Generating Swarm Satellite Data using LSTM and GAN-based models for Detecting Seismic Precursors

Arzaan Ahmed Kankudti; Y Bi

doi:10.1145/3719384.3719470

Generating Swarm Satellite Data using LSTM and GAN-based models for Detecting Seismic Precursors

Arzaan Ahmed Kankudti
, Y Bi

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

21 Downloads (Pure)

Abstract

Precise understanding of the evolution mechanisms of earthquakes remains challenging in geophysics,
seismology and artificial intelligence (AI) related areas. This work leverages the advance of deep learning
predictive (Long Short-Term Memory (LSTM)) and generative (Generative Adversarial Networks (GANs))
approaches to develop two models for generating synthetic time series data based on Swarm satellite data,
which is then used for detecting anomalies over seismic evolution. Our results demonstrated the LSTM-based
model is capable of making use of the dependence learnt from time-series data to predict synthetic data. By
contrast, although the GAN-based model is able to capture the data distribution of the time series, it is not
able to discriminate non-informative values produced by the generator very well. These findings highlight
both the promise and challenges associated with applying deep learning techniques to generate synthetic
electromagnetic satellite data, underscoring the potential of deep learning approaches for seismic anomaly
detection.

Original language	English
Title of host publication	AICCC '24: Proceedings of the 2024 7th Artificial Intelligence and Cloud Computing Conference
Publisher	Association for Computing Machinery
Pages	596-602
Number of pages	7
ISBN (Electronic)	9798400717925
DOIs	https://doi.org/10.1145/3719384.3719470
Publication status	Published online - 9 Jul 2025
Event	2024 7th Artificial Intelligence and Cloud Computing Conference - Tokyo, Japan, Tokyo, Japan Duration: 14 Dec 2024 → 16 Dec 2024

Publication series

Name	ACM International Conference Proceeding Series

Conference

Conference	2024 7th Artificial Intelligence and Cloud Computing Conference
Abbreviated title	AICCC’25
Country/Territory	Japan
City	Tokyo
Period	14/12/24 → 16/12/24

Bibliographical note

Publisher Copyright:
© 2024 Copyright held by the owner/author(s).

Funding

This study is supported by Dragon 5 project (ID: 59308) that is part of the European Space Agency s (ESA) collaboration with China s Ministry of Science and Technology (MOST).

Funders
European Space Agency - ESA

Keywords

LSTM
GAN
Generative AI
Synthetic data
Seismic Precursors

Access to Document

10.1145/3719384.3719470

3719384.3719470Final published version, 1.82 MBLicence: CC BY

Cite this

Kankudti, A. A., & Bi, Y. (2025). Generating Swarm Satellite Data using LSTM and GAN-based models for Detecting Seismic Precursors. In AICCC '24: Proceedings of the 2024 7th Artificial Intelligence and Cloud Computing Conference (pp. 596-602). (ACM International Conference Proceeding Series). Association for Computing Machinery. Advance online publication. https://doi.org/10.1145/3719384.3719470

@inproceedings{0208f4793c3c4bdfafcedb96e594dddc,

title = "Generating Swarm Satellite Data using LSTM and GAN-based models for Detecting Seismic Precursors",

abstract = "Precise understanding of the evolution mechanisms of earthquakes remains challenging in geophysics, seismology and artificial intelligence (AI) related areas. This work leverages the advance of deep learning predictive (Long Short-Term Memory (LSTM)) and generative (Generative Adversarial Networks (GANs)) approaches to develop two models for generating synthetic time series data based on Swarm satellite data, which is then used for detecting anomalies over seismic evolution. Our results demonstrated the LSTM-based model is capable of making use of the dependence learnt from time-series data to predict synthetic data. By contrast, although the GAN-based model is able to capture the data distribution of the time series, it is not able to discriminate non-informative values produced by the generator very well. These findings highlight both the promise and challenges associated with applying deep learning techniques to generate synthetic electromagnetic satellite data, underscoring the potential of deep learning approaches for seismic anomaly detection.",

keywords = "LSTM, GAN, Generative AI, Synthetic data, Seismic Precursors",

author = "Kankudti, \{Arzaan Ahmed\} and Y Bi",

note = "Publisher Copyright: {\textcopyright} 2024 Copyright held by the owner/author(s).; 2024 7th Artificial Intelligence and Cloud Computing Conference, AICCC{\textquoteright}25 ; Conference date: 14-12-2024 Through 16-12-2024",

