Methods for climate change detection and attribution

Detection and attribution (D&A) have played a central role within the assessment of the human influence on climate and within IPCC’s reports. Detection involves the statistical demonstration that a change has happened within climatic observations. Attribution consists in assessing the respective contributions of one or several causes to some observed change. Both require the use of climate model simulations, and are based on spatial or spatio temporal patterns of change. This paper provides a very short presentation of the classical ”optimal fingerprint” method for D&A. Some recent developments, regarding the use of ”error in variable” are introduced. Some of the challenging aspects of the method will be discussed too, in particular regarding the very large dimension of the typical datasets used.