An objective methodology for identifying oceanic provinces

An objective methodology for identifying oceanic provinces in hydrographic data is presented, although, the technique is sufficiently general so as to be applicable to a variety of data sets. The sub- and near-surface tempera ture (T) and salinity (S) of the Atlantic Ocean were measured on two Atlant ic Meridional Transect (AMT) cruises covering approximately 100 degrees of latitude and 50 degrees of longitude. The cruises covered the boreal autumn and austral spring (AMT-1) and the boreal spring and austral autumn (AMT-2 ) during 1995 and 1996, respectively. There are distinct patterns of change in T-S from 50 degrees N to 50 degrees S for both sub- and nearsurface mea surements, although this study concentrates primarily on the latter. Near-s urface density (sigma(t)) encapsulates the variations in T-S. The position of localized extrema in the sigma(t) first spatial derivative locates the c hange in bulk T-S properties and, hence, the extent of a physical province. The province identification methodology is validated by comparing the resu lts obtained using in situ density, a mixed layer depth space series constr ucted from expendable bathythermograph profiles, and climatological density . Additional validation of the methodology is achieved by intercomparing di fferent in situ data sources between the two cruise periods. The comparison s show the near-surface measurements are indicative of the mixed layer and that the AMT-1 and AMT-2 cruises are representative of the corresponding cl imatological months. A simple scheme for investigating how far a province e xtends away from the cruise track uses the climatological data in concert w ith a T-S description of each province from the AMT along-track observation s to produce a two-dimensional map of the geographical extent of each provi nce. Although exceptions between the province extents and the basic circula tion of the Atlantic Ocean occur, the majority of the provinces have the co rrect shape and extent, that is, their edges follow the flow fields associa ted with the currents that bound them and they extend over an appropriate a rea. The province identification methodology is sensitive and reliable enou gh to investigate temporal differences in oceanic provinces, which means it might be useful for parameterizing global budget calculations in numerical models. (C) 2000 Elsevier Science Ltd. All rights reserved.