Filtering of Milankovitch cycles by the thermohaline circulation

Abstract

A low-order, basin-averaged, coupled atmosphere-ocean paleoclimate model is developed and the results from a 3.2-Myr model paleointegration described. A three-basin version of the Wright-Stocker ocean model is used to compute the thermohaline circulation component of the climate system, with a six-basin energy balance atmosphere coupled to the ocean and land surfaces. As expected, amplitude spectra of the paleo-integration results show that the annually averaged global air temperature (T(atm)) closely follows the net radiation (Q(N)), with power in the obliquity (40 kyr) and eccentricity bands (100 kyr and 400 kyr). However, there are also some unexpected results: the globally and annually averaged ocean temperature (T(ocean)) is negatively correlated with T(atm) in the obliquity band, T(ocean) shows significant energy in the precessional band (20 kyr), and the response of T(ocean) to Q(N) variations is suppressed in the eccentricity band. Physical explanations for the above results are presented and supported by a simple box climate model. This model helps to isolate and clarify the mechanism by which the ocean temperature varies significantly at precessional periods while the atmospheric temperature does not. The same model also illustrates the cause of the 40 kyr atmosphere-ocean temperature anticorrelation. Model integrations and analysis confirm that convection serves to rectify the zero annual-mean precessional forcing, resulting in 20 kyr energy in the ocean, which shows up only weakly in the atmosphere. The 40 kyr anticorrelation is the result of the latitudinal distribution of net radiation at obliquity periods, and thus can be reproduced only by a climate model with horizontal resolution. Ocean convection plays a critical role in determining both the 20 kyr and 40 kyr responses. The suppressed response of the ocean in the eccentricity band is attributed to a combination of two effects. First, the larger albedo at high latitudes results in reduced variation of the air-sea heat flux at high latitudes so that variations in convection, and hence in deep water temperatures, are also reduced. Second, the nonlinearity of the equation of state for seawater contributes latitudinal variations in ocean densities, which result in changes in the overturning circulation, which further suppress the ocean temperature variations in the eccentricity band. The implications of the authors' results for interpretations of the paleoclimate record are discussed.