Machine Learning–Enabled Climate Risk Scenarios for Local Government Decision-Making

Abstract

Local governments face increasingly complex and compounding climate risks, yet the analytical outputs of climate risk assessment often remain weakly connected to municipal decision-making routines. Although machine learning (ML) has advanced hazard forecasting and impact estimation by exploiting high-dimensional data and non-linear relationships, a persistent gap remains between predictive sophistication and governance usability. This gap is driven by limited interpretability, weak treatment of uncertainty, and insufficient alignment with scenario-based planning needs under non-stationary climate conditions. This paper provides a systematic literature review and conceptual synthesis of peer-reviewed research published between 2015 and 2025 on ML applications across the climate risk chain—hazard estimation, exposure analysis, impact assessment, and vulnerability mapping—with an explicit focus on local policy relevance. Rather than evaluating models solely through accuracy-oriented benchmarks, the review codes studies by (i) risk component, (ii) ML technique, (iii) uncertainty and explainability practices, and (iv) explicit linkage to municipal planning and investment decisions. Building on the synthesis, the paper proposes a governance-oriented framework that repositions ML from a predictive endpoint to a scenario-enabled decision-support infrastructure. The framework emphasizes interpretability as a legitimacy requirement, uncertainty representation as a core planning input, and institutional embedding within municipal cycles and accountability structures. The study contributes actionable recommendations for municipalities seeking to operationalize ML-enabled climate risk scenarios in adaptation planning and investment prioritization while strengthening transparency, accountability, and decision robustness.