Abstract
The stochastic character of short time scale atmospheric variables induce unpredictable fluctuations in the climatic system. These fluctuations are treated here as a “forced noise level” to be taken into account in climate modeling. A simple statistical model based on recent work by Hasselmann (1976) is used to estimate the resulting error in the calculated means of climate variables over finite intervals. This error may be large for short record lengths, as illustrated for the sea surface temperature. Some consequences relevant to joint ocean-atmosphere models and climate change experiments are discussed.
