North Atlantic SST regime classification from Self-Organising Maps and k-means clustering based on HadISST dataset (1950-2024)

North Atlantic sea surface temperature (SST) regime classification used in the analysis of Murphy, V., Yaniv, R., Lembo, V., Bordoni, S., & Hochman, A. (2026). North Atlantic SST drives sub-seasonal variability and predictability of the large-scale circulation. npj Climate and Atmospheric Science (under review). The classification was derived by training a 7x4 Self-Organising Map (SOM) on monthly North Atlantic SST anomalies from HadISST (1950-2024), then clustering the 28 SOM nodes into three regimes via k-means. The daily cluster assignment is obtained by projecting daily SST anomalies onto the SOM grid and assigning each day to its best-matching cluster. File contents: cluster_sequence (integer cluster label 1-3 per day), and per-cluster boolean temporal masks (cluster_1_temporal_mask, cluster_2_temporal_mask, cluster_3_temporal_mask) aligned to the same daily time axis from 1 January 1950 to 31 December 2024. This file is the regime classification used as input to the Wave Activity Flux, Local Wave Activity, and dynamical-systems analyses in the associated manuscript. The code used to produce this classification (SOM + k-means pipeline) will be deposited as a separate Zenodo record; for now, it is available from the corresponding author on reasonable request. Source SST data: HadISST (Met Office Hadley Centre), available at https://www.metoffice.gov.uk/hadobs/hadisst/.