Climate change models simplify the complicated roles of physical, chemical, and biological processes in the earth system to represent possible future (and past) climates. Because different models simplify these processes in various ways, the collaborative “Coupled Model Intercomparison Project” (CMIP) helps compare model predictions in a standardized way, by evaluating each model under the same set of potential future scenarios (e.g., levels of greenhouse gas emissions and human activities). Some models, however, fall victim to the “Hot Model” problem – a problem, as the name suggests, where models predict future conditions that are significantly hotter than other lines of evidence suggest will occur. 

With issues like the “hot model” problem, how can decision-makers and researchers choose the best climate models to use in their work? A new publication co-authored by Alaska, North Central, and Southeast CASC scientists presents four different ways to deal with the “hot model” problem. 

As with any choice, each of the four approaches has advantages and disadvantages, making it important to choose based on the specific objectives of the study or decision, including the location, ecosystem type, resources, risk-tolerance, and management context. For example, the “All Models Approach” may be useful in risk-averse decision environments because it provides the entire range of possible future scenarios based on “all models.” But one caveat to this approach is that it may create scenarios where much planning time is spent discussing and considering unlikely scenarios. The authors illustrate how there is no single “best” approach by presenting a case study on how the four approaches can be used to assess the climate risk to “an imaginary at-risk species – the golden-billed raptor.” 

Climate model projections are widely used in research and policymaking. This work provides a reference for decision-makers to select the appropriate models for different research and planning goals based on considerations such as risk tolerance and management context, in addition to the location, objectives, and system focus of the application.