Skip to main content
U.S. flag

An official website of the United States government

Publication Date
21 November 2022

Reconciling Model-Satellite Differences in Tropical Atmospheric Warming

Subtitle
Model-versus-satellite warming difference is explained by natural climate variations and biases in the prescribed model forcing.
Print / PDF
Powerpoint Slide
Image
Figure showing the effect of removing a) internal variability (which reduces satellite-derived warming) and b) biomass burning aerosol emission biases (which enhance model simulated warming). Removing these two effects largely reconciles model-satellite warming differences in the tropical troposphere.
Science

Scientists at Lawrence Livermore National Laboratory worked with experts at CSU, NCAR, and UCLA to apply machine learning to large ensembles of climate model simulations to separate and quantify factors contributing to satellite-era warming. Their research shows that natural climate variations have reduced real-world warming and that biases in the model-prescribed biomass burning aerosol emissions have resulted in inflated climate model warming.

Impact

Climate models have persistently simulated greater warming than satellite observations of the tropical troposphere (the lowest 15km of the atmosphere), suggesting that models may be too sensitive to greenhouse gas emissions. This research disentangles the factors contributing to this warming discrepancy and indicates that model sensitivity biases are not needed to explain the model-versus-satellite warming difference.

Summary

The average simulated tropical tropospheric temperature change has outpaced satellite observations over the last three generations of climate models, with few simulations reproducing the observed rate of warming. One explanation for the model-satellite difference is that natural, internal climate variability has slowed real-world warming. The researchers applied machine learning to model-simulated maps of surface temperature change to disentangle the externally-forced and natural, unforced contributions to the satellite-era (1979 – present) rate of tropical tropospheric warming. The approach was successful in accurately quantifying the forced and unforced components of tropical tropospheric warming in model simulations. In applying machine learning to observed surface warming the researchers found that natural variations in climate had significantly reduced the satellite-era rate of warming. Recent research has also documented a spurious increase in the variability of biomass burning (BB) aerosol emissions, which enhances simulated surface warming after the mid-1990s. This study determined that this issue also affects tropical tropospheric warming and has the effect of artificially enhancing model-simulated warming. Real-world internal variability and biases in BB aerosol emissions largely explain the model-satellite warming gap.

Point of Contact
Dr. Stephen Klein
Institution(s)
Lawrence Livermore National Laboratory (LLNL) - PCMDI
Funding Program Area(s)
Publication