Detection and Attribution of Temperature Trends in the Presence of Natural Variability

Wednesday, 17 December 2014: 1:40 PM
John Michael Wallace, University of Washington, Seattle, WA, United States
The fingerprint of human-induced global warming stands out clearly above the noise In the time series of global-mean temperature, but not local temperature. At extratropical latitudes over land the standard error of 50-year linear temperature trends at a fixed point is as large as the cumulative rise in global-mean temperature over the past century. Much of the samping variability in local temperature trends is “dynamically-induced”, i.e., attributable to the fact that the seasonally-varying mean circulation varies substantially from one year to the next and anomalous circulation patterns are generally accompanied by anomalous temperature patterns. In the presence of such large sampling variability it is virtually impossible to identify the spatial signature of greenhouse warming based on observational data or to partition observed local temperature trends into natural and human-induced components. It follows that previous IPCC assessments, which have focused on the deterministic signature of human-induced climate change, are inherently limited as to what they can tell us about the attribution of the past record of local temperature change or about how much the temperature at a particular place is likely to rise in the next few decades in response to global warming. To obtain more informative assessments of regional and local climate variability and change it will be necessary to take a probabilistic approach. Just as the use of the ensembles has contributed to more informative extended range weather predictions, large ensembles of climate model simulations can provide a statistical context for interpreting observed climate change and for framing projections of future climate. For some purposes, statistics relating to the interannual variability in the historical record can serve as a surrogate for statistics relating to the diversity of climate change scenarios in large ensembles.