Spatial Grouping of Tornado-Relevant Wind Regimes Areas in Indonesia to Enhance Disaster Risk Mitigation Capacity

Ardelita, Sela Naren (Mathematics Department, Universitas Negeri Surabaya, Surabaya, 60231, Indonesia)
Sofro, A'yunin (Mathematics Department, Universitas Negeri Surabaya, Surabaya, 60231, Indonesia)

Zero : Jurnal Sains, Matematika, dan Terapan
Vol 9, No 3 (2025): Zero: Jurnal Sains Matematika dan Terapan

Tornadoes are major weather hazards in Indonesia, where wind variability is important for assessing disaster risk and supporting energy planning. This study conducts a short-term (one-year) analysis by identify similarities in regional wind speed patterns using a time-series clustering approach, treating monthly average wind speeds in 2024 as proxies for tornado-relevant wind regimes rather than direct tornado occurrence data. Agglomerative hierarchical clustering is integrated with three distance measures-Dynamic Time Warping (DTW), Autocorrelation Function (ACF), and Short Time Series (STS)-and optimized using Brain Storm Optimization (BSO) to determine optimal distance weighting and cluster numbers. The results indicate that DTW provides the best performance, yielding a two-cluster solution with a Silhouette Coefficient of 0.5292. The first cluster exhibits relatively stable wind patterns, while the second shows higher temporal variability. This framework provides a data-driven basis for region-specific wind energy planning and tornado-adaptive infrastructure considerations in Indonesia.