An increasing number of jurisdictions have issued long-term mitigation plans. Though the formats of these plans differ, they typically include one or several numerical scenarios. The present paper investigates what numerical model(s) are called upon to produce these scenarios, how they are used, and how models, modellers and, in some cases, other stakeholders contribute to the overall production of these Deep Decarbonization Strategies (DDS). Case studies in France, Brazil, the United States, Sweden show that for technical and institutional reasons large assemblages of models are typically used. The setting up of these assemblages requires significant work, especially when they are not (partly) inherited from past processes. The objective and configurations of these modelling assemblages can be very different, and the design of DDS is perceived to be successful when the political context and the objective of the process are aligned. Finally, there is always a tension between the granularity of the models -underpinning stakeholders' possibility to recognize their expertise in the modelling results -and the comprehensiveness of the model coverage -underpinning the planners' ability to get a systemic view of the proposed mitigation pathway(s). The paper concludes with tracks for further research on the issues raised by multi-model assemblages, which have not been considered by the academic literature.
Key policy insights• DDS raises unique challenges for policy making (e.g. long term horizon, integrating sectors beyond energy, evaluating economic impacts). • The setting up of these assemblages of models to evaluate DDS require time, resources, expecially when they are not (partly) inherited from past processes. • Policy makers should support the creation of 'hybrid' modelling collectives bringing together, over a long period of time, modellers and experts by major sectors. • Economic evaluations should be more at the heart of the process and consistent with sectorial analysis. • More effort should be conducted to address the tension between granularity of the models and policy makers' expectations.