Climate analysis & attribution
Recent Weakening of the Global Radiative Feedback
In one sentence. A CNN trained on climate models and applied to observed temperature reconstructions shows the global radiative feedback weakened from about -3 to -2 W/m2/K since the mid-1990s, driven partly by Northeast Pacific warming.
Abstract
Earth's climate stability, characterized by the global radiative feedback parameter ($λ$), varies decadally due to changing surface temperature patterns. Recent variations in $λ$ are poorly understood as coordinated model simulations typically end in 2014. We apply a convolutional neural network trained on climate model simulations to observation-based surface temperature reconstructions to estimate variations in $λ$ up to 2025. We find that $λ$ reached a minimum (maximum stability) around the mid 1990s ($λ\simeq -3 {\rm Wm^{-2}/K}$), but has since weakened significantly ($λ\simeq -2\, {\rm Wm^{-2}/K}$). We confirm these results with climate model simulations extended to 2022. The recent $λ$ weakening is not significantly affected by El Niño Southern Oscillation or Pacific Decadal Oscillation. Attribution reveals that warming in the subtropical Northeast Pacific is an important driver of the recently weakened feedback, confirmed by targeted experiments in E3SMv2. Our approach enables near real-time monitoring of Earth's climate stability.
Labels
- Topic
- Climate analysis & attribution
- Type
- Application study
- Method
- CNN / U-Net
- Time scale
- Decadal
- Domain
- Global
- Data
- CMIP / climate model output, Other
- Status
- Research
- Identifiers
- arXiv:2603.12515 DOI:10.1029/2026gl122980 OpenAlex:W7136500642
- Added
- 2026-10-22
Labels and summary are generated by Claude and reviewed weekly. Wrong? Edit this entry.