year = "2025",

month = jul,

day = "9",

doi = "10.1145/3719384.3719470",

language = "English",

series = "ACM International Conference Proceeding Series",

publisher = "Association for Computing Machinery",

pages = "596--602",

booktitle = "AICCC '24: Proceedings of the 2024 7th Artificial Intelligence and Cloud Computing Conference",

address = "United States",

}

Kankudti, AA & Bi, Y 2025, Generating Swarm Satellite Data using LSTM and GAN-based models for Detecting Seismic Precursors. in AICCC '24: Proceedings of the 2024 7th Artificial Intelligence and Cloud Computing Conference. ACM International Conference Proceeding Series, Association for Computing Machinery, pp. 596-602, 2024 7th Artificial Intelligence and Cloud Computing Conference, Tokyo, Japan, 14/12/24. https://doi.org/10.1145/3719384.3719470

Generating Swarm Satellite Data using LSTM and GAN-based models for Detecting Seismic Precursors. / Kankudti, Arzaan Ahmed; Bi, Y.
AICCC '24: Proceedings of the 2024 7th Artificial Intelligence and Cloud Computing Conference. Association for Computing Machinery, 2025. p. 596-602 (ACM International Conference Proceeding Series).

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

TY - GEN

T1 - Generating Swarm Satellite Data using LSTM and GAN-based models for Detecting Seismic Precursors

AU - Kankudti, Arzaan Ahmed

AU - Bi, Y

PY - 2025/7/9

Y1 - 2025/7/9

N2 - Precise understanding of the evolution mechanisms of earthquakes remains challenging in geophysics, seismology and artificial intelligence (AI) related areas. This work leverages the advance of deep learning predictive (Long Short-Term Memory (LSTM)) and generative (Generative Adversarial Networks (GANs)) approaches to develop two models for generating synthetic time series data based on Swarm satellite data, which is then used for detecting anomalies over seismic evolution. Our results demonstrated the LSTM-based model is capable of making use of the dependence learnt from time-series data to predict synthetic data. By contrast, although the GAN-based model is able to capture the data distribution of the time series, it is not able to discriminate non-informative values produced by the generator very well. These findings highlight both the promise and challenges associated with applying deep learning techniques to generate synthetic electromagnetic satellite data, underscoring the potential of deep learning approaches for seismic anomaly detection.

AB - Precise understanding of the evolution mechanisms of earthquakes remains challenging in geophysics, seismology and artificial intelligence (AI) related areas. This work leverages the advance of deep learning predictive (Long Short-Term Memory (LSTM)) and generative (Generative Adversarial Networks (GANs)) approaches to develop two models for generating synthetic time series data based on Swarm satellite data, which is then used for detecting anomalies over seismic evolution. Our results demonstrated the LSTM-based model is capable of making use of the dependence learnt from time-series data to predict synthetic data. By contrast, although the GAN-based model is able to capture the data distribution of the time series, it is not able to discriminate non-informative values produced by the generator very well. These findings highlight both the promise and challenges associated with applying deep learning techniques to generate synthetic electromagnetic satellite data, underscoring the potential of deep learning approaches for seismic anomaly detection.

KW - LSTM

KW - GAN

KW - Generative AI

KW - Synthetic data

KW - Seismic Precursors

UR - https://www.scopus.com/pages/publications/105011717132

U2 - 10.1145/3719384.3719470

DO - 10.1145/3719384.3719470

M3 - Conference contribution

T3 - ACM International Conference Proceeding Series

SP - 596

EP - 602

BT - AICCC '24: Proceedings of the 2024 7th Artificial Intelligence and Cloud Computing Conference

PB - Association for Computing Machinery

T2 - 2024 7th Artificial Intelligence and Cloud Computing Conference

Y2 - 14 December 2024 through 16 December 2024

ER -

Generating Swarm Satellite Data using LSTM and GAN-based models for Detecting Seismic Precursors

Abstract

Publication series

Conference

Bibliographical note

Funding

Keywords

Access to Document

Other files and links

Fingerprint

Cite this