Citation

Asaad AAB. Spat. Inf. Res. 2022; 30(1): 143-153.

Copyright

(Copyright © 2022, Korean Spatial Information Society, Publisher Holtzbrinck Springer Nature Publishing Group)

DOI

10.1007/s41324-021-00415-4

PMID

unavailable

Abstract

The aim of the paper is to explore the characteristics of the Philippine storm tracks using Directional or Circular Statistics. In particular, it investigates the circular distributions of the temporal and spatial variables of the cyclones, including its correlation to non-circular variables such as typhoon intensity and the corresponding central atmospheric pressure. In addition to that, circular autocorrelation function was also generated up to 12th temporal lag (or 3 days) for the spatial variables; and the classification of the Philippine storm tracks was done using hierarchical clustering. The data was taken from the Japan Meteorological Agency's (JMA's) Regional Specialized Meteorological Center (RMSC), with time points from 1950 to 2021. The findings suggest that Circular Statistics can answer key questions that are useful for further modeling of the cyclones, specifically on identifying the appropriate model architecture and model features.

Language: en

Keywords

Agglomerative clustering; Climate change; Correlation; Directional statistics; Julia; R