Forecasting of Global Ionosphere Maps With Multi‐Day Lead Time Using Transformer‐Based Neural Networks
Abstract Ionospheric total electron content (TEC) is a key indicator of the space environment. Geophysical forcing from above and below drives its spatial and temporal variations. A full understanding of physical and chemical principles, available and well‐representable driving inputs, and capable c...
Saved in:
| Main Authors: | , , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Wiley
2024-02-01
|
| Series: | Space Weather |
| Subjects: | |
| Online Access: | https://doi.org/10.1029/2023SW003579 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1841536403425460224 |
|---|---|
| author | Chung‐Yu Shih Cissi Ying‐tsen Lin Shu‐Yu Lin Cheng‐Hung Yeh Yu‐Ming Huang Feng‐Nan Hwang Chia‐Hui Chang |
| author_facet | Chung‐Yu Shih Cissi Ying‐tsen Lin Shu‐Yu Lin Cheng‐Hung Yeh Yu‐Ming Huang Feng‐Nan Hwang Chia‐Hui Chang |
| author_sort | Chung‐Yu Shih |
| collection | DOAJ |
| description | Abstract Ionospheric total electron content (TEC) is a key indicator of the space environment. Geophysical forcing from above and below drives its spatial and temporal variations. A full understanding of physical and chemical principles, available and well‐representable driving inputs, and capable computational power are required for physical models to reproduce simulations that agree with observations, which may be challenging at times. Recently, data‐driven approaches, such as deep learning, have therefore surged as means for TEC prediction. Owing to the fact that the geophysical world possesses a sequential nature in time and space, Transformer architectures are proposed and evaluated for sequence‐to‐sequence TEC predictions in this study. We discuss the impacts of time lengths of choice during the training process and analyze what the neural network has learned regarding the data sets. Our results suggest that 12‐layer, 128‐hidden‐unit Transformer architectures sufficiently provide multi‐step global TEC predictions for 48 hr with an overall root‐mean‐square error (RMSE) of ∼1.8 TECU. The hourly variation of RMSE increases from 0.6 TECU to about 2.0 TECU during the prediction time frame. |
| format | Article |
| id | doaj-art-595c6acd630040e3a1c9e2f1b57461a0 |
| institution | Kabale University |
| issn | 1542-7390 |
| language | English |
| publishDate | 2024-02-01 |
| publisher | Wiley |
| record_format | Article |
| series | Space Weather |
| spelling | doaj-art-595c6acd630040e3a1c9e2f1b57461a02025-01-14T16:30:41ZengWileySpace Weather1542-73902024-02-01222n/an/a10.1029/2023SW003579Forecasting of Global Ionosphere Maps With Multi‐Day Lead Time Using Transformer‐Based Neural NetworksChung‐Yu Shih0Cissi Ying‐tsen Lin1Shu‐Yu Lin2Cheng‐Hung Yeh3Yu‐Ming Huang4Feng‐Nan Hwang5Chia‐Hui Chang6Mathematics National Central University Taoyuan TaiwanSpace Science and Engineering National Central University Taoyuan TaiwanComputer Science and Information Engineering National Central University Taoyuan TaiwanComputer Science and Information Engineering National Central University Taoyuan TaiwanComputer Science and Information Engineering National Central University Taoyuan TaiwanMathematics National Central University Taoyuan TaiwanComputer Science and Information Engineering National Central University Taoyuan TaiwanAbstract Ionospheric total electron content (TEC) is a key indicator of the space environment. Geophysical forcing from above and below drives its spatial and temporal variations. A full understanding of physical and chemical principles, available and well‐representable driving inputs, and capable computational power are required for physical models to reproduce simulations that agree with observations, which may be challenging at times. Recently, data‐driven approaches, such as deep learning, have therefore surged as means for TEC prediction. Owing to the fact that the geophysical world possesses a sequential nature in time and space, Transformer architectures are proposed and evaluated for sequence‐to‐sequence TEC predictions in this study. We discuss the impacts of time lengths of choice during the training process and analyze what the neural network has learned regarding the data sets. Our results suggest that 12‐layer, 128‐hidden‐unit Transformer architectures sufficiently provide multi‐step global TEC predictions for 48 hr with an overall root‐mean‐square error (RMSE) of ∼1.8 TECU. The hourly variation of RMSE increases from 0.6 TECU to about 2.0 TECU during the prediction time frame.https://doi.org/10.1029/2023SW003579TEC predictionneural networkTransformer |
| spellingShingle | Chung‐Yu Shih Cissi Ying‐tsen Lin Shu‐Yu Lin Cheng‐Hung Yeh Yu‐Ming Huang Feng‐Nan Hwang Chia‐Hui Chang Forecasting of Global Ionosphere Maps With Multi‐Day Lead Time Using Transformer‐Based Neural Networks Space Weather TEC prediction neural network Transformer |
| title | Forecasting of Global Ionosphere Maps With Multi‐Day Lead Time Using Transformer‐Based Neural Networks |
| title_full | Forecasting of Global Ionosphere Maps With Multi‐Day Lead Time Using Transformer‐Based Neural Networks |
| title_fullStr | Forecasting of Global Ionosphere Maps With Multi‐Day Lead Time Using Transformer‐Based Neural Networks |
| title_full_unstemmed | Forecasting of Global Ionosphere Maps With Multi‐Day Lead Time Using Transformer‐Based Neural Networks |
| title_short | Forecasting of Global Ionosphere Maps With Multi‐Day Lead Time Using Transformer‐Based Neural Networks |
| title_sort | forecasting of global ionosphere maps with multi day lead time using transformer based neural networks |
| topic | TEC prediction neural network Transformer |
| url | https://doi.org/10.1029/2023SW003579 |
| work_keys_str_mv | AT chungyushih forecastingofglobalionospheremapswithmultidayleadtimeusingtransformerbasedneuralnetworks AT cissiyingtsenlin forecastingofglobalionospheremapswithmultidayleadtimeusingtransformerbasedneuralnetworks AT shuyulin forecastingofglobalionospheremapswithmultidayleadtimeusingtransformerbasedneuralnetworks AT chenghungyeh forecastingofglobalionospheremapswithmultidayleadtimeusingtransformerbasedneuralnetworks AT yuminghuang forecastingofglobalionospheremapswithmultidayleadtimeusingtransformerbasedneuralnetworks AT fengnanhwang forecastingofglobalionospheremapswithmultidayleadtimeusingtransformerbasedneuralnetworks AT chiahuichang forecastingofglobalionospheremapswithmultidayleadtimeusingtransformerbasedneuralnetworks |