GEOPHYSICAL RESEARCH LETTERS, VOL. 29, NO. 2, 10.1029/2001GL013781, 2002

3. Model Results

Figure 1. Seasonal effects on phase relations: atmospheric CO₂ (light solid, right axes, in ppm) and temperature (heavy, left axes, in °C) in the southernmost of the three atmospheric boxes. Upper panel: consecutive Novembers and lower panel: annual means.

[9] Figure 1 shows the results for temperatures and atmospheric CO₂ concentrations in the southern atmospheric box of a 300-kyr simulation. Records for model CO₂ and temperature taken for the month of November each year (upper panel) clearly show that CO₂ lags southern temperature by a few kyr at the beginning of glaciation. In contrast, during deglaciation both atmospheric temperature and CO₂ rise almost in phase. When both records represent the annual mean (lower panel), no phase difference arises in either glacial-cycle stage. This example demonstrates the care required when trying to correlate time series of different proxies, in order to identify lead-lag relations.

3.1. Seasonal Dependence of the CO₂-vs.-temperature Phase Lag

[10] Different processes control the atmospheric temperature and atmospheric CO₂ in high southern latitudes. In our model, the 100-kyr variations in atmospheric CO₂ are the result of changes in the rate of vertical mixing and in sea-ice cover in the Southern Hemisphere [Gildor and Tziperman, 2001b]. The CO₂ variations superimposed on the 100-kyr cycle are dominated by solubility variations in high northern latitudes, where the surface ocean water absorbs CO₂ from the atmosphere before sinking into the deep ocean.

Figure 2. Factors that control the phase relationships. a) Evolution of atmospheric CO₂ (light solid, right axes, in ppm) and oceanic average temperature (heavy, left axes, in °C) for consecutive Novembers (upper panel) and annual means (lower panel). b) Evolution of southern-box atmospheric temperature (heavy, left axes, in °C) and insolation (light solid, right axes, in W m^-2): (upper panel) for consecutive Novembers; (middle panel) for consecutive months of May; and (lower panel) for the annual mean.

[11] Figure 2 shows that CO₂ varies in phase with the average oceanic temperature, both in the annual mean and in November-only values. CO₂ solubility is affected by ocean temperature, while both oceanic and atmospheric temperatures in and above the northern subpolar ocean are largely affected by albedo, and hence by the extent of the northern ice sheet [Gildor and Tziperman, 2000]. The variations in the northern surface temperature are then carried by the thermohaline circulation and affect the deep-ocean temperature. Hence, atmospheric CO₂ is highly correlated with boreal summer insolation in northern latitudes.

[12] The model's southern polar atmospheric temperature is controlled by two main factors: sea-ice extent, via its albedo effect, and solar insolation. There is a high (inverse) correlation between sea-ice extent and atmospheric temperature, which is due to the high albedo of sea ice. This relation is observed throughout the year. However, while sea-ice extent is partially affected by the temperature of the deep ocean and hence by Northern Hemisphere processes [Gildor and Tziperman, 2001b], it is also modified by local insolation variations [Kim et al., 1998]. Thus the model's southern polar atmospheric temperature is affected by local insolation both directly and through the insolation effect on sea ice, but it is also affected by northern hemisphere condition.

[13] Figure 2b presents the relation between local insolation and atmospheric temperature in high southern latitudes during November, May, and for the annual mean. It is clear that insolation has a larger effect on atmospheric temperature during November, when the amount of insolation becomes significant. Hence, the relative contributions of local insolation and sea ice to southern atmospheric temperature differ from season to season, and change during the glacial cycle (see lower panel of Figure 2b). The dependence on the march of seasons is quite similar, albeit slightly less striking, when taking three-month seasons (NDJ and MJJ) into account (not shown), rather than single months (as in Figures 1 and 2b).

Citation: Gildor, H., and M. Ghil, Phase relations between climate proxy records: Potential effect of seasonal precipitation changes, Geophys. Res. Lett., 29(2), 10.1029/2001GL013781, 2002.