ISNet: Decomposed Dynamic Spatio‐Temporal Neural Network for Ionospheric Scintillation Forecasts

Abstract Accurate prediction of ionospheric scintillation is essential for ensuring the reliability of spaceborne and ground‐based radio wave technology infrastructures, including but not limited to navigation and communication systems. In this study, we propose a deep learning‐based Ionospheric Sci...

Full description

Saved in:
Bibliographic Details
Main Authors: Zhixu Gao, Yanhong Chen, Xianzhi Ao, Fulu Yue, Hong Chen, Hao Deng, Bingxian Luo, Xin Wang, Tianjiao Yuan
Format: Article
Language:English
Published: Wiley 2025-06-01
Series:Space Weather
Online Access:https://doi.org/10.1029/2024SW004239
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1849415811215654912
author Zhixu Gao
Yanhong Chen
Xianzhi Ao
Fulu Yue
Hong Chen
Hao Deng
Bingxian Luo
Xin Wang
Tianjiao Yuan
author_facet Zhixu Gao
Yanhong Chen
Xianzhi Ao
Fulu Yue
Hong Chen
Hao Deng
Bingxian Luo
Xin Wang
Tianjiao Yuan
author_sort Zhixu Gao
collection DOAJ
description Abstract Accurate prediction of ionospheric scintillation is essential for ensuring the reliability of spaceborne and ground‐based radio wave technology infrastructures, including but not limited to navigation and communication systems. In this study, we propose a deep learning‐based Ionospheric Scintillation Network (ISNet), which can predict the regional scintillation index S4 1 hour in advance. The novel ISNet decomposes the S4 index into background and disturbance fields and treats them separately. The model also employs specialized modules to capture the time‐delay effect between external disturbances and the associated scintillation. A flexible dynamic data reconstruction strategy is adopted, which allows the model to learn directly from high‐fidelity scintillation observations. Our results show that ISNet can provide accurate regional ionospheric scintillation forecasts in the low‐latitude regions of China. The RMSEs for weak, moderate, and strong scintillations in the test set are 0.053, 0.124, and 0.183, respectively. Compared with 12 other methods, ISNet displays a significant advantage in predicting moderate‐to‐strong scintillations. We also quantified the relative importance of input features to model predictions through the interpretable technique, which may provide valuable insights into model interpretation and feature selection.
format Article
id doaj-art-3dc7d68ba4aa497183c8c0c2fc220f12
institution Kabale University
issn 1542-7390
language English
publishDate 2025-06-01
publisher Wiley
record_format Article
series Space Weather
spelling doaj-art-3dc7d68ba4aa497183c8c0c2fc220f122025-08-20T03:33:25ZengWileySpace Weather1542-73902025-06-01236n/an/a10.1029/2024SW004239ISNet: Decomposed Dynamic Spatio‐Temporal Neural Network for Ionospheric Scintillation ForecastsZhixu Gao0Yanhong Chen1Xianzhi Ao2Fulu Yue3Hong Chen4Hao Deng5Bingxian Luo6Xin Wang7Tianjiao Yuan8Key Laboratory of Solar Activity and Space Weather National Space Science Center Chinese Academy of Sciences Beijing ChinaKey Laboratory of Solar Activity and Space Weather National Space Science Center Chinese Academy of Sciences Beijing ChinaKey Laboratory of Solar Activity and Space Weather National Space Science Center Chinese Academy of Sciences Beijing ChinaState Key Laboratory of Astronautic Dynamics Xi'an ChinaCollege of Informatics Huazhong Agricultural University Wuhan ChinaCollege of Informatics Huazhong Agricultural University Wuhan ChinaKey Laboratory of Solar Activity and Space Weather National Space Science Center Chinese Academy of Sciences Beijing ChinaKey Laboratory of Solar Activity and Space Weather National Space Science Center Chinese Academy of Sciences Beijing ChinaKey Laboratory of Solar Activity and Space Weather National Space Science Center Chinese Academy of Sciences Beijing ChinaAbstract Accurate prediction of ionospheric scintillation is essential for ensuring the reliability of spaceborne and ground‐based radio wave technology infrastructures, including but not limited to navigation and communication systems. In this study, we propose a deep learning‐based Ionospheric Scintillation Network (ISNet), which can predict the regional scintillation index S4 1 hour in advance. The novel ISNet decomposes the S4 index into background and disturbance fields and treats them separately. The model also employs specialized modules to capture the time‐delay effect between external disturbances and the associated scintillation. A flexible dynamic data reconstruction strategy is adopted, which allows the model to learn directly from high‐fidelity scintillation observations. Our results show that ISNet can provide accurate regional ionospheric scintillation forecasts in the low‐latitude regions of China. The RMSEs for weak, moderate, and strong scintillations in the test set are 0.053, 0.124, and 0.183, respectively. Compared with 12 other methods, ISNet displays a significant advantage in predicting moderate‐to‐strong scintillations. We also quantified the relative importance of input features to model predictions through the interpretable technique, which may provide valuable insights into model interpretation and feature selection.https://doi.org/10.1029/2024SW004239
spellingShingle Zhixu Gao
Yanhong Chen
Xianzhi Ao
Fulu Yue
Hong Chen
Hao Deng
Bingxian Luo
Xin Wang
Tianjiao Yuan
ISNet: Decomposed Dynamic Spatio‐Temporal Neural Network for Ionospheric Scintillation Forecasts
Space Weather
title ISNet: Decomposed Dynamic Spatio‐Temporal Neural Network for Ionospheric Scintillation Forecasts
title_full ISNet: Decomposed Dynamic Spatio‐Temporal Neural Network for Ionospheric Scintillation Forecasts
title_fullStr ISNet: Decomposed Dynamic Spatio‐Temporal Neural Network for Ionospheric Scintillation Forecasts
title_full_unstemmed ISNet: Decomposed Dynamic Spatio‐Temporal Neural Network for Ionospheric Scintillation Forecasts
title_short ISNet: Decomposed Dynamic Spatio‐Temporal Neural Network for Ionospheric Scintillation Forecasts
title_sort isnet decomposed dynamic spatio temporal neural network for ionospheric scintillation forecasts
url https://doi.org/10.1029/2024SW004239
work_keys_str_mv AT zhixugao isnetdecomposeddynamicspatiotemporalneuralnetworkforionosphericscintillationforecasts
AT yanhongchen isnetdecomposeddynamicspatiotemporalneuralnetworkforionosphericscintillationforecasts
AT xianzhiao isnetdecomposeddynamicspatiotemporalneuralnetworkforionosphericscintillationforecasts
AT fuluyue isnetdecomposeddynamicspatiotemporalneuralnetworkforionosphericscintillationforecasts
AT hongchen isnetdecomposeddynamicspatiotemporalneuralnetworkforionosphericscintillationforecasts
AT haodeng isnetdecomposeddynamicspatiotemporalneuralnetworkforionosphericscintillationforecasts
AT bingxianluo isnetdecomposeddynamicspatiotemporalneuralnetworkforionosphericscintillationforecasts
AT xinwang isnetdecomposeddynamicspatiotemporalneuralnetworkforionosphericscintillationforecasts
AT tianjiaoyuan isnetdecomposeddynamicspatiotemporalneuralnetworkforionosphericscintillationforecasts