A deep learning approach to solar radio flux forecasting

Stevenson, E. C.; Rodríguez-Fernández, Víctor; Minisci, Edmondo; Camacho, David

doi:10.1016/j.actaastro.2021.08.004

Cited by 31 publications

(45 citation statements)

References 24 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…This work will hopefully help lead to future improvements made in prediction of not only F 10 but many of the other solar and geomagnetic indices. Data splitting into training and validation data was performed to be consistent with the previous work done by Stevenson et al (2022), but this validation set may not include values large and small enough to correctly validate the training data.…”

Section: Discussionmentioning

confidence: 99%

“…The sensitivity of model performance to this back average value will be investigated. Another method which considers the variation of lookback window, L, has been used as a method for promoting ensemble diversity (Stevenson et al, 2022;Oreshkin et al, 2019;Al-Shareef & Abbod, 2010). The sensitivity of model performance on lookback will also be investigated to determine the best lookback values to use during training (Table 1).…”

Section: Ensemble Diversitymentioning

confidence: 99%

“…Stevenson et al. (2022) discussed an approach for F 10 forecasting using neural networks. The authors implemented the network architecture created by Oreshkin et al.…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Probabilistic Solar Proxy Forecasting With Neural Network Ensembles

Daniell,

Mehta

2023

Space Weather

View full text Add to dashboard Cite

Space weather indices are used commonly to drive forecasts of thermosphere density, which affects objects in low‐Earth orbit (LEO) through atmospheric drag. One commonly used space weather proxy, F10.7cm, correlates well with solar extreme ultra‐violet (EUV) energy deposition into the thermosphere. Currently, the USAF contracts Space Environment Technologies (SET), which uses a linear algorithm to forecast F10.7cm. In this work, we introduce methods using neural network ensembles with multi‐layer perceptrons (MLPs) and long‐short term memory (LSTMs) to improve on the SET predictions. We make predictions only from historical F10.7cm values. We investigate data manipulation methods (backwards averaging and lookback) as well as multi step and dynamic forecasting. This work shows an improvement over the popular persistence and the operational SET model when using ensemble methods. The best models found in this work are ensemble approaches using multi step or a combination of multi step and dynamic predictions. Nearly all approaches offer an improvement, with the best models improving between 48% and 59% on relative MSE with respect to persistence. Other relative error metrics were shown to improve greatly when ensembles methods were used. We were also able to leverage the ensemble approach to provide a distribution of predicted values; allowing an investigation into forecast uncertainty. Our work found models that produced less biased predictions at elevated and high solar activity levels. Uncertainty was also investigated through the use of a calibration error score metric (CES), our best ensemble reached similar CES as other work.

show abstract

Section: Discussionmentioning

confidence: 99%

Section: Ensemble Diversitymentioning

confidence: 99%

See 1 more Smart Citation

Probabilistic Solar Proxy Forecasting With Neural Network Ensembles

Daniell,

Mehta

2023

Space Weather

View full text Add to dashboard Cite

show abstract

“…The use of machine learning for space weather indexes, such as the 𝐹 10.7 , which is a proxy for EUV light, dates back to the 90's, when a ANN was used to predict solar flux [78] or Gleisner et al [79] predicted geomagnetic storms using Time delay Neural Networks (TDNN). On the whole, due to the public availability of space weather indexes, and their natural periodic features, this area has an ample array of scientific publications [80][81][82][83][84] and for an extensive review of all machine learning applications in this particular field we recommend the excellent survey done by Camporeale [85].…”

Section: Thermospheric Density Mass Modelsmentioning

confidence: 99%

Machine Learning in Orbit Estimation: a Survey

Caldas¹,

Soares²

2022

Preprint

View full text Add to dashboard Cite

Since the late '50s, when the first artificial satellite was launched, the number of resident space objects (RSOs) has steadily increased. It is estimated that around 1 Million objects larger than 1 cm are currently orbiting the Earth, with only 30,000, larger than 10 cm, presently being tracked. To avert a chain reaction of collisions, termed Kessler Syndrome [1], it is indispensable to accurately track and predict space debris and satellites' orbit alike. Current physics-based methods have errors in the order of kilometres for 7 days predictions, which is insufficient when considering space debris that have mostly less than 1 meter. Typically, this failure is due to uncertainty around the state of the space object at the beginning of the trajectory, forecasting errors in environmental conditions such as atmospheric drag, as well as specific unknown characteristics such as mass or geometry of the RSO. Leveraging datadriven techniques, namely machine learning, the orbit prediction accuracy can be enhanced: by deriving unmeasured objects' characteristics, improving non-conservative forces' effects, and by the superior abstraction capacity that Deep Learning models have of modelling highly complex non-linear systems. In this survey, we provide an overview of the current work being done in this field.

show abstract

“…Moreover, the processing speed of this method was maximum during processing. However, the error produced by this method was maximumTo reduce the error value, Emma Stevensonet al [20] introduced the deep learning approach to estimate the SRF . Here, this method was extended to generate the forecast uncertainty estimates based on deep ensembles.…”

Section: Literature Surveymentioning

confidence: 99%

Deep Learning-Based Solar Activity Prediction Using Sunspot Number and Solar Radio Flux

Waykar

Kale

Karlekar

2022

Advances in Parallel Computing Algorithms, Tools and Paradigms

View full text Add to dashboard Cite

Nowadays, solar spectral irradiances are modelled by solar activity indices, which are used to identify the solar energy absorbed in the environment. This paper devises the Deep LSTM model for predicting solar activity using Sunspot Number (SSN) and Solar Radio Flux (SRF). The processing steps involved in the solar activity prediction are technical indicator extraction and solar activity prediction. In this paper, the solar indices are considered as an input of solar activity prediction, which is acquired from the solar cycle progression dataset. The technical indicators, like Simple Moving Average (SMA), Average True Range (ATR), Relative Strength Index (RSI), William’s %R, Stochastic %D, and Commodity Channel Index (CCI)are extracted for attaining a better prediction performance. In addition, the solar activity prediction is carried out using Deep LSTM based on SSN and SRF. The Deep LSTM is an effective deep learning technique that is widely utilized for prediction purposes since it has a better prediction ability. Moreover, the experimental result demonstrates that the devised Deep LSTM attained the minimum Mean Absolute Error (MAE), Mean Squared Error (MSE) and Root Mean Squared Error (RMSE) of 1.186, 2.869 and 1.693, correspondingly.

show abstract

A deep learning approach to solar radio flux forecasting

Cited by 31 publications

References 24 publications

Probabilistic Solar Proxy Forecasting With Neural Network Ensembles

Probabilistic Solar Proxy Forecasting With Neural Network Ensembles

Machine Learning in Orbit Estimation: a Survey

Deep Learning-Based Solar Activity Prediction Using Sunspot Number and Solar Radio Flux

Contact Info

Product

Resources

About