The COMPASS project recently released a “Report on understanding of weather types and large-scale climate drivers“.

This report presents a novel framework (outlined in Figure 1) to identify, assess and understand large-scale drivers which give rise to compound extremes, utilising weather patterns to bridge the gap. Improving understanding of the large-scale drivers of compound events is crucial to operationalising the climate and impact attribution of these events. However, it is challenging to directly link large-scale drivers of climate variability to compound events. Weather patterns, a set of atmospheric circulation types over a defined geographical region on a given day, are an established method used to assess the conditions that give rise to weather extremes. A significant amount research on weather patterns exists that links meteorological hazard to their associated impacts. Here, we propose a framework to extend this research to understand the large-scale drivers of compound events and pave the way for future climate and impact attribution studies.

This report presents a demonstration of the COMPASS framework for a case study of the 2013/2014 UK winter storms, a discussion of the findings and recommendations for future work. In addition, we have conducted research which begins to explore whether weather patterns remain consistent over long time periods to be reliably used in climate attribution studies. It aims to clarify what is meant by non-stationarity and identify the main challenges and uncertainties in assessing it.

Key findings – case study demonstration:

• There are statistically significant associations between a set of large-scale drivers of winter North Atlantic atmospheric circulation and a previously defined, well-used set of daily weather patterns for circulation over the UK (Neal et al., 2016).
• A review of the literature shows that weather patterns 29 (cyclonic south–south-westerly with a deep low centre of pressure west of Ireland) and 30 (cyclonic west-north-westerly with deep low pressure southeast of Iceland) from Neal et al. (2016) are linked to the 2013/2014 winter storms in the UK.
• Whilst the performance of the large-scale drivers in predicting the frequency of weather patterns 29 and 30 in January and February through an initial causal network is poor, a number of areas for further research and improvements have been identified.

Key findings – assessing the stationarity of weather patterns:

• We use a machine learning clustering algorithm to create historical weather patterns for the UK based on mean sea level pressure data. These patterns are compared across different time windows to a set of reference patterns to assess how they differ. Our findings suggest that some of the spatial patterns of generated clusters show changes across time windows over the last 40 years. However, it is currently not possible to determine whether these changes reflect real shifts, natural variability, or limitations of the method itself.
• This research highlights the complexity and uncertainty involved in assessing weather pattern stationarity and suggests that using weather patterns as constraints – and the assumption of stationarity – in climate attribution should be done with caution and warrants further research.

The report is available on the COMPASS research repository Zenodo at the following DOI: 10.5281/zenodo.17077013.

More information can be found in the following COMPASS project Deliverable:
Rushby, I., Etheridge, D., Perks, R., Bernie, D., Munday, G., Cotterill, D. (2025): Report on understanding of weather types and large-scale climate drivers. Horizon Europe project COMPASS. Deliverable D2.2.