Approaches for using CMIP projections in climate model ensembles to address the ‘hot model’ problem

Several recent generation global-climate models were found to have anomalously high climate sensitivities and may not be useful for certain applications. Four approaches for developing ensembles of climate projections for applications that address this issue are:

1. Using an “all models” approach;
2. Screening using equilibrium climate sensitivity and (or) transient climate response;
3. Bayesian model averaging; and
4. Using global warming levels.

Advantages and disadvantages of each approach are described by using example applications to study the effects of climate change on an imaginary at-risk species. Choosing the right approach is dependent on the location, goals, and system focus of each application and the risk-tolerance and resource-management context.