Keywords
scenario discovery, deep uncertainty, inequality, adaptation planning, simulation models
Start Date
15-9-2020 4:00 PM
End Date
15-9-2020 4:20 PM
Abstract
That climate change and the implementation of adaptation policies can exacerbate existing or create new inequalities is widely acknowledged. However, most model-based adaptation studies look either at inequality implications of climate uncertainties, or at inequality patterns resulting from taking certain policies. Here, we propose a modified scenario discovery approach to explore plausible inequality patterns resulting from the interplay of uncertainty and adaptation policies. The approach entails three steps: (i) performance of large-scale simulation experiments to take into account multiple realizations of the uncertainties and different policies, (ii) clustering of simulation results to identify distinct inequality patterns, (iii) identification of combinations of uncertainty and policy realizations that lead to each clusters of inequality patterns. We apply this approach to a model-based agricultural adaptation planning study in the Vietnam Mekong Delta. In step (i), we use Latin hypercube sampling to generate 10,000 combinations of realization of the uncertainty and policies. For step (ii), we empirically select an appropriate clustering algorithm and the number of clusters based on their degree of consistency, specifically by using the silhouette metric. In step (iii), we use the random forest classifier algorithm to map the uncertainty and the policy space with the output space (i.e., the clusters of inequality patterns). We then select a decision tree that has the highest degree of accuracy among the identified trees from the random forest. The selected decision tree shows which inequality pattern emerges conditional on which realization of the uncertainties happens and which policies are implemented. We showcase two use cases of this tree: as a tool for visual exploration of inequality patterns and as a metamodel to predict inequality patterns. From a policy perspective, understanding how inequality patterns emerge from the interplay of uncertainty and policies can help decision makers in preparing additional measures for ameliorating climate-induced inequalities.
A modified scenario discovery approach to explore inequality patterns in model-based adaptation planning
That climate change and the implementation of adaptation policies can exacerbate existing or create new inequalities is widely acknowledged. However, most model-based adaptation studies look either at inequality implications of climate uncertainties, or at inequality patterns resulting from taking certain policies. Here, we propose a modified scenario discovery approach to explore plausible inequality patterns resulting from the interplay of uncertainty and adaptation policies. The approach entails three steps: (i) performance of large-scale simulation experiments to take into account multiple realizations of the uncertainties and different policies, (ii) clustering of simulation results to identify distinct inequality patterns, (iii) identification of combinations of uncertainty and policy realizations that lead to each clusters of inequality patterns. We apply this approach to a model-based agricultural adaptation planning study in the Vietnam Mekong Delta. In step (i), we use Latin hypercube sampling to generate 10,000 combinations of realization of the uncertainty and policies. For step (ii), we empirically select an appropriate clustering algorithm and the number of clusters based on their degree of consistency, specifically by using the silhouette metric. In step (iii), we use the random forest classifier algorithm to map the uncertainty and the policy space with the output space (i.e., the clusters of inequality patterns). We then select a decision tree that has the highest degree of accuracy among the identified trees from the random forest. The selected decision tree shows which inequality pattern emerges conditional on which realization of the uncertainties happens and which policies are implemented. We showcase two use cases of this tree: as a tool for visual exploration of inequality patterns and as a metamodel to predict inequality patterns. From a policy perspective, understanding how inequality patterns emerge from the interplay of uncertainty and policies can help decision makers in preparing additional measures for ameliorating climate-induced inequalities.
Stream and Session
false