pihos (1994) performed a regression analysis
between T/P sea-level anomalies (computed as
deviations from the time mean) and pressure fluctuations. They found that the sea-level response
over regions poleward of 30° latitude was weaker
than IB with a regression coefficient of 90.84<
0.29 (1 standard deviation) cm mbar
91 . (The
minus sign indicates an inverse relationship.) They
also performed a multivariate regression analysis
to remove the effects of wind forcing, which is correlated to pressure. The result indicated a response
much closer to IB at 90.96<0.32 cm mbar
91
.
Their results in the tropics showed a complete
breakdown of the IB effect. However, after the
removal of the wind effects by using the simulation of a tropical ocean model driven by wind,
the sea-level response became IB-like even in the
tropics where the pressure forcing is particularly
weak.
Gaspar and Ponte (1997) and Ponte and Gaspar
(1999) used the T/P cross-over differences within
10-day repeat cycles to investigate the IB effects
at high frequencies (primarily at periods of 3–4
days). Outside the tropics, their results were
similar to those of Fu and Pihos (1994), whose
study, based on repeat-track analysis, did not filter
out any particular frequencies. The high-frequency
sea-level response is generally weaker than IB with
a regression coefficient about 90.9 cm mbar
91
.
The coefficient dropped to 90.7 cm mbar
91 in a
number of regions in the Southern Ocean. They
were able to simulate the observed relation
between sea level and pressure using a numerical
model of a homogeneous ocean driven by both
wind and pressure. They found that the extratropical sea-level response became much closer to IB if
the model was driven only by pressure. Most of
the apparent non-IB response in the extratropical
regions was due to the wind effects, consistent
with the multivariate regression analysis of Fu and
Pihos (1994) and the modelling study of Ponte
(1994). However, the pressure-driven simulations still revealed significant non-IB response
(ϳ90.9 cm mbar
91 ) in the Southern Ocean where
semi-closed f/H contours occurred and hence provided conditions favourable for resonant response.
Another extensive region of pressure-driven nonIB response (from 90.7 to 90.8 cm mbar
91 ) is in
the tropics, where the frequency band of free
waves is wider than that in the extratropics and is
thus more prone to resonant response. However,
the atmospheric pressure variability is very small
(rms value less than 2 mb) in the tropics and hence
the magnitude of the ocean’s non-IB response is
also extremely small there.
Steric sea-level variability and the heat budget
of the ocean
The effects of the ocean’s exchange of heat (heating/cooling) and fresh water (evaporation/precipitation) with the atmosphere change the density of
the ocean and hence the sea level (the ‘steric’
effect). The steric variability of sea level is to first
order due to the change of the heat content in the
ocean on seasonal time scales, with the effects of
freshwater exchange (via salinity) playing a secondary role (Gill and Niiler, 1973). Therefore sealevel variability has been used to infer the heat
storage in the upper ocean and the rate of heat
exchange between the ocean and the atmosphere.
Leuliette and Wahr (1999) showed global patterns
of the thermal effects on sea level using a coupledpattern analysis of simultaneous altimeter data
and sea-surface-temperature data. However, the
3.3 Ocean Circulation and Variability from Satellite Altimetry
149
Fu
movement of water associated with the ocean’s
response. If not corrected for such static response,
sea-level data would be very difficult to analyse
for studying the variability caused by ocean circulation. The variability of atmospheric pressure
has a global rms magnitude of about 7 mbar with
a high degree of spatial variability. The resulting
inverted barometer (IB) sea-level variations can
be as large as 15 cm (rms) in the Southern Ocean
where the atmospheric pressure variability is
the largest. The ‘inverted barometer’ correction is
thus a very important issue in analysing altimeter
data.
How well can the ocean’s response be represented by the IB approximation? This subject has
recently been reviewed by Wunsch and Stammer
(1997). Their conclusion is that an IB response is
expected at all frequencies and wavenumbers
except those falling on certain dispersion curves,
where dynamic response occurs. The dynamic
response assumes various forms of gravity and
Rossby waves. Resonance is possible depending on
the reflecting properties of the ocean bottom and
specific geometry of the ocean basin. Analysis of
the T/P altimeter data has indeed shown a near IB
response in most of the non-tropical regions. Fu
and Pihos (1994) performed a regression analysis
between T/P sea-level anomalies (computed as
deviations from the time mean) and pressure fluctuations. They found that the sea-level response
over regions poleward of 30° latitude was weaker
than IB with a regression coefficient of 90.84<
0.29 (1 standard deviation) cm mbar
91 . (The
minus sign indicates an inverse relationship.) They
also performed a multivariate regression analysis
to remove the effects of wind forcing, which is correlated to pressure. The result indicated a response
much closer to IB at 90.96<0.32 cm mbar
91
.
Their results in the tropics showed a complete
breakdown of the IB effect. However, after the
removal of the wind effects by using the simulation of a tropical ocean model driven by wind,
the sea-level response became IB-like even in the
tropics where the pressure forcing is particularly
weak.
Gaspar and Ponte (1997) and Ponte and Gaspar
(1999) used the T/P cross-over differences within
10-day repeat cycles to investigate the IB effects
at high frequencies (primarily at periods of 3–4
days). Outside the tropics, their results were
similar to those of Fu and Pihos (1994), whose
study, based on repeat-track analysis, did not filter
out any particular frequencies. The high-frequency
sea-level response is generally weaker than IB with
a regression coefficient about 90.9 cm mbar
91
.
The coefficient dropped to 90.7 cm mbar
91 in a
number of regions in the Southern Ocean. They
were able to simulate the observed relation
between sea level and pressure using a numerical
model of a homogeneous ocean driven by both
wind and pressure. They found that the extratropical sea-level response became much closer to IB if
the model was driven only by pressure. Most of
the apparent non-IB response in the extratropical
regions was due to the wind effects, consistent
with the multivariate regression analysis of Fu and
Pihos (1994) and the modelling study of Ponte
(1994). However, the pressure-driven simulations still revealed significant non-IB response
(ϳ90.9 cm mbar
91 ) in the Southern Ocean where
semi-closed f/H contours occurred and hence provided conditions favourable for resonant response.
Another extensive region of pressure-driven nonIB response (from 90.7 to 90.8 cm mbar
91 ) is in
the tropics, where the frequency band of free
waves is wider than that in the extratropics and is
thus more prone to resonant response. However,
the atmospheric pressure variability is very small
(rms value less than 2 mb) in the tropics and hence
the magnitude of the ocean’s non-IB response is
also extremely small there.
Steric sea-level variability and the heat budget
of the ocean
The effects of the ocean’s exchange of heat (heating/cooling) and fresh water (evaporation/precipitation) with the atmosphere change the density of
the ocean and hence the sea level (the ‘steric’
effect). The steric variability of sea level is to first
order due to the change of the heat content in the
ocean on seasonal time scales, with the effects of
freshwater exchange (via salinity) playing a secondary role (Gill and Niiler, 1973). Therefore sealevel variability has been used to infer the heat
storage in the upper ocean and the rate of heat
exchange between the ocean and the atmosphere.
Leuliette and Wahr (1999) showed global patterns
of the thermal effects on sea level using a coupledpattern analysis of simultaneous altimeter data
and sea-surface-temperature data. However, the
3.3 Ocean Circulation and Variability from Satellite Altimetry
149
Fu