Chapter 09 · Sustainability & AISustainability & AI

AI climate risk modelling

Meaning statusEstablishedSource recordDirect source requestedWhy these are different

Definition

Applying machine learning to assess physical and transition climate risks — hazards, exposure and financial impact — for assets, portfolios and supply chains.

References

NGFSNetwork for Greening the Financial System — climate scenarios

This source provides part of the technical or institutional basis for the definition.

Climate InformaticsClimate Informatics workshop series (since 2011)

This source supports the explanation of how the term is applied, measured or governed in practice.

Overview

What it means

Machine learning enhances hazard modelling (downscaling projections, detecting extremes), exposure mapping from imagery, and translation into financial loss estimates. The results feed scenario analyses aligned with frameworks such as the NGFS reference scenarios used by central banks and financial institutions.

How it is used

Banks, insurers and asset managers use AI-assisted models for climate stress-testing and disclosure; corporates use them for site and supply-chain resilience planning. Model opacity and scenario dependence are recognised limitations requiring validation.

Why it matters

Climate risk is a data-sparse, deep-uncertainty problem — exactly where AI both helps and overpromises. Outputs inform real capital allocation, so model governance and honest uncertainty treatment are as material as the projections themselves.