On the Late Pleistocene ocean geochemistry and circulation


  • Robin S. Keir


A box model of the atmosphere and ocean was developed to investigate how geochemical distributions extant during the late Pleistocene may have come about. The model simulates the regional distribution of calcium carbonate dissolution as well as the chemical oceanography and atmospheric CO2, δ13C, and radiocarbon. If the downward biological flux of particulate carbon increases by a factor of 2 to 3 in the Antarctic and if this increase is combined with a relative increase of the Atlantic sector Antarctic Bottom Water (AABW) versus North Atlantic Deep Water (NADW) source ratio from 1∶3 to about 2∶1, then the model predicts several changes that seem to be recorded in the sedimentary record, as follows: (1) A global redistribution of nutrients and 12C from the intermediate to deep water takes place with the Atlantic intermediate water phosphate decreasing 0.6 µmole kg−1 and the δ13C increasing 0.3 to 0.5‰. (2) The dissolved oxygen level of the deep sea is reduced from an average of about 180 to 70 µmole kg−1, but the intermediate water oxygen declines only a small amount. (3) The decrease in intermediate water nutrient concentration results in lower average organic carbon and calcium carbonate production in the warm surface ocean. (4) The atmospheric CO2 decreases by 90 to 110 ppm. (5) Initially, a global increase in calcium carbonate dissolution occurs, which is followed by a relaxation toward better preservation than exists for the present ocean. In the model in this paper the reduction of NADW by itself does not produce these effects. Rather, the nutrient decrease that does occur is found mostly in North Pacific intermediate water, and the model atmospheric CO2 decrease is only 10 to 30 ppm. It is observed that 92% of the atmospheric CO2 change takes place according to a 200-year time constant in the model. This corresponds to the response time of the upper ocean and atmosphere to a change in the stationary state atmospheric PCO2. Thus, according to this model, the time lag between the nutrient-based cause and the atmospheric CO2 response is not expected to be particularly large.