CHANGE IN GLOBAL TEMPERATURE - A STATISTICAL-ANALYSIS

This paper investigates several issues relating to global climatic cha nge using statistical techniques that impose minimal restrictions on t he data. The main findings are as follows: 1) The global temperature i ncrease since the last century is a systematic development. 2) Short-t erm variations in temperature do not have long-lasting effects on the final realizations of the series; over time, stochastic perturbations dissipate and temperature reverts to trend. 3) Multivariate tests for causality demonstrate that atmospheric CO2 is a significant forcing fa ctor. The implied change in temperature with respect to a doubling of atmospheric CO2 lies in a range of 2.17-degrees to 2.57-degrees-C, wit h a mean value of 2.34-degrees-C. The contributions of solar irradianc e and volcanic loading are much smaller. 4) In a multivariate system, shocks to forcing factors generate stochastic cycles in temperature co mparable to the results from unforced simulations of climatological mo dels. 5) Extrapolation of regression equations predict changes in glob al temperature that are marginally lower than the results from climato logical simulation models.