Statistical Considerations in the Evaluation of Climatic Experiments with Atmospheric General Circulation Models

View More View Less
  • 1 The Rand Corporation, Santa Monica, Calif. 90406
© Get Permissions Rent on DeepDyve
Restricted access

Abstract

Numerical simulations made with atmospheric general circulation models contain short-term fluctuations that need to be taken into account when considering the significance of long-term climatic means. When such models are used in the study of the climatic changes resulting from either altered forcing or changed boundary conditions, a quantitative evaluation of the effects of the fluctuations, as measured by the variance of model-generated climatic estimates, is required. Such fluctuations are generally of large magnitude, and necessitate extended simulations or repeated Monte-Carlo calculations in order to determine the climatic means with confidence.

Based on a review of statistical questions that are important for climate modeling, this paper develops a rationale for planning the climate simulation trials needed to establish the significance of proposed climatic change experiments. In particular, a quantitative prescription is given for determining the extent of model calculations required to establish the presence of a climatic change at any chosen confidence level between two numerical experiments of unequal variance.

Abstract

Numerical simulations made with atmospheric general circulation models contain short-term fluctuations that need to be taken into account when considering the significance of long-term climatic means. When such models are used in the study of the climatic changes resulting from either altered forcing or changed boundary conditions, a quantitative evaluation of the effects of the fluctuations, as measured by the variance of model-generated climatic estimates, is required. Such fluctuations are generally of large magnitude, and necessitate extended simulations or repeated Monte-Carlo calculations in order to determine the climatic means with confidence.

Based on a review of statistical questions that are important for climate modeling, this paper develops a rationale for planning the climate simulation trials needed to establish the significance of proposed climatic change experiments. In particular, a quantitative prescription is given for determining the extent of model calculations required to establish the presence of a climatic change at any chosen confidence level between two numerical experiments of unequal variance.

Save