Key messages

Uncertainty in climate planning arises from future human behaviour and the complexity of the climate system. Climate models provide valuable insights into how decisions may influence future outcomes but require careful interpretation. Effective adaptation depends on understanding assumptions, limitations and uncertainty to support informed decision-making under incomplete information.

Climate models

Climate models are detailed computational representations of the Earth’s systems, incorporating physics, chemistry and biology to simulate processes such as atmospheric circulation, heat transfer and carbon cycles. They generate projections, not predictions, offering plausible futures based on different assumptions. While highly valuable, their precision can be misinterpreted, leading to inappropriate use in decision-making.

Precision v Accuracy

Model outputs may appear precise, with results extending to many decimal places over long timeframes, but precision does not guarantee accuracy. Misunderstanding this distinction can lead to overconfidence in single simulations. Reliable interpretation requires considering multiple simulations to assess the likelihood of different outcomes rather than relying on isolated results.

Assumptions for simulations

Climate projections depend heavily on assumptions about future human behaviour, including economic development, technological change, energy systems and policy measures such as carbon pricing. These assumptions determine greenhouse gas emissions pathways, represented through scenarios such as RCPs and SSPs, and are the largest source of uncertainty in future climate estimates.

Describing climate processes: Variability

The climate system exhibits internal variability, which is unpredictable and occurs without external forcing, and forced variability driven by factors such as emissions and solar radiation. Models use ensembles to capture these dynamics, but variability remains difficult to quantify, particularly as it evolves with changing external conditions.

Describing climate processes: Representation in models

Differences in model design, including spatial resolution, included processes and institutional priorities, lead to variation in results even under identical emissions scenarios. Multi-model ensembles, such as CMIP, reflect this diversity, producing a wide range of possible outcomes that must be considered collectively rather than individually.

Limits of technology: Computing power

Computational constraints limit model resolution and complexity. Simulations require substantial resources, often taking months on supercomputers, yet still operate at coarse spatial scales (hundreds of kilometres). This necessitates approximations of key processes, particularly small-scale phenomena such as intense rainfall, introducing further uncertainty.

Uncertainty at local scales

Climate model outputs are typically too coarse for local decision-making, requiring downscaling to generate higher-resolution projections. Downscaling supports applications such as infrastructure planning and insurance pricing but introduces additional uncertainty, as multiple valid approaches can yield differing results. Only a limited subset of models can be downscaled due to resource constraints.

Working with the information we have

Decision-makers must work with incomplete information, recognising both known and unknown uncertainties. Interpreting projections requires analysing ensembles rather than single simulations to understand probability distributions. Two practical approaches are outlined: using scenario-based ‘storylines’ to explore plausible futures, and developing probabilistic estimates through advanced modelling techniques, albeit with higher cost and expertise requirements.

How should we use information from climate models?

Effective use of climate information involves assessing how projected changes affect specific risks, such as supply chains or asset exposure. Some variables, such as temperature, show clearer trends, while others, like precipitation, remain highly uncertain, particularly at local scales. Decision-makers should prioritise resilience across a range of outcomes and consider broader system impacts beyond direct climate effects.

Conclusion

Climate models are essential tools for informing adaptation and mitigation but remain conditional on assumptions and incomplete system representation. Understanding their limitations and uncertainties enables more effective use of projections. Acknowledging uncertainty is critical for managing risk and making robust decisions in climate planning.