ricevuto il 2 Febbraio 2005; approvato il 30 Maggio 2005; pubblicato online il 23 Settembre 2005) Summary. — Conveyor belt circulation controls global climate through heat and water fluxes with atmosphere and from tropical to polar regions and vice versa. This circulation, commonly referred to as thermohaline circulation (THC), seems to have millennium time scale and nowadays—a non-glacial period—appears to be as rather stable. However, concern is raised by the buildup of CO2 and other green-house gases in the atmosphere (IPCC, Third assessment report: Climate Change 2001. A contribution of working group I, II and III to the Third Assessment Report of the Intergovernmental Panel on Climate Change (Cambridge Univ. Press, UK) 2001, http://www.ipcc.ch) as these may affect the THC conveyor paths. Since it is widely recognized that dense-water formation sites act as primary sources in strengthening quasi-stable THC paths (Stommel H., Tellus, 13 (1961) 224), in or-der to simulate properly the consequences of such scenarios a better understanding of these oceanic processes is needed. To successfully model these processes, air-sea-ice–integrated modelling approaches are often required. Here we focus on two polar regions using the Regional Ocean Modeling System (ROMS). In the first re-gion investigated, the North Atlantic-Arctic, where open-ocean deep convection and open-sea ice formation and dispersion under the intense air-sea interactions are the major engines, we use a new version of the coupled hydrodynamic-ice ROMS model. The second area belongs to the Antarctica region inside the Southern Ocean, where brine rejections during ice formation inside shelf seas origin dense water that, flowing along the continental slope, overflow becoming eventually abyssal waters. Results show how nowadays integrated-modelling tasks have become more and more feasible and effective; numerical simulations dealing with large computational domains or challenging different climate scenarios can be run on multi-processors platforms and on systems like LINUX clusters, made of the same hardware as PCs, and codes have been accordingly modified.