Narrowing the Uncertainty in Radiative Forcing (Invited Presentation)

Radiative forcing, the chance in radiative flux caused by a change in the state or composition of the atmosphere, is the fuel for climate change. The fundamental understanding of radiative forcing is among the most mature in the atmospheric sciences. Yet the range of radiative forcings produced by climate models has historically been large even when the change in atmospheric properties is well-specified. The range of forcings inferred by climate models for specified changes from 1850-2005, for example, is large enough to explain much of the variability in warming observed during this period; discrepancies in the treatment of absorption of solar radiation by water vapor are responsible for the inter-model diversity in hydrologic sensitivity (the increase in precipitation per degree warming).

The Radiative Forcing Model Intercomparison Project is characterizing and assessing calculations of radiative forcing for the climate models participating in the current generation of intercomparisons (CMIP6). RFMIP complements the parallel Aerosol and Chemistry MIP, which focuses on the impacts of aerosols and atmospheric composition on air quality and climate, and the Precipitation Driver and Response MIP, which is focused on how perturbations to composition affect the net cooling of the atmosphere, and hence the global-mean precipitation.

RFMIP includes two efforts to assess the radiative parameterizations used in climate models. These build on a long history of such comparisons in the radiation community but extend the scope to the global scale using sampling strategies that accurately represent the global, annual mean using a number of profiles small enough to make benchmark calculations feasible. In this talk I will share results from the global assessment of radiative transfer parameterizations in radiative forcing in clear, aerosol-free conditions under conditions ranging from the present-day (to which parameterizations are tuned) through the pre-industrial to the Last Glacial Maximum, and into the future conditions of quadrupled and octupled carbon dioxide concentrations used to probe climate sensitivity. Line-by-line calculations agree with one another to within a percent or so. This largely reflects broad reliance on the same underlying spectroscopic data bu suggests that benchmark estimates are feasible.

At this writing few results from parameterizations are available but initial results suggest that the range of errors is diverse, especially in conditions far from present-day. I will interpret in light of results from PDRMIP and, to the extent available, AerChemMIP.

A second effort, to assess the treatment of aerosols, is described in a separate presentation.