QUARCC: Quantifying Uncertainty in the Attribution of Recent Climate Change

OBJECTIVES:
  • To quantify the robustness of recent reports of an attributable anthropogenic influence on global climate through the application of a consistent optimal detection methodology to a wider range of alternative climate change mechanisms, a wider range of model predictions and a wider range of observational data sources than have been addressed to date.
  • To investigate the simulation of natural climate variability in the coupled models used for climate change detection and attribution to establish the extent to which uncertainty estimates based on these models may under- (or over-) estimate the true uncertainties, and to develop methods of correcting any uncertainty analysis when model variability is found to be deficient.
  • To quantify the implications of detection and attribution results in terms of physically-interpretable processes and parameters using simplified coupled ocean-atmosphere climate models.

    • WORK CONTENT:
    A common procedure for detection and attribution studies will be developed, based on optimal fingerprinting but revised to (i) allow for noise in the model-predicted patterns, (ii) incorporate diagnostic checks to ensure that uncertainty estimates are consistent with the dataset used in the attribution study, (iii) standardize the choice of truncation and treatment of missing data and (iv) incorporate prior expectations in a maximum-likelihood framework.

    Following detailed statistical evaluation, the algorithm will be used to provide quantitative estimates of uncertainties in the climate response to greenhouse gases, sulphate aerosols, ozone depletion, volcanic aerosols and solar variability. These estimates will be based on the analysis of the historical surface temperature record and vertically-resolved radiosonde record of atmospheric temperatures using patterns derived from both the Hadley Centre and suit of ECHAM-n climate models.

    All current uncertainty estimates in climate change detection and attribution depend on model simulations of internal climate variability. The adequacy of model-simulated variability over the full range of spatio-temporal scales will be evaluated and advanced statistical techniques will be applied to quantitify any bias in uncertainty estimates.

    Current approaches to detection and attribution involve estimating the range of possible amplitudes of model-predicted response-patterns in the observational record. For these results to inform model development, these pattern-amplitude-ranges must be interpreted in terms of physically meaningful parameters. A reduced-physics (zonally averaged) climate model will be calibrated to mimic the behaviour of the GCMs under the full range of forcing scenarios. This will allow a physical interpretation of detection and attribution results in the framework of a non-linear model and suggest which physical parameters are constrained (or left unconstrained) by detection results.

    Throughout, this project will provide information, as required, to EU policy makers, both directly and through the Intergovernmental Panel on Climate Change (IPCC).

    PARTNERS:

    | Jim Hansen's Home | EAPS Home |