observations shown in Fig. 5.7.4 do not seem to
support this argument. Originally the calculated
barotropic or total transport showed a mean value
of 0.086 Sv, which is about twice of the expected
value based on estimates of net evaporation over
the Mediterranean Basin (Garrett et al., 1993b).
This pointed to uncertainties in the estimates of
the upper-layer inflow, where the cross-section is
wider and transport estimates based on only midstrait current measurements were considered unreliable. Therefore currents in the upper 100 m
were reduced by 9% in order to have a mean total
transport value of 0.04 Sv over the 2 years of
observations, in accordance with historical net
evaporation estimates. Apart from this mean correction the seasonal cycle present in the total flow
has the correct phase to be the principal contribution to the observed seasonal sea-level rise within
the Sea, although to explain fully the observed sealevel change one has to take into account also the
effects due to evaporation and seasonal changes in
heat content. Evaporation and heat loss from the
Mediterranean also have seasonal cycles, with
maximum loss of both heat and moisture during
winter; heat is actually gained from the atmosphere by the Mediterranean during April–August
(Bunker, 1976b). The effect of seasonal air–sea
forcing over the Mediterranean on the instantaneous transport through the strait, however, may
be quite small, as the residence time involved in
the transformation of Atlantic Water to Mediterranean Water is on the order of decades (Lacombe
et al., 1981). It is more likely that seasonal fluctuations of transport result from dynamic effects
more local to the strait, or from mechanisms of
draining and filling of the LIW and/or WMDW
reservoirs, as suggested by Bormans et al. (1986).
At periodicities of days to months there is a clear
increase of variability of the exchange in both
layers during the late fall and winter periods
(Astraldi et al., 1999).
At interannual time scales, the formation of
Mediterranean Water masses may occur under
different conditions in different years, changing
the characteristic temperatures and salinities of the
outflow (Lacombe et al., 1985). As a result, different volumes of outflow in different years might
transport the same flux properties. This variability
in the characteristics of the outflow is generally
thought to be a small effect, but investigators have
noted a wide range in the maximum of salinity
observed west of the sill: Schott (1928) found a
maximum of 37.25‰ over several years of observations, Boyum (1963) a maximum of 38.2‰, and
Kinder and Parrilla (1987) a maximum of 38.44‰.
Based on hydrographic sections in the Gulf of
Cadiz, Ochoa and Bray (1991) estimated an equivalent transport of ‘pure’ Mediterranean Water of
0.7 Sv, assuming salinities of 38.4‰ and 35.6‰
for Mediterranean and Atlantic Waters. They
also estimated a freshwater flux equivalent to
0.53 m yr
91 of evaporation excess over the Mediterranean and a heat flux higher in the autumn of
1986 than in the spring (6.0 W m
92 versus
2.2 W m
92
), due to warmer surface temperatures
in the autumn. Salt fluxes from our observations,
computed by the same method as Bryden et al.
(1994), show a mean value of 1.3 Sv ‰ with a
large standard deviation of 0.4 Sv ‰ in the 2 years,
which implies a mean net evaporation over the
Mediterranean Sea of 0.45 m yr
91 , when mass and
salt conservation equations for the whole Mediterranean Basin are taken into account. This
net evaporation is similar to previous estimates
(Bryden et al., 1994), but shows appreciable interannual differences of 0.15 m yr
91
. Temperature
fluxes, computed as in Macdonald et al. (1994),
give a mean inward flux of 3 Sv °C, with a considerable interannual difference of 1 Sv °C. This mean
temperature flux through Gibraltar implies a mean
heat loss through the Mediterranean Sea surface of
about 3 W m
92
, considering that the estimated
0.04 Sv of net inward flow leaves the surface of the
Mediterranean at 25°C. This heat loss is a factor
of two smaller than the 6 W m
92 obtained by
Macdonald et al. (1994) based on direct current
and temperature measurements done in the strait
between October 1985 and October 1986. However, in contrast, climatological estimates of the
basin mean heat flux based on sea surface measurements have typically shown a heat gain by the
Sea of 20–30 W m
92 (Garrett et al., 1993b; Josey
et al., 1997). Further research is required to
resolve this discrepancy. Although both observed
salt and heat flux interannual differences seem to
be small with respect to their effect on the North
Atlantic water mass structure, it is still not possible to conclude that they are negligible.
