ACCESS-CM2 SSP scenario ensemble runs

Earth System Experiments
1

Experiment title:
ACCESS-CM2 SSP scenario ensemble run from R6 to R10

Summary:
To extend the ensemble member of ACCESS-CM2 scenario experiments for CMIP6, we extend the previous 5 ensemble members to 10. The ensemble members from the r6 to the r10 were being run here recently. Totally 12 experiments were running on lg87: 4 for SSP126, 4 for SSP245, 4 for SSP370 and 4 for SSP585.

Scientific motivation:
This ensemble member runs are multiple simulations of ACCESS-CM2 model that differ in initial conditions. Increasing the number of ensemble members in CMIP6 scenario, such as those, provides several benefits:

  1. Reduce Uncertainty: Increasing the number of ensemble members helps reduce the uncertainty in projections due to internal climate variability. Internal variability is the natural variability in the climate system when there are no changes in external forcing factors like greenhouse gas concentrations or solar radiation. More ensemble members mean a wider range of this natural variability can be captured and understood.

  2. Better Capture of Extreme Events: More ensemble members can help better capture rare or extreme climatic events, which might not be sufficiently represented with fewer runs. This can be particularly important for understanding and preparing for extreme weather events under future climate scenarios.

  3. Statistical Significance: More ensemble members can provide a larger dataset for analysis, which can improve the statistical significance of the results.

  4. Multiple Initial Conditions: Increasing the ensemble members allows more initial condition sets to be used, which improves the robustness of the projections, as initial condition uncertainty is a key source of forecast uncertainty in the short-to-medium term.

  5. Improved Regional Detail: A greater number of ensemble members can provide more information about regional details and variability, which can be crucial for policy-making and climate adaptation strategies at the local to regional scale.

However, increasing the number of ensemble members also requires significantly more computational resources and data storage, which is the reason why we applied the computational resource from the project lg87. We appreciate ACCESS-NRI with the project lg87 for providing us with additional computing resources, which enabled us to successfully complete the run by the end of Q2.

Experiment Name: SSP126/SSP245/SSP370/SSP585
People: CSIRO ACCESS model development team
Model: ACCESS-CM2
Configuration:
Initial conditions: Historical initial conditions
Run plan: 4 ensemble members for each SSP run. 86 years in total from 2015 to 2100
Simulation details:
Total KSUs required:
Total storage required:
Outputs:
Restarts:
Related articles:

Analysis:

Conclusion:

The additional computational resources enabled us to successfully complete a part of runs by the end of Q2.