An Analysis of Typhoon Paths and Intensity Based on Feature Engineering and Machine Learning

Authors

  • Dongming Ma

DOI:

https://doi.org/10.54097/6p53gy30

Keywords:

K-means clustering algorithm, Average profile factor, Spearman's correlation coefficient.

Abstract

Given the increasing frequency of typhoon disasters and persistent prediction challenges, this study develops a typhoon classification model using machine learning to improve forecast accuracy. The research analyzed correlations between typhoon characteristics and environmental factors using Spearman coefficients, then visualized typhoon paths of different intensities. Recognizing multiple influencing factors, the study performed clustering analysis based on two key features: typhoon intensity and moving path, integrating these results for a comprehensive classification model. The model was validated by predicting typhoons in 2024. Results showed typhoon intensity significantly correlates positively with class and wind speed, while negatively with air pressure, and typhoon paths exhibit strong seasonal and geographical differences. K-means clustering determined the optimal number of clusters as 3 for intensity (silhouette coefficient: 0.545) and 4 for paths (silhouette coefficient: 0.615), indicating effective clustering. This model optimizes disaster response strategies and improves emergency management efficiency, offering substantial socioeconomic value for protecting lives and property in the context of increasing extreme weather events due to climate change.

Downloads

Download data is not yet available.

References

[1] Zhou Junhua, Shi Peijun, Chen Xuewen. Spatiotemporal differentiation of tropical cyclone activity in the Northwest Pacific from 1949 to 1999 [J]. Journal of Natural Disasters, 2002, (03): 44 - 49.

[2] Qin Dahuo, Sun Honglie, Sun Shu, Liu Yingjin. Development strategy for China's meteorological services from 2005 to 2020 [J]. Advances in Earth Science, 2005, (03): 268 - 274.

[3] Ma Suohong, Shen Xueshun, Gong Jiandong, et al. Development of the National Meteorological Center's tropical cyclone numerical forecasting system[J]. Meteorological Monthly, 2021, 47 (6): 685 - 692.

[4] Liao Fei, Li Wenting, Zhang Ziran, et al. Statistical characteristics of rapid intensity and path changes of tropical cyclones in the South China Sea from 1949 to 2017 [J]. Acta Oceanological Sinica, 2019, 41 (09): 126 - 135.

[5] Qian Qifeng, Wang Chuan, Xu Yajing, et al. A typhoon intensity estimation technique based on deep learning[J]. Meteorological Monthly, 2021, 47 (5): 601 - 608.

[6] Zhou Guanbo, Qian Qifeng, Xu Yinglong. Application of artificial intelligence-based typhoon intensity sudden change discrimination technology[J]. Journal of Natural Disasters, 2023, 32 (06): 96 - 103.

[7] Zhao Danfeng, Xu Qiang, Huang Zhou, et al. A typhoon track classification method based on similarity threshold learning [J]. Marine Environmental Science, 2022, 41 (4): 644 - 652.

[8] Zhou Xiaotian, Zhang Feng, Du Zhenhong, et al. Optimization of typhoon track forecasting based on neural network ensemble forecasting [J]. Journal of Zhejiang University (Science Edition), 2020, 47 (02): 196 - 202.

[9] Mary Balogun, Obeng-Gyasi Emmanuel. Association of Combined PFOA, PFOS, Metals and Allostatic Load on Hepatic Disease Risk [J]. J Xenobiot, 2024, 14 (2): 516 - 536. DOI: https://doi.org/10.3390/jox14020031

[10] Kristina-P Sinaga, Yang Miin-Shen. Unsupervised K-Means Clustering Algorithm [J]. IEEE Access, 2020, 880716 - 80727. DOI: https://doi.org/10.1109/ACCESS.2020.2988796

Downloads

Published

11-05-2025

How to Cite

Ma, D. (2025). An Analysis of Typhoon Paths and Intensity Based on Feature Engineering and Machine Learning. Highlights in Science, Engineering and Technology, 138, 195-202. https://doi.org/10.54097/6p53gy30