Atmospheric radiative cooling (Q) is closely tied to latent heat release during precipitation, and its changes offer critical insight into how changes in the Earth’s energy balance might impact future rainfall. This study analyzes trends in Q (dQ) over 2001-2024 period using CERES observation-constrained estimates, two reanalysis datasets (ERA5 and MERRA2) and the CMIP6 multi-model mean. The datasets show significant divergence in global mean dQ. CERES indicates a negative trend of −0.21 ± 0.45 W m−2 decade−1 (shifting to +0.19 after 2006), ERA5 shows a positive trend of +0.56 ± 0.27 W m−2 decade−1, while MERRA2 and CMIP6 display weaker trends. These differences arise from divergent longwave radiative cooling trends over land, driven by surface temperature biases prior to 2006 and strong temperature and water vapor anomalies associated with ENSO events. Shortwave radiative cooling generally decreases, although regional aerosol variability, including intense Amazonian fires in 2024, may further influence global trends. In reanalyses, cloud representations often mask aerosol radiative effects, complicating attribution.
While consistent trends across independent datasets would suggest a robust signal, our results highlight the limitations that influence dQ, including input dataset inconsistencies, differing treatments of compensating radiative effects of key drivers such as near-surface temperature and water vapor—which amplify Q variability under strong forcing—and the realistic or unrealistic representation of cloud and aerosol trends.
As a result, both measurements and simulations of a dQ indicate a positive trend, but the reliability of its magnitude remains uncertain, highlighting the need to reconcile differences across multiple datasets.
CORE (COnnecting REpositories)
CORE (COnnecting REpositories)