MILLENNIAL-SCALE CLIMATE VARIABILITY CONTROLLED BY VERTICAL EXCHANGE IN THE OCEAN

Polar ice core records have revealed strong, millennial-scale (MS) climate variability during the last glacial period. Sediment records from the North Atlantic Ocean show persistent MS variability back across the last five, 100 kyr, glacial-interglacial cycles. A variety of climate records document significant MS variability even during the present interglacial period after ice sheet demise (e.g. Medieval Warm Period/Little Ice Age). Although modulation of the thermohaline circulation and climate by glacial melt water has been a favored explanation for MS variability, the persistence of such variability across interglacial periods argues against such melt water forcing. Here we present an alternative explanation for this variability.

Small-scale vertical diffusion in the ocean at low latitudes heats the deep ocean while convection at high latitudes cools it. During periods of strong and weak thermohaline circulation and convection, the ocean interior may therefore cool and warm, promoting weak and strong convection at high latitudes. This lays the basis for an oscillation of the thermohaline circulation, convection and high latitude SST on the diffusive (millennial) time scale of the ocean. Such "deep decoupling" oscillations have indeed been found in two- and three-dimensional ocean general circulation models. We have developed a new simple, coupled climate model (calibrated to reproduce present day conditions) which includes oxygen isotopes in carbonates. Model simulations for interglacial and glacial boundary conditions exhibit MS climate variability with dynamics akin to the "deep decoupling" oscillations. Even without forcing by glacial melt water, these model climate cycles mimic many key features of the MS variability found in the ice and sediment core records. Model cycles are found to be quite sensitive to the strength and nature of vertical exchange in the ocean. More complete models of the climate system should model such exchange realistically in order to correctly capture MS climate cycles and their modulation by climate state.