(1958), Cochrane (1958), and Pollak (1958) still stand,but I believe that their classes are too large. One cansee that by merging the present classes into divisionsas large as 0.5%o of salinity, as Cochrane and Pollakhave done, or 1.0°%o, as Montgomery has done, thehigh-volume warm prongs (figure 2.9) would be artifi-cially blurred and merged with one another.One or two colleagues have asked whether I couldnot use a logarithmic volume scale in such presenta-tions as figures 2.2-2.5 so that the warm water masses(if included) could be made to stand out more clearly,but one of the principal virtues of the volumetric T-Sdiagram is that it displays the relative abundances ofthe water masses as they actually exist. The concen-tration of water in the most abundant North Pacificclass exceeds that in the warm-water prongs (shown infigure 2.9) by a ratio of about 25,973 to 10 or less. Thisis analogous to comparing the elevation of Mount Ev-erest to that of Water Street, Woods Hole, near theoriginal building of the Woods Hole OceanographicInstitution. In fact, if we were able to sample and meas-ure salinity more perfectly, the apparent elevationsshown in the deep water in figure 2.8 would probablybe even higher.The feelings I have about the census are compoundedequally of fascination and frustration. The frustrationis the result of the decrease in the rate of acquisitionof new high-quality data. This decrease is due in partto the trends in modem physical oceanography inwhich the dramatic improvements in direct currentmeasurements have understandably taken priority overroutine measurements of water properties on a largescale. It is also clear that there is a long delay (as muchas 5 years) between the time hydrographic data areobtained at sea and the time these data become avail-able on tape from the National Oceanographic DataCenter (in part because some investigators take a longtime to turn their data in to the Data Center). I havebeen reluctant to obtain new data informally, fromfriendly colleagues, however, because I do not thinkthat the Data Center should be bypassed at present; itsfunction would be impaired if data were only ex-changed between a cabal of skilled observers.The fascination results from the precise but peculiarway in which the water masses of the oceans are ar-ranged-particularly the deep water masses that makeup the greater part of the oceans. Why, for instance, arethe big, exclusive North Pacific classes fresher thanexisting circumpolar and South Pacific waters? Arethey fossil water masses that were formed in some pastmillennium when the oceans were somewhat fresher,or are they still undergoing a change toward the fresheras the result of slow vertical mixing (across densitysurfaces) with the still fresher water that lies abovethem at the present time? I do not think that we cansupply answers to such questions at present, and an-swers will not be available even in the future withoutpainstaking observations. There are indications thisstyle of observations may be coming back into vogue.The authorless Scripps data report of the INDOPACexpedition (Scripps Institution of Oceanography Ref-erence: 78-21) is an excellent example. It should beworthwhile to reactivate this census (which was closedas of June 1977) when more high-quality data of thiskind are available from NODC, and I shall probablypropose to do so at some time in the future.2.4 The Formation of Water MassesThere is only one hypothesis about water-mass for-mation that is universally agreed upon, that is, thatthe cold, dense water that fills the great ocean basinshas been formed at high latitudes. The manner inwhich the thermocline-halocline is formed is underdispute, and there are almost as many notions of therate at which all the various water masses are formedas there are investigators.Given the extraordinary regularity of the T-S curvesthat are found in much of the oceans, it is natural toassume that these curves are the result of vertical mix-ing between two end water masses. Very simply stated,this assumption implies that the bottom water (as allagree) has been formed at high latitudes, that the sur-face water at middle and low latitudes has received itsT-S characteristics from the atmosphere by the unevenprocesses of evaporation and heating, and that the re-mainder of the water column is a mixture of surfaceand bottom water. Wiist (1935) clearly recognized thatthis was an oversimplification, and his use of the "core-layer" method reflects his conviction that differentwater masses can be traced to a small number of more-or-less point sources at the sea surface over a widerange of latitude.The notion that all the thermocline water massescan be traced to the sea surface is generally attributedto Iselin (1939a). He constructed a T-S diagram fromwinter observations at the surface of the western NorthAtlantic, and found that it corresponded closely to theT-S diagram obtained from a typical hydrographic sta-tion in that ocean. It is worth noting that Wiist (1935,p. 3) anticipated Iselin (in the South Atlantic) by 4years. He wrote, "The vertical structure of the Subant-arctic Intermediate Water, with its horizontal spread-ing at depths, is analogous to a vertical figure of thehorizontal arrangement of temperature and salinity atthe surface of the formation region." Wiist did notdwell on this subject further, and it is clear that heregarded core layers as more important as indices ofocean circulation.In his 1939a paper, Iselin stressed "lateral mixing"as responsible for the T-S curve in the western NorthAtlantic. Sverdrup, in chapter XV of The Oceans (Sver-57Water Masses of the World Ocean: A Fine-Scale Censusdrup, Johnson, and Fleming, 1942) amplified Iselin'sconcept; he suggested that "subtropical convergences"were the dominant source of the waters in the ther-mocline-halocline. In these convergences, according toSverdrup, surface water sinks, in late winter, over awide range of latitude. He compared late-winter, sea-surface T-S points to the T-S curves obtained fromsubsurface hydrographic data and found a close
View Full Document
Unlocking...