From a climatic point of view it is important to
know the mean values of the inflow and outflow
as accurately as possible as well as any interannual variabilities. Recent numerical model studies
5.7 Mediterranean Water and Global Circulation
425
Candela
support this argument. Originally the calculated
barotropic or total transport showed a mean value
of 0.086 Sv, which is about twice of the expected
value based on estimates of net evaporation over
the Mediterranean Basin (Garrett et al., 1993b).
This pointed to uncertainties in the estimates of
the upper-layer inflow, where the cross-section is
wider and transport estimates based on only midstrait current measurements were considered unreliable. Therefore currents in the upper 100 m
were reduced by 9% in order to have a mean total
transport value of 0.04 Sv over the 2 years of
observations, in accordance with historical net
evaporation estimates. Apart from this mean correction the seasonal cycle present in the total flow
has the correct phase to be the principal contribution to the observed seasonal sea-level rise within
the Sea, although to explain fully the observed sealevel change one has to take into account also the
effects due to evaporation and seasonal changes in
heat content. Evaporation and heat loss from the
Mediterranean also have seasonal cycles, with
maximum loss of both heat and moisture during
winter; heat is actually gained from the atmosphere by the Mediterranean during April–August
(Bunker, 1976b). The effect of seasonal air–sea
forcing over the Mediterranean on the instantaneous transport through the strait, however, may
be quite small, as the residence time involved in
the transformation of Atlantic Water to Mediterranean Water is on the order of decades (Lacombe
et al., 1981). It is more likely that seasonal fluctuations of transport result from dynamic effects
more local to the strait, or from mechanisms of
draining and filling of the LIW and/or WMDW
reservoirs, as suggested by Bormans et al. (1986).
At periodicities of days to months there is a clear
increase of variability of the exchange in both
layers during the late fall and winter periods
(Astraldi et al., 1999).
At interannual time scales, the formation of
Mediterranean Water masses may occur under
different conditions in different years, changing
the characteristic temperatures and salinities of the
outflow (Lacombe et al., 1985). As a result, different volumes of outflow in different years might
transport the same flux properties. This variability
in the characteristics of the outflow is generally
thought to be a small effect, but investigators have
noted a wide range in the maximum of salinity
observed west of the sill: Schott (1928) found a
maximum of 37.25‰ over several years of observations, Boyum (1963) a maximum of 38.2‰, and
Kinder and Parrilla (1987) a maximum of 38.44‰.
Based on hydrographic sections in the Gulf of
Cadiz, Ochoa and Bray (1991) estimated an equivalent transport of ‘pure’ Mediterranean Water of
0.7 Sv, assuming salinities of 38.4‰ and 35.6‰
for Mediterranean and Atlantic Waters. They
also estimated a freshwater flux equivalent to
0.53 m yr
91 of evaporation excess over the Mediterranean and a heat flux higher in the autumn of
1986 than in the spring (6.0 W m
92 versus
2.2 W m
92
), due to warmer surface temperatures
in the autumn. Salt fluxes from our observations,
computed by the same method as Bryden et al.
(1994), show a mean value of 1.3 Sv ‰ with a
large standard deviation of 0.4 Sv ‰ in the 2 years,
which implies a mean net evaporation over the
Mediterranean Sea of 0.45 m yr
91 , when mass and
salt conservation equations for the whole Mediterranean Basin are taken into account. This
net evaporation is similar to previous estimates
(Bryden et al., 1994), but shows appreciable interannual differences of 0.15 m yr
91
. Temperature
fluxes, computed as in Macdonald et al. (1994),
give a mean inward flux of 3 Sv °C, with a considerable interannual difference of 1 Sv °C. This mean
temperature flux through Gibraltar implies a mean
heat loss through the Mediterranean Sea surface of
about 3 W m
92
, considering that the estimated
0.04 Sv of net inward flow leaves the surface of the
Mediterranean at 25°C. This heat loss is a factor
of two smaller than the 6 W m
92 obtained by
Macdonald et al. (1994) based on direct current
and temperature measurements done in the strait
between October 1985 and October 1986. However, in contrast, climatological estimates of the
basin mean heat flux based on sea surface measurements have typically shown a heat gain by the
Sea of 20–30 W m
92 (Garrett et al., 1993b; Josey
et al., 1997). Further research is required to
resolve this discrepancy. Although both observed
salt and heat flux interannual differences seem to
be small with respect to their effect on the North
Atlantic water mass structure, it is still not possible to conclude that they are negligible.
From a climatic point of view it is important to
know the mean values of the inflow and outflow
as accurately as possible as well as any interannual variabilities. Recent numerical model studies
5.7 Mediterranean Water and Global Circulation
425
Candela