Clustering Indian Ocean tropical cyclone tracks by the standard deviational ellipse

Abstract

The standard deviational ellipse is useful to analyze the shape and the length of a tropical cyclone (TC) track. Cyclone intensity at each six-hour position is used as the weight at that location. Only named cyclones in the Indian Ocean since 1981 are considered for this study. The K-means clustering algorithm is used to cluster Indian Ocean cyclones based on the five parameters: x-y coordinates of the mean center, variances along zonal and meridional directions, and covariance between zonal and meridional locations of the cyclone track. Four clusters are identified across the Indian Ocean; among them, only one cluster is in the North Indian Ocean (NIO) and the rest of them are in the South Indian Ocean (SIO). Other characteristics associated with each cluster, such as wind speed, lifespan, track length, track orientation, seasonality, landfall, category during landfall, total accumulated cyclone energy (ACE), and cyclone trend, are analyzed and discussed. Cyclone frequency and energy of Cluster 4 (in the NIO) have been following a linear increasing trend. Cluster 4 also has a higher number of landfall cyclones compared to other clusters. Cluster 2, located in the middle of the SIO, is characterized by the long track, high intensity, long lifespan, and high accumulated energy. Sea surface temperature (SST) and outgoing longwave radiation (OLR) associated with genesis of TCs are also examined in each cluster. Cyclone genesis is co-located with the negative OLR anomaly and the positive SST anomaly. Localized SST anomalies are associated with clusters in the SIO; however, TC geneses of Cluster 4 are associated with SSTA all over the Indian Ocean (IO).