The Detection of Climate Change

Abstract

This paper reviews some of the techniques that have been proposed for climate change detection during the past fifteen years and describes several applications. We also briefly consider the question of the attribution of climate change. An attempt is made to use uniform notation and nomenclature throughout so that similarities are more readily apparent.

The methodologies fall into two groups; those that use ‘optimal detectors’ and those that use ‘pattern similarity’ statistics. Both approaches require the use of climate models for estimating the signals that are to be searched for, and also for estimating the natural variability of the detectors in the absence of signal. The latter is necessary because the detection statistics are influenced by low frequency climate variations that are not well sampled in the relatively short instrumental record that is currently available. Formally, the use of optimal detectors enhances the prospects for early climate change detection by increasing the signal-to-noise ratio by of the order of 20%. However, real-world differences in performance are likely more strongly affected by how well models represent the climate’s response to changes in forcing from human activity and by variations in the intricate processing of the data that precedes the formal detection step. Recent applications of both types of detection methodology have uncovered ‘smoking gun’ evidence that human activity is having an effect on the recent climate.