VERBI VEnice Research BIbliography

This paper illustrates the results of a classification analysis of around 200 spatial distributions of temperature and salinity in the Northern Adriatic Sea.
Maps are obtained by spatial interpolation of experimental observations collected during many oceanographic research projects performed in this area during the last 20 years. The classification is based on Growing Hierarchical Self Organisizing Map, a numerical procedure which combines advantages of competitive unsupervised artificial neural networks and hierarchical clustering procedure. The classification returns a typology of spatial patterns of the thermohaline features of the system. Meteorological conditions associated to each pattern, together with information of the regime of Po river, are used to infer on relationship between these forcing and the emerging structures.
Results represent a contribution to the characterization of the waters of Northern Adriatic Sea, to the comprehension of interactions between major forcing of the Northern Adriatic Sea, its general circulation patterns and spatial distribution of biogeochemical properties.