Dangerous Scenarios – Designing climate change mitigation policy for uncertain futures

Inherent uncertainty in Land Use Change (LUC) has largely been ignored in the design of historic climate change adaptation and mitigation policy.  Instead, large or poorly understood uncertainties (typically those linked to future social/societal change) are handled with scenario analysis, with LUC outcomes evaluated for a pre-selected set of potential futures, each treated as alternative but certain outcomes.  Here we show that such approaches are dangerously simplistic; to the extent that some LUC designs, which present favourably in a scenario-analysis approach, risk worse outcomes than if we simply do nothing.  Adopting a net zero-focused tree-planting case study for Great Britain, we show that alternative assumptions about the future (in particular, in our example, future dietary change and food demand) can present policymakers with vastly differing land use change strategies to achieve cost-effective climate change mitigation. 

In this study we have used a large ensemble of an integrated assessment model projections to investigate the range of potential future LUC outcomes; comparing ensemble average outcomes (the Expected Value of decisionmakers); and the Conditional Value at Risk (CVaR – borrowed from financial economics, and quantifying average outcomes in the worst cases).  We show that tipping points in the LUC system can generate outcome distributions that are poorly represented by statistical measures typical used to quantify LUC outcomes; and hence risk misleading decisionmakers on the potential benefits and risks of a given LUC strategy.

The second part of our study examines the potential to improve on traditional LUC design approaches, by resolving a range of potential LUC futures.  Using the ensemble of LUC projections, we use multi-objective linear optimisation to identify LUC strategies that maximise different combinations of expected value and CvaR.  We find a large trade-off between strategies that priorities the best expected outcomes, and outcomes in the worst cases; and that these uncertainty-resolving design approaches offer significant improvement on traditional LUC design methods. 

More broadly, our results demonstrate the necessity for a shift in the way LUC schemes are designed and evaluated.  There is a need to incorporate better uncertainty quantification, and improved communication of uncertainty and risk to LUC decisionmakers; and as part of this seminar we’re keen to explore with you the opportunities to develop and apply alternative approaches and techniques.