license: apache-2.0
ACE2-SOM
Ai2 Climate Emulator (ACE) is a family of models designed to simulate atmospheric variability from the time scale of days to centuries.
Disclaimer: ACE models are research tools and should not be used for operational climate predictions.
ACE2-SOM is the first model we have trained to emulate the sensitivity of climate to substantial increases in carbon dioxide. It consists of ACE2 coupled to a highly simplified physics-based ocean model, called a slab ocean model (SOM), and is trained on equilibrium-climate output from GFDL's SHiELD model coupled to an identical slab ocean with carbon dioxide concentrations of the present-day, double the present-day, and quadruple the present-day (1xCO2, 2xCO2, and 4xCO2).
Quick links:
- 📃 Paper
- 💻 Code
- 💬 Docs
- 📂 All Models
Briefly, the strengths of ACE2-SOM are:
- accurate emulation of equilibrium climates with CO2 concentrations ranging from 1xCO2 to 4xCO2, including changes in the spatial patterns of temperature and precipitation, and changes in precipitation extremes between climates
- realistic emulation of surface and lower-to-middle atmosphere fields with gradually changing CO2 concentration between 1xCO2 and 4xCO2.
Some known weaknesses are:
- Sea-ice coverage, ocean heat transport, and ocean mixed layer depth are prescribed based on a present-day climatologies and therefore do not respond to changes in CO2. Prescribed sea-ice in particular means projections with ACE2-SOM are missing an important feedback mechanism known to amplify warming in the polar regions.
- Abrupt regime shifts in stratospheric temperature and moisture can occur in inference runs with changing CO2 concentration.
- Adjustment to an abrupt change in CO2 concentration occurs too quickly, violating energy conservation.
- Sensitivities of the surface and top of atmosphere radiative fluxes to changes in CO2 with the atmospheric state held fixed are not always realistic.