Key Messages

The report finds attribution of extreme events is most robust where high-quality observations, strong process understanding, and reliable climate models exist. Heat-related events, including large-scale land and marine heatwaves, can often be partly attributed to global warming. In contrast, attribution for storms, east coast lows, and long-term droughts remains beyond current capability. Limitations do not negate climate change influence but reflect insufficient evidence for specific event-level attribution.

Introduction

The report distinguishes between long-term detection and attribution of climate trends and the emerging field of extreme event attribution. While human-induced warming is unequivocal, attributing individual events to climate change is more complex and contested. This complexity has implications for legal and financial accountability, including potential attribution of damages to emissions sources. The report summarises expert consensus on the current state of attribution science in Australia.

Australian context

Extreme events arise from multiple interacting processes influenced by human-induced climate change. Attribution reliability varies by event type, region, and timescale, depending on observational data quality, scientific understanding, and model capability. Robust attribution requires all three elements.

The report categorises confidence levels (low, medium, high) across phenomena and timescales. High confidence exists for temperature-related extremes, particularly over 1–3 month periods. Moderate confidence applies to rainfall, drought, and fire-related phenomena in some contexts. Low confidence dominates short-duration rainfall, storms, tropical systems, and multi-year droughts, reflecting modelling and knowledge gaps.

Quantitative assessment shows climate models perform best for extreme heat and marine heatwaves, with high confidence across modelling, understanding, and observations. Conversely, phenomena such as severe convective storms and east coast lows consistently show low model capability and understanding, limiting attribution.

Summary of event types and modelling capability

Temperature extremes demonstrate the strongest attribution capability, with medium to high confidence even at shorter timescales. Rainfall and drought can be assessed with moderate confidence over monthly scales, though uncertainty increases for short-duration events due to coarse model resolution.

Multi-year drought attribution is particularly constrained by limited understanding of sustaining mechanisms. Similarly, short-duration rainfall and storm events are difficult to simulate accurately. Improvements in model resolution and process understanding are identified as necessary to enhance attribution reliability.

Decision framework for attribution

The report presents a decision tree to assess attribution feasibility and confidence. Attribution is only robust where events are well observed, models reproduce observed statistics, and physical processes are accurately simulated.

If these criteria are unmet, attribution may be impossible or require caveated statements based on observed trends rather than model-based evidence. This framework highlights that robust attribution remains rare for several event types and underscores the need for cautious interpretation of attribution claims.

Future opportunities

The report outlines three main strategies to improve attribution. First, expand and enhance observational datasets, including remote sensing and palaeoclimate proxies. Second, deepen process-based scientific understanding, particularly for poorly understood phenomena such as hail and long-term droughts. Third, significantly improve climate models, including adopting higher-resolution, convection-permitting models.

Additional approaches include multi-method attribution, integrating seasonal prediction tools, and developing impact-based attribution through collaboration between climate and impacts researchers. These actions aim to reduce uncertainty and improve practical applicability.

Summary

Extreme event attribution is advancing but remains uneven across event types in Australia. Confidence is highest for temperature extremes and lower for complex or short-duration phenomena. Attribution statements require careful interpretation due to data, modelling, and knowledge limitations. Continued investment in data, modelling, and interdisciplinary research is essential to improve robustness and support scientific, policy, and financial applications.