156
F. Raicich
examined for interannual and interdecadal time scales. All the time series were
filtered to separate those time scales by means ofNg and Young's (1990) technique;
the cutoff period was set at about 10 years. However, since the SL is affected by a
"secular" trend, time scales longer than about 100 years have been preliminarily
removed from the time series. Due to the absence of information, the effects of
possible vertical land motion on the SL time series could not be checked and
corrected, therefore, in the discussion, it is assumed that SL changes are really
related to SL only, but this assumption has to be taken cautiously. For all the
parameters standardized monthly and seasonal anomalies were computed relative
to the period 1958-87, for which enough SL data is available for all locations to
establish a common climatological period.
3 Discussion
The undetrended SL time series show the well known increasing trend of about \1.5 cm/decade, but also reveal that during the last 30-40 years the tendency has
changed, with a SL decrease rate of o. 5 -1.5 cm/decade. By comparing SL and AP
trends, at least part of the trend inversion can be ascribed to the recent AP increase
that affected the Mediterranean region mainly in winter. In fact, overall, an inverse
relationship is observed between AP and SL at all locations. It is related to the static
inverse barometer effect (IB) and is more pronounced in winter and in the NW Med
and Adriatic, where in this season the water column is generally barotropic. In other
seasons and/or locations SL changes are affected to a large extent also by baroclinic
processes, as the presence of stratification, particularly in summer, and, in the case
of Alboran, the water exchange through the Gibraltar Strait. In winter the IB is
effective both at interannual and, although weaker, interdecadal time scales at all
location except Alboran, where at interdecadal time scales the IB is stronger than
at interannual time scales. In summer the IB is negligible at interdecadal time
scales, indicating that baroclinic processes play the most important role, and in other
seasons it is observed only in the Adriatic.
AP and SL in the Mediterranean region are teleconnected with large scale
patterns of the atmospheric circulation characteristic of surrounding regions. In
winter a major climatic pattern is the North Atlantic Oscillation (NAO, Hurrell,
1995). The AP-NAO teleconnection is highest at the western end of the
Mediterranean (Alboran) and decreases eastwards, although remaining significant
all over the basin. This occurs both at interannual and interdecadal time scales.
Concerning the SL-NAO connection, it is mainly an indirect effect of the AP-SL
correlation. In summer the Mediterranean participates in the circulation of the
Tropics (Ward, 1996; 1998; Raicich et aI., 200 I). It is in fact anticorrelated at
interannual time scales with the interannual anomalies of the precipitation regimes
F. Raicich
examined for interannual and interdecadal time scales. All the time series were
filtered to separate those time scales by means ofNg and Young's (1990) technique;
the cutoff period was set at about 10 years. However, since the SL is affected by a
"secular" trend, time scales longer than about 100 years have been preliminarily
removed from the time series. Due to the absence of information, the effects of
possible vertical land motion on the SL time series could not be checked and
corrected, therefore, in the discussion, it is assumed that SL changes are really
related to SL only, but this assumption has to be taken cautiously. For all the
parameters standardized monthly and seasonal anomalies were computed relative
to the period 1958-87, for which enough SL data is available for all locations to
establish a common climatological period.
3 Discussion
The undetrended SL time series show the well known increasing trend of about \1.5 cm/decade, but also reveal that during the last 30-40 years the tendency has
changed, with a SL decrease rate of o. 5 -1.5 cm/decade. By comparing SL and AP
trends, at least part of the trend inversion can be ascribed to the recent AP increase
that affected the Mediterranean region mainly in winter. In fact, overall, an inverse
relationship is observed between AP and SL at all locations. It is related to the static
inverse barometer effect (IB) and is more pronounced in winter and in the NW Med
and Adriatic, where in this season the water column is generally barotropic. In other
seasons and/or locations SL changes are affected to a large extent also by baroclinic
processes, as the presence of stratification, particularly in summer, and, in the case
of Alboran, the water exchange through the Gibraltar Strait. In winter the IB is
effective both at interannual and, although weaker, interdecadal time scales at all
location except Alboran, where at interdecadal time scales the IB is stronger than
at interannual time scales. In summer the IB is negligible at interdecadal time
scales, indicating that baroclinic processes play the most important role, and in other
seasons it is observed only in the Adriatic.
AP and SL in the Mediterranean region are teleconnected with large scale
patterns of the atmospheric circulation characteristic of surrounding regions. In
winter a major climatic pattern is the North Atlantic Oscillation (NAO, Hurrell,
1995). The AP-NAO teleconnection is highest at the western end of the
Mediterranean (Alboran) and decreases eastwards, although remaining significant
all over the basin. This occurs both at interannual and interdecadal time scales.
Concerning the SL-NAO connection, it is mainly an indirect effect of the AP-SL
correlation. In summer the Mediterranean participates in the circulation of the
Tropics (Ward, 1996; 1998; Raicich et aI., 200 I). It is in fact anticorrelated at
interannual time scales with the interannual anomalies of the precipitation regimes