Current climate change research faces a fundamental challenge: how to link complex physical knowledge of climate with practical risk management and adaptation planning. Scientific study published in the magazine Nature Climate Change presents an innovative climate taxonomy that successfully bridges this gap. This tool provides a common language for scientists, policymakers, and practitioners.

Bridging Two Worlds: Physical Science and Adaptation

The Intergovernmental Panel on Climate Change (IPCC) in its Sixth Assessment Report (AR6) highlighted the need for a stronger link between the physical science assessed by Working Group I (WGI) and the assessments of impacts, adaptation and vulnerability undertaken by Working Group II (WGII). The Climate Drivers Framework (CIDs) was introduced to facilitate this link. However, until now, a systematically applied, practice-oriented product that would consistently link these domains has been lacking.

The new climate taxonomy acts as an operational bridge. It allows for the translation of abstract physical climate indicators (e.g. average temperature changes) into concrete and actionable information (e.g. heat stress for ecosystems). This integrated approach is key for designing comprehensive and forward-looking adaptation strategies at regional, sectoral and temporal scales.

The Heart of the Taxonomy: Linking CID and Key Risks

The basis of the developed taxonomy is systematic pairing 35 climate drivers (CIDs) with 8 representative key risks (RKRs). CIDs represent physical climate conditions that directly affect society or ecosystems (e.g. drought, extreme heat, or flooding). RKRs, in turn, define eight groups of critical climate risks that intensify with warming, including impacts on food security, human health, or water security.

From the original 280 possible combinations, experts evaluated their real-world validity and eliminated 17 unlikely connections (for example, linking water safety risks to the open ocean). The resulting taxonomy thus contains 263 unique RKR-CID combinations.

Analysis of these combinations reveals important characteristics:

• Regional character: More than half (59 %) of the combinations are predominantly regional in nature. This reflects the fact that climate impacts are strongly determined by local vulnerability and exposure.
• Time and type balance: In terms of time scale, trends dominate in the horizon of years to decades (36 %) and hours to days (24 %). The types of changes are divided exactly in half: 50 % represent changes in climate extremes (e.g. cyclones) and 50 % represent changes in climate averages (e.g. gradual sea level rise). This balance shows the need to prepare not only for sudden extreme events, but also for long-term planning over a horizon of decades to centuries.

Comprehensive metadata and scientific evaluation

Each of the 263 unique linkages is enriched in the taxonomy with a robust set of structured metadata. This metadata includes spatial scale, type of change, temporal nature, and IPCC assessment of relevant subsystems for both human and natural systems.

Based on the IPCC scientific literature, the taxonomy determines the level of relevance of CIDs for individual sectors (high, medium/low, no/low confidence). For example, for the key risk „food security“, climate drivers such as drought and salinity are highly relevant for agricultural crops and pastures. The taxonomy also defines research needs and suggests the most appropriate modelling approaches for each link, ideally combining a physical model and an impact model (e.g. the Agricultural Production Systems Simulator – APSIM).

In building this first version, researchers piloted the analytical capabilities of large language models (e.g., OpenAI GPT-4 and Anthropic Claude) to preprocess metadata, subjecting their outputs to rigorous expert review and revision to ensure factual accuracy and consistency. In the future, they plan to use machine learning to accelerate scientific literature mapping.

Practical links to adaptation and financing

For successful application in practice, the taxonomy introduces specific examples of adaptation measures, which are divided according to three basic components of climate risk:

1. Measures aimed at hazard (threat): Physical or engineering measures reducing the intensity of a phenomenon, such as restoring mangroves against tidal waves or controlled recharge of aquifers.
2. Vulnerability-focused measures: Social, technical and institutional steps to increase resilience, such as promoting drought-resistant crops or precision irrigation.
3. Exposure-targeted measures: Changes in spatial organization and planning, such as the relocation of agriculture to less arid areas or land-use planning.

To facilitate financing, these measures were linked to investment opportunities under Climate Bonds Resilience Taxonomy (CBRT) 2024, the first classification system for credible climate resilience investments at the activity level. This allows investors and governments to easily identify projects suitable for capital markets. All adaptation linkages are also linked to specific goals of the United Nations Framework Convention on Climate Change (UNFCCC) under the Global Goal on Adaptation.

Mitigation synergies and transition to practice

The taxonomy also includes the area of climate change mitigation. It links individual risks to critical levels of global warming at which there is a risk of transition to high risk (Reasons for Concern – RFC). Almost half of the RKR-CID combinations become critical already at warming of 1.5 °C, while the other half reach the critical point above 2 °C. This fact emphasizes the urgency of the synergy between strict mitigation and proactive adaptation. The taxonomy therefore also provides examples of sectoral emission reduction potential for each combination.

In order to make these findings available to the general public and the scientific community, an open online platform has been created at https://climate-impact-taxonomy.iiasa.ac.at. Here, users can interactively filter data, search for specific links, and provide feedback via a structured form, thereby directly participating in the creation and improvement of future versions of this important tool.

This prototype climate taxonomy has thus laid a solid foundation for a more understandable, transparent and, above all, more actionable assessment of climate risks worldwide. JRI&CO2AI