Extreme weather events emerge out of the interaction of many physical processes, and understanding how is a key challenge in atmospheric science. Early statistical studies focused on single variables, trying to see if storm surges, wind speeds or temperatures, for example, could be used to predict episodes of extreme weather. Later, studies moved on to consider the coincidental occurrence of large or small values of several variables, such as wind speed and storm surge. In the past decade, researchers have come to realise that extreme weather often occurs without any precursory extremes in single or joint variables. It seems there’s no easy way to predict extreme weather with low-dimensional data.
Understanding the real causes of large-scale weather extremes, it now appears, requires a detailed study of all the relevant variables in space and time over extended regions. Heat waves or unusually cold days, for example, arise from the interaction of persistent atmospheric circulation regimes and patterns of temperature. Even soil moisture levels may play an important role. Perhaps the only way to make progress on disentangling causation in this high-dimensional setting is through numerical experiments with multi-parameter models.
In a new paper, LML External Fellow Davide Faranda and colleagues adopt this view, and combine a multivariate view of extremes with insights from the generic behaviour of chaotic systems. They propose a new methodology which exploits recent mathematical results from dynamical systems theory, and apply it to the study of temperature extremes (warm or cold days) over North America. The research builds upon previous studies which point to a significant coupling between atmospheric circulation variables and temperature in driving weather extremes. In particular, Faranda and colleagues study the coupling between the atmospheric circulation, tracked using the sea-level pressure maps and the thermodynamics, via 2-m temperature maps . They determine that winter temperature extremes are largely due to the atmospheric circulation while thermodynamics effects drive summer extremes. The authors hope this study will demonstrate the applicability of a novel analytical approach to multivariate atmospheric dynamics. They plan to apply this methodology to study the modifications induced by climate change in the coupling between dynamic and thermodynamical processes in the atmosphere.
The paper is available at https://link.springer.com/article/10.1007/s00382-019-05106-3