The dynamic fingerprint of ENSO on European summer weather regimes

The El-Niño Southern Oscillation (ENSO) is the main known predictor of interannual natural climate variability. It presents several known teleconnections to weather in different regions of the world. When it comes to the influence of ENSO on European climate, the existing literature is mainly focused on the variability of precipitation and, more recently, temperature. In most of these studies, the variability modes of the target variables are directly recovered with techniques such as empirical orthogonal functions, and the role of atmospheric dynamics is assessed a posteriori by considering composites of dynamical variables. Here we take on a different approach, and we directly assess the influence of ENSO on atmospheric dynamics over the Europe and North Atlantic (ENA) region. This is particularly relevant when considering summer, given the vulnerability of parts of this region to increasing risk of drought and summer heatwaves. To conduct this assessment, we use Latent Dirichlet Allocation (LDA), a generative statistical topic model traditionally used in natural language processing, which can also be utilized as a weather pattern recognition technique. In particular, we apply LDA to the ERA5 daily 500 hPa geopotential height anomaly field (Z500). As a result, we obtain a prescribed number of positive and negative Z500 anomaly patterns, each associated with a daily weight. The analysis of the weights time series for each LDA pattern shows a significant association between the ENSO phase and the importance of some Z500 patterns during summer over ENA. By clustering the pattern weights after conditioning on the ENSO phase, we show how this identified teleconnection affects precipitation and temperature variability over the ENA region.