Observed and CMIP6 Modeled Internal Variability Feedbacks and Their Relation to Forced Climate Feedbacks

Abstract

Inter model variations in global temperature response to increasing atmospheric carbon dioxide
stem mostly from uncertainties in modeled climate feedbacks. To study potential reductions in model feedback
uncertainties, we estimate observed feedbacks in response to internal variability using changes in Top Of
the Atmosphere energy balance with temperature. We compare those observations with internal variability
feedbacks from historical simulations of coupled and atmosphere-only experiments from the sixth phase of
the Coupled Model Intercomparison Project (CMIP6) to identify that simulated feedbacks exhibit biases in the
tropics, subtropics, and the Southern Ocean. Furthermore, we find a relation between simulated longwave and
shortwave internal variability feedbacks and those where atmospheric carbon dioxide is abruptly quadrupled. In
the model range of internal variability feedbacks, the observations are more consistent with moderately negative
longwave feedback, and weak shortwave feedback, but the observations can't be used to rule out any models or
their long-term feedback.

Plain Language Summary

We investigate how Earth's radiation balance changes during natural
variations of temperature, comparing satellite observations to global climate models. Around the globe, we
find predominant negative radiative feedbacks, that is, that the temperature perturbation is dampened by the
longwave radiation emitted from Earth, as well as positive feedbacks from reduced reflected of sunlight during
warmer periods. The magnitudes of feedback vary across latitudes, particularly in the tropics and subtropics,
something which we can relate mostly to clouds. When compared to observations, models in the latest
generation as a group continue to misrepresent the negative longwave feedbacks in the tropics, subtropics, and
the shortwave feedback in the Southern Ocean, although a few of the new models now match the observed.
Moreover, we show that the shortwave and longwave internal variability feedbacks are related to the long-term
feedback to increasing carbon dioxide. Among the models, some underestimate and some overestimate the
negative longwave feedback range indicated by observations. In the shortwave, the observational range is less
consistent with the strongest positive and strongest negative model estimates.

Data Availability Statement

The CERES EBAF-TOA Ed4.1, CERES-MODIS/VIIRS, and GEO, data sets used for estimating observed TOA
fluxes and cloud properties are available at the NASA Langley Research Center via https://ceres.larc.nasa.gov/
data/. The Gridded temperature anomalies HadCRUT (5) data set used to estimate observed internal variability
feedbacks can be obtained from the Met Office Hadley Centre observations datasets at https://www.metoffice.
gov.uk/hadobs/hadcrut5/. Used CMIP6 models can be found in Table S1 in Supporting Information S1 and are
available from ESGF at https://esgf-node.llnl.gov.