ED‐ConvLSTM: A Novel Global Ionospheric Total Electron Content Medium‐Term Forecast Model

In this paper, we proposed an innovative encoder‐decoder structure with a convolution long short‐term memory (ED‐ConvLSTM) network to forecast global total electron content (TEC) based on the International GNSS Service (IGS) TEC maps from 2005 to 2018 with 1‐hr time cadence. The ED‐ConvLSTM model is...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Veröffentlicht in:Space Weather Jg. 20; H. 8
Hauptverfasser: Xia, Guozhen, Zhang, Fubin, Wang, Cheng, Zhou, Chen
Format: Journal Article
Sprache:Englisch
Veröffentlicht: Washington John Wiley & Sons, Inc 01.08.2022
Wiley
Schlagworte:
Aerospace environments
Astronautics
CNN
Coders
ConvLSTM
encoder‐decoder
Equatorial regions
Geomagnetic storms
Geomagnetism
Global navigation satellite system
Ionosphere
Ionospheric electron content
Ionospheric models
ionospheric TEC
Magnetic storms
Mathematical models
medium‐term global prediction
Modelling
Neural networks
Root-mean-square errors
Total Electron Content
ISSN:1542-7390, 1539-4964, 1542-7390
Online-Zugang:Volltext
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
Beschreibung
Zusammenfassung:In this paper, we proposed an innovative encoder‐decoder structure with a convolution long short‐term memory (ED‐ConvLSTM) network to forecast global total electron content (TEC) based on the International GNSS Service (IGS) TEC maps from 2005 to 2018 with 1‐hr time cadence. The ED‐ConvLSTM model is used to forecast TEC maps 1–7 days in advance through iterations. To investigate the model's performance, we compared the model with International Reference Ionosphere (IRI2016) model in 2014 and 2018, and compared the model with 1‐day Beijing University of Aeronautics and Astronautics (BUAA) model in 2018. The results show that our 7‐day ED‐ConvLSTM model (ED‐ConvLSTM model that forecasts 7 days in advance) outperforms IRI2016 in 2014 and 2018, and our 5‐day ED‐ConvLSTM model (ED‐ConvLSTM model that forecasts 5 days in advance) outperforms 1‐day BUAA model. Furthermore, the root mean square error (RMSE) from the 1‐day ED‐ConvLSTM model with respect to the IGS TEC maps decreases by 51.5% and 43%, respectively, in 2014 and 2018 compared with that from IRI2016 model. The RMSE from the 1‐day ED‐ConvLSTM model is 20.3% lower than that from the 1‐day BUAA model in 2018. In addition, our model has the highest RMSE in the Equatorial Ionospheric Anomaly (EIA) region, but can roughly predict the features and locations of EIA. However, the model fails to forecast localized TEC enhancement and the sudden ionospheric response to the geomagnetic storms. Overall, the model shows competitive performance in medium‐term global TEC maps prediction during geomagnetic quiet periods. Plain Language Summary The ionosphere is an important part of the earth's space environment, which can affect communication and satellite positioning. Accurately predicting the changes of ionospheric total electron content (TEC) will help improve communication quality and positioning accuracy. In this paper, we combine two different neural networks to predict the TEC on a global scale. The results show that our model can predict the TEC variations in the next seven days and the prediction accuracy is improved compared with the other two models. Key Points The 7‐day encoder‐decoder structure with a convolution long short‐term memory (ED‐ConvLSTM) model outperforms International Reference Ionosphere 2016 model in 2014 and 2018, and 5‐day ED‐ConvLSTM model outperforms 1‐day Beijing University of Aeronautics and Astronautics model in 2018 The 1‐day ED‐ConvLSTM model has the highest root mean square error in the Equatorial Ionospheric Anomaly (EIA) region, but can roughly predict the features and locations of EIA The 1‐day ED‐ConvLSTM model fails to forecast localized total electron content enhancement and the sudden ionospheric response to the geomagnetic storms
Bibliographie:ObjectType-Article-1
SourceType-Scholarly Journals-1
ObjectType-Feature-2
content type line 14
ISSN:1542-7390
1539-4964
1542-7390
DOI:10.1029/2021SW002959