Dept. of Oceanology and Environmental Geophysics
Deep Water Formation Experiment in the Adriatic Sea is being repeated again this Year

M. Gacic , B.B. Manca, R. Mosetti, P. Scarazzato and D. Viezzoli


The Adriatic Sea, due to its location (the northernmost part of the Mediterranean Sea), to mountain orography, and to the relatively large amount of the freshwater river run-off, represents a dilution basin. In addition, due to strong winter heat losses, it has been identified as one of the regions of the World Ocean where deep water formation processes take place. Leaman and Schott (1991), however, did not mention the Adriatic Sea as one of the sites where deep water is formed, probably because of the almost complete lack of evidence of winter vertical convection processes for this area available so far in the literature. The Adriatic Deep Water (ADW), which spreads over the Eastern Mediterranean bottom layer (Pollak, 1951), has distinct characteristics with respect to other Mediterranean water masses, being fresher and colder (salinity S ~ 38.68 and temperature T ~ 13.3 ° C). Experimental evidence of a vertical winter overturning event down to considerable depth, that presumably takes place in the center of the Southern Adriatic, was presented by Ovchinikov et al. (1985) from the analysis of Russian in situ data for three winter situations. Even if the data resolution was apparently not sufficient, they concluded that the horizontal scales of the vertical mixing events are typically a few tens of miles with time scales of the order of few days. More recently, some evidence of violent mixing process accompanied by a ventilation of the water column down to 600 m in winter 1996 has been presented by Manca and Bregant (1998).


The purpose of the deep water formation experiment carried out within the framework of the EU project MATER, was to provide more reliable experimental evidence on the occurrence of this process in the Southern Adriatic. The study was planned to acquire data for the understanding of all three phases of open-ocean convection in the Southern Adriatic. To achieve these goals, firstly from December 28, 1997 to January 5, 1998 a basin-scale survey was carried out in the presumed period of the pre-conditioning phase in order to localize sites of possible convective processes (Fig. 1). Subsequently, from January 16 to January 30, 1998 during the expected vertical convection phase, a mesoscale survey in the center of a cyclonic gyre was undertaken. Finally, a CTD sampling cruise was carried out in the post-convection period in early spring (March 12 through March 23, 1998). During this last cruise, CTD measurements along a single transect across the center of the South Adriatic gyre were carried out. In addition to temperature and salinity, some chemical parameters were measured. More specifically, dissolved oxygen and nutrients were considered important indicators of the vertical convection and/or good tracers of the various water masses that possibly participate in forming the ADW. In situ measurements were complemented with IR and surface color satellite imagery.

In addition, the ADW outflow has been monitored with three bottom mounted ADCP?s in the Strait of Otranto (Fig. 1) since March 1997 and the measurements will hopefully be extended up to summer 1999. The current measurements have been performed in a portion of the Otranto transect, where on the basis of a previous EU project (OTRANTO), a quasi-permanent bottom outflow was evidenced (Gacic et al., 1996).


Low-frequency sub-inertial variability of the Adriatic outflowing current component has typical time scales on the order of a week. These variations are superimposed on the seasonal and year-to-year variability (Fig. 2). The spatial distribution of the bottom flow suggests that the vein of the outflowing Adriatic dense water has horizontal dimensions of about 15 km, while in the vertical it is about 100 m thick. Sub-inertial variations on weekly time scales are manifested by occasional current reversals of the bottom water outflow. Seasonal variations are of the order of 1 cm/s while the interannual variability is slightly stronger. Seasonal variations show a maximum in the outflow in May and a minimum in November which can be explained in terms of the filling up/emptying the South Adriatic dense water reservoir. Year-to-year variations in the bottom water outflow are correlated to the surface buoyancy losses, and consequently to the intensity of the vertical convection and deep-water formation. Estimates of the ADW outflow rate give values of about 0.1 Sv which is two to three times smaller than the estimates obtained from the measurements carried out in 1995 within project OTRANTO (Poulain et al., 1996).

