S-SOM v1.0: a structural self-organizing map algorithm for weather typing

Abstract. This study proposes a novel structural self-organizing map (S-SOM) algorithm for synoptic weather typing. A feature of the S-SOM compared with traditional SOMs is its ability to deal input data spatial or temporal structures. In detail, search scheme best matching unit (BMU) in built based on similarity (S-SIM) index rather than by using Euclidean distance (ED). S-SIM enables BMU consider correlation space between states, such as locations highs lows, that impossible when ED. The performance evaluated multiple demo simulations clustering patterns over Japan ERA-Interim sea-level pressure data. results show S-SOM's superiority standard SOM ED (or ED-SOM) two respects: quality silhouette analysis and topological preservation error. Better versus consistent from different tests node-size configurations. performs better Pearson coefficient COR-SOM), though difference not clear it ED-SOM.