LIMITATIONS TO THE QUANTITATIVE APPLICATION OF CD AS A PALEOCEANOGRAPHIC TRACER, BASED ON RESULTS OF A MULTI-BOX MODEL (MENU) AND STATISTICAL CONSIDERATIONS

Citation
Pm. Saager et Hjw. Debaar, LIMITATIONS TO THE QUANTITATIVE APPLICATION OF CD AS A PALEOCEANOGRAPHIC TRACER, BASED ON RESULTS OF A MULTI-BOX MODEL (MENU) AND STATISTICAL CONSIDERATIONS, Global and planetary change, 8(1-2), 1993, pp. 69-92
Citations number
85
Categorie Soggetti
Geosciences, Interdisciplinary
Journal title
ISSN journal
09218181
Volume
8
Issue
1-2
Year of publication
1993
Pages
69 - 92
Database
ISI
SICI code
0921-8181(1993)8:1-2<69:LTTQAO>2.0.ZU;2-L
Abstract
The distribution of cadmium in the modern ocean gives important inform ation about ocean circulation and nutrient distributions. As the Cd/Ca -ratio of foraminiferal shells is proportional to that of seawater, Cd /Ca-records of benthic foraminifera have been used to reconstruct the Glacial oceanic distributions of dissolved Cd and PO4 quantitatively. The results, in turn, have served as boundary conditions for ocean car bon cycle models. This quantitative reconstruction, however, requires that the present oceanic Cd-PO4 relationship and the distribution coef ficient converting Cd/Ca-ratios into dissolved Cd and PO4 concentratio ns, remain constant through geological time. We have constructed a mul ti-box model (MENU: Metal-nutrients) to test various hypotheses about the biogeochemical mechanisms determining the modern oceanic Cd-PO4 re lationship and its behaviour in time. The results indicate that the oc eanic Cd-PO4 relationship may change in response to changing oceanic c onditions, such as circulation, productivity changes and upwelling int ensity, for example during a glacial period. In addition, we discuss t he frequent omission in the literature of statistical considerations p ertaining to regression analysis, resulting in an unduly optimistic es timate of the distribution coefficient, thereby neglecting uncertainti es on the order of 30% or more. In combination with spatial variabilit y of the distribution coefficient, this conceivably poses intrinsic re strictions to its use as a conversion factor. We conclude that, awaiti ng a significant reduction of the aforementioned uncertainties, inter- pretation of the sedimentary Cd/Ca record should be performed only qua litatively.