The basin-wide pre-conditioning hydrographic survey covered the entire area of the Southern Adriatic. The typical distance between stations along the same transect was 5 - 10 km, while the distance between neighboring transects is 20 km. The surface, temperature, salinity and density distributions clearly reveal an elongated sub-basin scale cyclonic gyre (Fig. 3). The gyre was embedded in a basin-scale cyclonic circulation and was determined both from temperature and salinity fields since in its center a water mass of higher salinity and lower temperature could be observed. The vertical temperature, salinity and density distributions across the center of the gyre (Fig. 4), in the pre-conditioning phase clearly evidences the doming of surfaces of constant properties in the upper 600 meters. However, it is also evident that even in the center of the gyre, the vertical stability of the water column was rather high with a pycnocline situated at a depth of about 50 m. In the center of the gyre, the outcropping isopycnal was 28.95 kg/m3, whereas the typical density of the ADW is 29.24 kg/m3, which means that the vertical stratification was still rather high with a high buoyancy content. The subsequent cruise which took place in the second half of January, showed that the circulation pattern had changed only to a small extent. The sub-basin cyclonic gyre remained almost unchanged except for the density in its center that had slightly increased so that the outcropping isopycnal became 29.05 kg/m3. An intensification of the upwelling in the center of the cyclone is evident (Fig. 5). It is however important to note that the stratification on the border of the gyre weakened, but the outcropping density in the center of the gyre, was still smaller than the typical ADW density. The March transect (Fig. 6) reveals the occurrence of a chimney having a diameter of about 50 km which delimits a vertically mixed water column down to about 500 m depth. Thus, it appears that at least in the investigated winter, an intermediate and not a deep convection took place in the center of the cyclonic gyre, probably due to the relatively weak surface buoyancy losses and a rather high buoyancy content as well as to a strong vertical stability of the water column.

It can therefore be concluded that the winter 1997/98 deep water formation experiment in the Adriatic Sea has provided important observational evidence of the open-ocean convection process which takes place in the Southern Adriatic. Furthermore, on the basis of the presented data on the field of mass and the ADW outflow, it has been shown that a prominent interannual signal characterizes the temporal variability of the South Adriatic open-sea convection. These results lead us to decide to undertake another deep-water formation experiment in the winter 1998/99 following the same sampling strategy as in the previous winter and hoping this time to experience much stronger winter and to observe at last a fully developed deep convection.


Gacic, M., V. Kovacevic, B. Manca, E. Papageorgiou, P.-M. Poulain, P. Scarazzato and A. Vetrano, 1996: Thermohaline properties and circulation in the Strait of Otranto. In Dynamics of Mediterranean Straits and Channels, edited by F. Briand. Bull. Inst. Oceanogr., Spec. Iss., 17, CIESM Science Series, 2, 117-145.

Leaman, K.D. and F. Schott, 1991: Hydrographic structure of the convective regime in the Gulf of Lions: Winter 1987. Journal of Physical Oceanography, 21, 573-596.

Manca, B. and D. Bregant, 1998: Dense water formation in the Southern Adriatic Sea during winter 1996. Rapp. Comm. Int. Mer Médit., 35, 176-177.

Ovchinnikov, I.M., V.I. Zats, V.G. Krivosheya and A.I. Udodov, 1985: A forming of deep eastern Mediterranean water in the Adriatic Sea (in Russian), Okeanologia, 25, 911-917.

Pollak, M.I., 1951: The sources of the deep water in the eastern Mediterranean. Journal Marine Research, 10, 1, 128-152.

Poulain, P.-M., M. Gacic and A. Vetrano, 1996: Current measurements in the Strait of Otranto reveal unforeseen aspects of its hydrodynamics. EOS Transactions, AGU, 77, 3, 345-348.

Figure captions:

Fig. 1: CTD station (dots) network for the pre-conditioning basin-wide survey (Dec. 28, 1997 - Jan. 5, 1998). ADCP moorings are denoted by asterisks. Depths are given in meters.

Fig. 2: Time-series of the low-pass ADCP data (north current component) for the entire measurement period at the central mooring. The annual and interannual signals are fitted with the harmonic and quadratic functions, respectively. In the upper left, the polar vector diagram shows the year-to-year variations (crosses) and the complete harmonic and quadratic functions fit (dots).

Fig. 3: The surface distributions of the temperature, salinity and density in the pre-conditioning phase. The dots represent sampling stations.

Fig. 4: Vertical distributions of the temperature, salinity and density in the center of the South Adriatic Gyre in the pre-conditioning phase (Dec. 28, 1997 - Jan. 5, 1998).

Fig. 5: The same as in Fig. 4 for the cruise Jan. 17 -31, 1998, but limited to the upper 400 m layer.

Fig. 6: The same as in Fig. 4, but for the post-convection March cruise.


Istituto Nazionale di Oceanografia
e di Geofisica Sperimentale - OGS

Borgo Grotta Gigante 42/c
34010 Sgonico (TS), Italy
Phone +39 040 2140.1
   Fax +39 040 327307

mail to webmaster :
Roberto Cecco