Title: Identification of sea level pressure anomaly patterns using Latent Dirichlet Allocation
Abstract: Mid-latitude circulation dynamics is often described in terms of weather regimes, which are typical field configurations of relevant observables (such as geopotential height or sea level pressure) determined by pattern recognition techniques. Each weather regime can be considered as a combination of basic synoptic objects, which are cyclones and anticyclones. Such combination makes it arduous to disentangle shifts in these structures recurrence and intensity and in particular those relevant to extreme events. Here we propose a change of perspective by applying Latent Dirichlet Allocation (LDA), a generative statistical model for collections of discrete data typically used as a topic model for text documents, to a set of snapshots featuring daily sea level pressure anomaly. LDA acts as a soft clustering technique providing a representation of a daily map in terms of a combination of motifs, which are latent spatial patterns, named motifs, inferred from the dataset. We notice that the motifs correspond to cyclones and anticyclones, the building blocks of weather regimes. Furthermore, we show that the weights provided by LDA are a practical way to characterize the effects of climate change on the recurrence and intensity of these structures and to identify potential precursors of extreme events.
The seminar will be held via Zoom. A link to the Zoom meeting will be posted at the morning of the event on the CIM mail list and when registering to the event.
Registration to the event.