Regional Sensitivity to CO2 forcing: Detection and Attribution Approach to Analyzing CO2 Component of Late Quaternary Climate Variability

Abstract:

We present a detection and attribution approach to interpret the forcings and feedbacks that shaped Quaternary climate stemming from known variations in greenhouse gases. Because future climate changes will be driven largely by only one forcing, CO₂, it is important to separate and understand the individual contributions of different forcings in producing past recorded changes. We use idealized simulations of the coupled atmosphere-ocean-sea ice system as represented by the GFDL CM2.1 and the NCAR CESM to estimate fingerprints of the annual mean responses to changes in obliquity, precession, CO₂, and ice sheets. For each location where we have observations, we can estimate the model’s time varying response to glacial cycle forcings by scaling and summing the fingerprints by known time series of past forcings. A multiple linear regression is conducted using Bayesian inference between the components of the linear reconstruction and long proxy time series for annual mean temperature from various geologic archives. In particular we determine scaling factors that would allow the modeled response to each forcing to better match the data. Estimates of the uncertainties in the data and model fingerprint patterns can be incorporated to establish the adequacy of different hypotheses (or models) to explain observed regional response patterns.