Interactive decision support and trade-off analysis for sustainable forest landscape planning under deep uncertainty

Abstract

Sustainable environmental management often involves long-term time horizons, multiple conflicting objectives, and by nature, is affected by different sources of uncertainty. Many sources of uncertainty, such as climate change or government policies, cannot be addressed using probabilistic models, and, therefore, they can be seen to contain deep uncertainty. In this setting, the variety of possible future states is represented as a set of scenarios lacking any information about the likelihood of occurring. Integrating deep uncertainty into multiobjective decision support increases complexity, calling for the elaboration of appropriate methods and tools. This paper proposes a novel interactive multi-scenario multiobjective approach to support decision-making and trade-off analysis in sustainable forest landscape planning under multiple sources of uncertainty. It includes new preference simulation models aimed at reducing the decision-maker's cognitive load and supporting the preference elicitation process. The proposed approach is applied in a case study of long-term forest landscape planning with four sustainability objectives in twelve scenarios and a forestry expert as the decision-maker. The approach is demonstrated to be efficient in exploring trade-offs in different scenarios, helping the expert gain deep insights into the problem, understand the consequences of alternative strategies, and find the most preferred robust strategy.

Visualizations for decision support in scenario-based multiobjective optimization

“…Even gathering new data through conducting a forest inventory before harvest scheduling cannot wholly remove existing uncertainty in available timber volumes. Since fitting the proper statistical distribution might not always be possible, this classifies as deep uncertainty (Walker et al 2013, Shavazipour & Stewart 2021, Shavazipour et al 2022.…”

“…In deep uncertainty, finding an optimal plan for an average or the most probable scenario (also called nominal or base-line scenario) may lead to failure (Shavazipour & Stewart 2021). Instead, identifying a robust plan that works relatively well in a broader range of scenarios is recommended, which requires specific uncertainty and robustness analyses (Lempert et al 2006, Shavazipour & Stewart 2021, Shavazipour et al 2022. Although, in the last decades, various methodologies and tools have been developed for dealing with deep uncertainty in other disciplines (Lempert et al 2006, Kasprzyk et al 2013, Herman et al 2015, Quinn et al 2017, Shavazipour, Kwakkel & Miettinen 2021, Shavazipour & Stewart 2023, only a few studies consider handling deep uncertainty in the forest management context, mainly in strategic planning (Yousefpour & Hanewinkel 2016, Radke et al 2017, Augustynczik & Yousefpour 2019, Radke et al 2020, Hörl et al 2020, Shavazipour et al 2022.…”