2007

The WCRP CMIP3 Multimodel Dataset: A New Era in Climate Change Research

G. A. Meehl, et al.

Observation and Context

In the 1980s, climate change modeling was isolated, as individual research groups developed and analyzed their own models without sharing data externally. During the late 1990s lead-up to the IPCC Third Assessment Report (TAR), a set of 30 SRES emission scenarios was introduced, but because of a tight timeline and computing limits, only a few groups could run a couple of scenarios. Very little data was collected or shared.

To prepare for the IPCC Fourth Assessment Report (AR4), the international Climate Variability and Predictability (CLIVAR) Working Group on Coupled Models (WGCM) observed that they needed a coordinated approach. They recognized that standardizing experiments, providing more lead time, and establishing open access to model data would allow a broader community of scientists to analyze climate variability and change.

Hypothesis

If international modeling groups perform a standardized set of global coupled climate experiments and centralize their atmospheric, land, ocean, and sea ice data in an open-access archive, then researchers worldwide can collaborate to produce a comprehensive, stable assessment of twentieth-century climate simulation and future greenhouse-warming projections.

Experiment and Methodology

The WCRP committee coordinated an international experiment, inviting modeling groups to use state-of-the-art atmosphere-ocean general circulation models (AOGCMs). They defined a set of standard experiments, which included:

• Historical Run: A twentieth-century simulation up to the year 2000.
• Future Scenarios: Twenty-first-century projections using low (SRES B1), medium (SRES A1B), and high (SRES A2) greenhouse gas forcing scenarios.
• Climate Commitment Runs: Idealized stabilization experiments where carbon concentrations were fixed at year 2000 or 2100 levels to observe delayed warming.
• Calibration Experiments: Standardized 1% compounded CO₂ increase experiments to calculate the transient climate response.

Because the volume of data was too large for online transfers, groups copied their netCDF-formatted model components (such as monthly atmosphere, land, and ocean grids, daily averages, and extremes indexes) onto physical hard disks and mailed them to the Program for Climate Model Diagnosis and Intercomparison (PCMDI). This centralized database became known as the WCRP CMIP3 multimodel dataset. WGCM then organized the analysis phase, granting open-access permissions to hundreds of international scientists and students.

Results and Data

By early 2005, a total of 16 modeling groups from 11 countries had submitted output from 23 different models, compiling over 30 terabytes of data. The coordinated dataset yielded key scientific findings:

• Twentieth-Century Warming: The models successfully simulated the observed 0.6 °C warming during the twentieth century.
• Twenty-First-Century Projections: Models projected average global warmings of 1.8 °C, 2.8 °C, and 3.4 °C for the B1, A1B, and A2 scenarios, respectively, by the decade 2090–2099.
• Climate Commitment: Stabilizing greenhouse gases at year 2000 levels still resulted in an additional warming of approximately 0.6 °C by 2100.
• Extreme Events: Analysis of the extremes indexes showed a global increase in precipitation intensity (harder rain) alongside an increase in consecutive dry days in several areas.
• Sea Ice & Ocean Circulation: The models captured the observed dipole pattern of Antarctic sea ice. Weighted ensembles also projected a gradual 25% weakening of the Atlantic Meridional Overturning Circulation (MOC) by 2100, with no sudden shutdown.

Conclusion and Climate Impact

The experiment supported the hypothesis. CMIP3 successfully established a new era of open, collaborative climate research. Centralizing and sharing massive data archives allowed hundreds of independent scientists to validate model physics and quantify projections. By shifting the field away from isolated modeling groups, CMIP3 provided the scientific foundation for the IPCC AR4 and significantly advanced humanity’s collective ability to analyze climate risks, extreme weather, and greenhouse warming pathways.