Professor Myles Allen CBE FRS

Abstract:

The detection and attribution of climate change in the observed record play a central role in synthesizing knowledge of the climate system. Unfortunately, the traditional method for detecting and attributing changes due to multiple forcings requires large numbers of general circulation model (GCM) simulations incorporating different initial conditions and forcing scenarios, and these have only been performed with a small number of GCMs. This paper presents an extension to the fingerprinting technique that permits the inclusion of GCMs in the multisignal analysis of surface temperature even when the required families of ensembles have not been generated. This is achieved by fitting a series of energy balance models (EBMs) to the GCM output in order to estimate the temporal response patterns to the various forcings. This methodology is applied to the very large Challenge ensemble of 62 simulations of historical climate conducted with the NCAR Community Climate System Model version 1.4 (CCSM1.4) GCM, as well as some simulations from other GCMs. Considerable uncertainty exists in the estimates of the parameters in fitted EBMs. Nevertheless, temporal response patterns from these EBMs are more reliable and the combined EBM time series closely mimics the GCM in the context of transient forcing. In particular, detection and attribution results from this technique appear self-consistent and consistent with results from other methods provided that all major forcings are included in the analysis. Using this technique on the Challenge ensemble, the estimated responses to changes in greenhouse gases, tropospheric sulfate aerosols, and stratospheric volcanic aerosols are all detected in the observed record, and the responses to the greenhouse gases and tropospheric sulfate aerosols are both consistent with the observed record without a scaling of the amplitude being required. The result is that the temperature difference of the 1996-2005 decade relative to the 1940-49 decade can be attributed to greenhouse gas emissions, with a partially offsetting cooling from sulfate emissions and little contribution from natural sources. The results support the viability of the new methodology as an extension to current analysis tools for the detection and attribution of climate change, which will allow the inclusion of many more GCMs. Shortcomings remain, however, and so it should not be considered a replacement to traditional techniques. © 2007 American Meteorological Society.

Abstract:

There has been considerable recent interest in paleoclimate reconstructions of the temperature history of the last millennium. A wide variety of techniques have been used. The interrelation among the techniques is sometimes unclear, as different studies often use distinct data sources as well as distinct methodologies. Here recent work is reviewed and some new calculations performed with an aim to clarifying the consequences of the different approaches used. A range of proxy data collections introduced by different authors is used to estimate Northern Hemispheric annual mean temperatures with two reconstruction algorithms: (1) inverse regression and, (2) compositing followed by variance matching (CVM). It is found that inverse regression tends to give large weighting to a small number of proxies and that the second approach (CVM) is more robust to varying proxy input. The choice of proxy records is one reason why different reconstructions show different ranges. A reconstruction using 13 proxy records extending back to AD 1000 shows a maximum pre-industrial temperature of 0.25 K (relative to the 1866 to 1970 mean). The standard error on this estimate, based on the residual in the calibration period, is 0.14 K. Instrumental temperatures for two recent years (1998 and 2005) have exceeded the pre-industrial estimated maximum by more than 4 standard deviations of the calibration period residual.