Skip to main content
U.S. flag

An official website of the United States government

Publication Date
30 March 2021

Natural Variability Helps to Explain the Gap Between Atmospheric Warming in Satellite Observations and Climate Models

Greater modeled-versus-observed warming represents a longstanding discrepancy; new research sheds light on the issue.
Print / PDF
Powerpoint Slide

Scientists at Lawrence Livermore National Laboratory in collaboration with colleagues from Princeton University and the University of Washington compared more than 400 climate model simulations to satellite observations of tropical tropospheric (lowest ~10km of the atmosphere) warming. Even though model average warming exceeds observational estimates, the researchers find that many simulations can reproduce satellite observed warming estimates, particularly when natural, but random fluctuations in Pacific climate variability are similar in model simulations and observations.


Past studies have suggested that greater-than-observed model warming results because climate models are overly sensitive to increases in the atmospheric concentration of greenhouse gases. This study shows that natural climate variability has reduced the rate of observed tropical tropospheric warming over the past four decades, which helps to explain the model-observational difference, and must be accounted for when evaluating climate models.


The tropical troposphere has warmed rapidly over the past 40 years in response to greenhouse gas forcing, but climate models simulate approximately two times more warming than satellite observations. As a result, it has been hypothesized that climate models are overly sensitive to greenhouse gas changes. This study makes use of large initial condition ensembles which are comprised of a number of simulations from a single model. The rate of warming across the large ensemble varies widely due to the influence of natural, stochastic climate variability. Although ensemble average warming exceeds satellite observations, some model realizations simulate reduced tropical tropospheric warming that is within the range of satellite observations. These realizations exhibit a spatial pattern of ocean warming that is similar to observations and has the imprint of multidecadal climate variability. It is therefore likely that satellite-era (1979 – present) tropical tropospheric warming has been reduced by natural climate variability. Climate models with both small and large sensitivities to greenhouse gas changes have individual realizations that have tropical and global tropospheric temperature trends in the range of observations (13 and 24% of CMIP6 simulations, respectively). Natural climate variability is an important consideration when evaluating climate models and can be significant even when considering four decades of satellite observations.

Point of Contact
Stephen Po-Chedley
Lawrence Livermore National Laboratory (LLNL)
Funding Program Area(s)