Machine learning evidence map reveals global differences in adaptation action

A.J. Sietsma*, Emily Theokritoff, G.R. Biesbroek, Iván Villaverde Canosa, Adelle Thomas, Max Callaghan, Jan Minx, James D. Ford

*Corresponding author for this work

Research output: Contribution to journalArticleAcademicpeer-review

3 Citations (Scopus)

Abstract

Climate change adaptation policies are urgently needed, but the large volume and variety of evidence limits the ability of practitioners to make informed decisions. Here, we create an evidence map of adaptation policy research, selecting and categorizing 8,691 documents using state-of-the-art transformers-based machine learning models. We combine policy-relevant categories, such as the NATO-typology and governance levels, with automatically extracted locations and a structural topic model to provide a detailed global assessment of the tools governments are using to address climate change risks and impacts. We find that international-level policies, as well as policies in North America and much of the Global South, emphasize financial instruments, whereas national policies, particularly in Europe and Oceania, favor authority-based legislation. Collaborative approaches are most common at the local level. Despite a rapidly expanding evidence base overall, we note persistent geographic inequalities and limited evidence on information-based policies, policy implementation, and structural reforms.
Original languageEnglish
Pages (from-to)280-292
JournalOne Earth
Volume7
Issue number2
Early online date12 Jan 2024
DOIs
Publication statusPublished - Feb 2024

Fingerprint

Dive into the research topics of 'Machine learning evidence map reveals global differences in adaptation action'. Together they form a unique fingerprint.

Cite this