Classification of cyclones and anticyclones by a LDA model 

Romeo Hatchi

Latent Dirichlet Allocation (LDA) is a statistical model usually used in natural language processing and machine learning for topic modeling. We apply LDA to build a weather classification. Daily weather maps are transformed into word documents that are then viewed as combinations of topics, here spatial patterns. These patterns correspond to cyclones and anticyclones. Finally we can make a comparision with analogues used in attribution science. An attribution study calculates whether a particular weather phenomenon is more likely or more intense due to climate change.