Equilibrium Climate Statistics of a General Circulation Model as a Function of Atmospheric Carbon Dioxide. Part I: Geographic Distributions of Primary Variables

Abstract

In order to estimate quantitatively the climatic response as a continuous function of CO2 (for plausible earth values), we have performed an extended series of simulations with a general circulation model, the NCAR CCM1. A set of six ``primary{''} simulations has been made, with 100 ppm, 200 ppm, 3 30 ppm, 460 ppm, 660 ppm, and 1000 PPM CO2. From these simulations, logarithmic sensitivity coefficients are computed for primary climatic variables. (Previous work by ourselves and others has suggested that the basic nature of the sensitivity should be logarithmic rather than, say, linear. This means that a decrease in CO2, relative to present-day values, has a bigger effect than an equivalent increase.) Variables considered include surface temperature, sea level pressure, the 500-mb winds, specific humidity, precipitation, and sea-ice cover. Two measures of the uncertainties of these coefficients, the least-squares log misfits and the jackknife standard deviations, have also been computed. Model climatic variables can be broadly classified into two groups: those with a large sensitivity to CO2 changes (surface temperature, specific humidity, and sea-ice cover) and those with less sensitivity to CO2 changes (the winds and surface pressure). Much of the large surface temperature sensitivity can be explained in terms of changing specific humidity (water vapor feedback) at low and midlatitudes and in terms of changing sea-ice cover (albedo feedback) at high latitudes. The water vapor feedback in particular appears to have a strongly logarithmic behavior, possibly helping to account for the overall loglike response of most model variables to changes in CO2. Important regional responses are also seen, even in those variables that have a relatively small overall sensitivity to CO2 changes.