Abstract

Remineralization ratios (-O2:P, C(org.):P, N:P) in the ocean are estimated from ocean tracer data using a new approach, which takes into account the effects of local exchange across neutral surfaces. This approach is applied to temperature, salinity, phosphate, nitrate, dissolved oxygen, alkalinity, and dissolved inorganic carbon data from the low- and mid-latitude Pacific, Indian, and South Atlantic Oceans. The consideration of local exchange effects tends to reduce the -O2:P and C(org.):P remineralization estimates above 1500 m compared to earlier estimates. Below 1500 m, exchange effects can be neglected (except in the South Atlantic) and earlier estimates appear robust. In the deep South Atlantic, the consideration of these effects leads to increased -O2 :P and C(org.):P remineralization ratio estimates, bringing them more in line with the robust deep ocean estimates. For reasonable, open ocean mixing coefficient values and several choices for phosphate remineralization rate profiles, -O2:P (C(org.):P) remineralization ratios in the ocean increase from about 140 (100) at 750 m depth to about 170 (130) at 1500 m and remain so deeper down. Such an increase down through the upper ocean thermocline implies significant fractionation during remineralization of organic matter-nutrients are released higher in the water column than inorganic carbon. These results also argue for a -O2:P (C(org.):P) uptake ratio in new production of about 140-150 (100-110). N:P remineralization ratios decrease from about 15 at 750 m to about 12 at 1500-2000 m. This may reflect a 'true' N:P remineralization (and uptake) ratio of about 16, modified by denitrification. These results imply that applications of derived, quasi-conservative tracers, based on the assumption of constant remineralization ratios, may be subject to significant error for depths less than 1500 m. In addition, present Ocean General Circulation Models of the natural carbon cycle in the ocean-atmosphere system assume remineralization to occur without fractionation but have problems simulating observed, pre-industrial levels of atmospheric pCO2, given observed ocean inventories of alkalinity and dissolved inorganic carbon. Implementation of uptake and (depth-dependent) remineralization ratios estimated here would likely reduce this problem considerably. Furthermore, calculations with a simple global carbon cycle model show that fractionation in the modern ocean, as estimated in the present work, has reduced atmospheric pCO2 by more than 20 ppm below the level it would have had without fractionation.