Tuesday, 11 February 2003
Long-term Variations in the 1000+ Year PCM Control Run
A 1200-year control run using the Parallel Climate Model (PCM), a fully coupled atmosphere-ocean-land-sea ice global climate model with no flux adjustments and relatively high resolution (~2.8deg for the atmosphere and 2/3deg for the oceans), is analyzed for unforced, long-term (decadal to multi-century) variations in surface temperature, precipitation, El Niņo-Southern Oscillation (ENSO), the Arctic Oscillation (AO), and ocean circulations. The PCM produces relatively stable surface fields without flux adjustments, with a global surface cooling of -0.3degC per 1000 years and regional temperature trends comparable to those of flux-adjusted models. Global mean precipitation is highly correlated with global mean temperature, especially on decadal and longer time scales. Relatively rapid changes in global mean temperature and precipitation occurred a few times during the 1200-year integration. The PCM-simulated surface air temperature at the Greenland Ice Sheet Project Two (GISP2) site shows variation patterns comparable to those of GISP2 ice-core d18O record, although with much smaller variance. ENSO in this control run shows large variability at decadal to centennial time scales, with several periods (of decades) during which El Niņos predominate. The PCM-simulated AO shows spatial and temporal patterns comparable to observed and exhibits little change during the integration. In this control run, the upper, clockwise overturning circulation in the Atlantic Ocean becomes increasingly shallow and weak while the bottom, anti-clockwise circulation gains depth and strength. The overturning changes are accompanied with an eastward shift of the central North Atlantic Drift Current. These ocean circulation changes are qualitatively similar to those in greenhouse gas-forced climate change simulations by the same model, although at a much slower pace. Nevertheless, this result suggests that interpretation of ocean circulation changes in greenhouse gas-forced model runs should be cautious and be put in the context of the changes in unforced control runs.