Abstract
A long time series of data simulated by the NCAR Community Climate Model is used to empirically determine the effects of imperfect spatial and temporal sampling on estimates of the model's global-mean surface air temperature. Results determined from a simple statistical sampling equation and those determined from a simple Monte Carlo experiment are shown to be reasonably similar to the empirically determined ones. The results provide insights into the spatial sampling problem, and it is proposed that variations of the simple models can be used in studies of real data to arrive at reasonable estimates of error that are directly applicable to the actual global-mean surface temperature.
