The authors designed ClimateBench v2 to compare physics-based, machine-learned and hybrid climate models under the same train-test contract.
A common test for probabilistic climate projections across model types
ClimateBench v2 evaluates physical, data-driven and hybrid models using held-out observations, paleoclimate states and cross-model prediction tests.
Big Tech
Duncan Watson-Parris · Willa Tobin · Aytaç Paçal · Manuel Schlund · V. Balaji · Kevin Bowman · +29 more
Scripps Institution of Oceanography · University of California San Diego · Deutsches Zentrum für Luft- und Raumfahrt e.V. (DLR) · Schmidt Sciences · Jet Propulsion Laboratory
Research Digest··2 min read
Watson-Parris and colleagues propose a standardized, three-tier protocol for testing models that project regional temperature and precipitation change around mid-century.
Why this paper
From Google Research and 19 others
In one line
ClimateBench v2 standardizes evaluation of climate models on mid-century regional temperature and precipitation projections.
What we could check
- ·No code link found
- ·No weights link found
- ·No dataset link found
- ·No compute details found
- ✓Limitations stated by the authors (2 noted)
- ·No benchmark numbers found
Observed from the paper text and links we have. Absence here means we did not find it, not that it does not exist.
§