How to Trigger a Super El Niño?

Quantifying Downstream Climate Impacts of Sea Surface Temperature Patterns in the Eastern Tropical Pacific Using Clustering

“…where b(i) represents the smallest average distance of a point (SST field in our study) to points in other clusters, and a(i) represents the average distance of a point from the other points in the cluster or group to which it belongs [21]. If the average of ( 2) is close to 1, then most, if not all, points are correctly assigned, and the clusters are well separated [51]. If the average coefficient is close to 0, then clusters may overlap, and some points may be between clusters.…”

“…However, the resulting clusters were not sensitive to the random initialization (the algorithm converged consistently on the same solution). K-means clustering is repeated for different numbers of clusters [21,51] because the underlying cluster structure is often not known.…”

“…If the average coefficient is close to 0, then clusters may overlap, and some points may be between clusters. If the value is negative, then many points are likely incorrectly assigned [21,51].…”