Reframing model priorities

Abstract

Earth systems models — physical climate models that incorporate biogeochemistry, such as the carbon cycle — are used for understanding and projecting human impacts on climate. Over the years, models have become more complex as greater understanding allows a more realistic representation of the Earth system and processes. This increase in realism aimed to improve prediction accuracy but it is increasingly clear that it has resulted in increased uncertainty, which has important implications for the usability of results in decision-making. Model development needs a new approach according to Matthew Smith, of Microsoft’s Computational Science Laboratory, Cambridge, UK, and colleagues. The focus should shift from improving real-world representation to understanding the level of complexity required for predictive, actionable science and effective decision-making. The authors suggest a new ‘balanced complexity’ approach: a formal method enabling the identification of key areas of uncertainty, key earth-system components and the level of complexity required to most effectively address specific questions. Adopting such an approach should greatly improve the ability of models to provide policy-relevant information for more effective decision making. BW CLIMATE