7A.1 Estimating Radiative Feedbacks from Stochastic Fluctuations in Surface Temperature and Energy Imbalance (Invited Presentation)

Tuesday, 9 January 2018: 1:30 PM
Salon F (Hilton) (Austin, Texas)
Cristian Proistosescu, Univ. of Washington, Seattle, WA; and K. Armour, A. Donohoe, G. H. Roe, M. F. Stuecker, and C. M. Bitz

Joint observations of global surface temperature and energy imbalance provide for a unique opportunity to empirically constrain radiative feedbacks. However, the satellite record of Earth’s radiative imbalance is relatively short and dominated by stochastic fluctuations. Estimates of radiative feedbacks obtained by regressing energy imbalance againstsurface temperature depend strongly on sampling choices and on assumptions about whether the stochastic fluctuations are primarily forced by atmospheric or oceanic variability (e.g. Murphy and Forster 2010, Dessler 2011, Spencer and Braswell 2011, Forster 2016).
We develop a framework around a stochastic energy balance model that allows us to parse the different contributions of atmospheric and oceanic forcing based on their differing impacts on the covariance structure – or lagged regression – of temperature and radiative imbalance. We validate the framework in a hierarchy of general circulation models: the impact of atmospheric forcing is examined in unforced control simulations of fixed sea-surface temperature and slab ocean model versions; the impact of oceanic forcing is examined in coupled simulations with prescribed ENSO variability. With the impact of atmospheric and oceanic forcing constrained, we are able to predict the relationship between temperature and radiative imbalance in a fully coupled control simulation, finding that both forcing sources are needed to explain the structure of the lagged-regression. We further model the dependence of feedback estimates on sampling interval by considering the effects of a finite equilibration time for the atmosphere, and issues of smoothing and aliasing.
Finally, we develop a method to fit the stochastic model to the short timeseries of temperature and radiative imbalance by performing a Bayesian inference based on a modified version of the spectral Whittle likelihood. We are thus able to place realistic joint uncertainty estimates on both stochastic forcing and radiative feedbacks derived from observational records. We find that these records are, as of yet, too short to be useful in constraining radiative feedbacks, and we provide estimates of how the uncertainty narrows as a function of record length.
- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner