Machine learning–based spatiotemporal modeling of earthquake occurrence for disaster mitigation: A case study of a seismically active region
Journal of Geospatial Science and Analytics · 31.03.2026 · 2 (1) · 27-40
DOI: 10.58524/jgsa.v2i1.119 ↗Kurz gefasst
Fachartikel in der Zeitschrift Journal of Geospatial Science and Analytics, online erschienen am 31.03.2026. Verfasst von Muhammad Harun Yahya und 4 weiteren Autor:innen. Der Volltext ist frei zugänglich (Open Access, CC BY-SA 4.0).
Abstract (Originalsprache)
Natural disasters, especially earthquakes, pose a significant threat that can cause significant damage to infrastructure and threaten public safety. West Java, a seismically active region, requires an effective mitigation system to minimize the impact of earthquakes. This study aims to develop an earthquake prediction model in West Java using a machine learning approach. The methods used are Long Short-Term Memory (LSTM) and Convolutional LSTM (ConvLSTM). LSTM is used to predict earthquake magnitude and depth and also prediction for 2025 - 2029, while ConvLSTM is used to predict the location of the earthquake epicenter. Historical earthquake data including location, magnitude, and depth are used to train the model. The results of the study showed that the ConvLSTM model was able to predict 104 earthquake events over the next five years with an RMSE of 7.907, MAPE of 32.65%, and of 0.245, while LSTM for magnitude produced an RMSE of 0.413, MAPE of 7.00%, and of 0.457, resulting in predictions with a magnitude range of 4.10 to 5.03 for all earthquake events for the next 5 years. Depth prediction with LSTM also showed significant results with an RMSE of 1.464, MAPE of 45.50%, and of 0.399, where most earthquakes were predicted to occur in Zone 2 and Zone 3, indicating that the depth of the earthquake tended to be shallow to medium in the next 5 years. These results can be used to increase public awareness and regional resilience through mapping earthquake-prone areas, as well as planning infrastructure strengthening and developing early warning systems.
Zitieren
APA
Yahya, M. H., Syazali, M., Syahbana, A. J., Umam, R., & Takahashi, H. (2026). Machine learning–based spatiotemporal modeling of earthquake occurrence for disaster mitigation: A case study of a seismically active region. Journal of Geospatial Science and Analytics, 2(1), 27-40. https://doi.org/10.58524/jgsa.v2i1.119
Vancouver
Yahya MH, Syazali M, Syahbana AJ, Umam R, Takahashi H. Machine learning–based spatiotemporal modeling of earthquake occurrence for disaster mitigation: A case study of a seismically active region. Journal of Geospatial Science and Analytics. 2026;2(1):27-40. doi:10.58524/jgsa.v2i1.119
BibTeX
@article{Yahya2026,
author = {Yahya, Muhammad Harun and Syazali, Muhamad and Syahbana, Arifan Jaya and Umam, Rofiqul and Takahashi, Hirotaka},
title = {Machine learning–based spatiotemporal modeling of earthquake occurrence for disaster mitigation: A case study of a seismically active region},
journal = {Journal of Geospatial Science and Analytics},
year = {2026},
volume = {2},
number = {1},
pages = {27--40},
doi = {10.58524/jgsa.v2i1.119}
}
EUVIMED