Comparison of Deep Learning Techniques to Model Connections Between Solar Wind and Ground Magnetic Perturbations

Keesee, A. M.; Pinto, V. A.; Coughlan, Michael J.; Lennox, Connor; Mahmud, Shaad; Connor, H. K.

doi:10.3389/fspas.2020.550874

Cited by 30 publications

(87 citation statements)

References 14 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…A future large-scale thermospheric density data base can be used in studies involving Machine Learning (ML) applications. ML studies have become very popular in the field of Earth and Space Sciences in the past decade (e.g., Keesee et al, 2020;Smith et al, 2020;Bortnik and Camporeale, 2021;Haines et al, 2021). For example, historical data sets (geomagnetic and solar indices, sunspot numbers) prior to 2000 can be used for training a model to predict storm drivers and the subsequent global thermospheric density and orbital drag of a LEO satellite in a given location (Licata et al, 2020;2021c).…”

Section: Discussionmentioning

confidence: 99%

The Current State and Future Directions of Modeling Thermosphere Density Enhancements During Extreme Magnetic Storms

Oliveira

Zesta

Mehta

et al. 2021

Front. Astron. Space Sci.

View full text Add to dashboard Cite

Satellites, crewed spacecraft and stations in low-Earth orbit (LEO) are very sensitive to atmospheric drag. A satellite’s lifetime and orbital tracking become increasingly inaccurate or uncertain during magnetic storms. Given the planned increase of government and private satellite presence in LEO, the need for accurate density predictions for collision avoidance and lifetime optimization, particularly during extreme events, has become an urgent matter and requires comprehensive international collaboration. Additionally, long-term solar activity models and historical data suggest that solar activity will significantly increase in the following years and decades. In this article, we briefly summarize the main achievements in the research of thermosphere response to extreme magnetic storms occurring particularly after the launching of many satellites with state-of-the-art accelerometers from which high-accuracy density can be determined. We find that the performance of an empirical model with data assimilation is higher than its performance without data assimilation during all extreme storm phases. We discuss how forecasting models can be improved by looking into two directions: first, to the past, by adapting historical extreme storm datasets for density predictions, and second, to the future, by facilitating the assimilation of large-scale thermosphere data sets that will be collected in future events. Therefore, this topic is relevant to the scientific community, government agencies that operate satellites, and the private sector with assets operating in LEO.

show abstract

Section: Discussionmentioning

confidence: 99%

The Current State and Future Directions of Modeling Thermosphere Density Enhancements During Extreme Magnetic Storms

Oliveira

Zesta

Mehta

et al. 2021

Front. Astron. Space Sci.

View full text Add to dashboard Cite

show abstract

“…(2015), followed by Lotz and Cilliers (2015) and recently Keesee et al. (2020) and Tasistro‐Hart et al. (2021).…”

Section: Introductionmentioning

confidence: 99%

Forecasting GICs and Geoelectric Fields From Solar Wind Data Using LSTMs: Application in Austria

et al. 2022

View full text Add to dashboard Cite

The forecasting of local GIC effects has largely relied on the forecasting of dB/dt as a proxy and, to date, little attention has been paid to directly forecasting the geoelectric field or GICs themselves. We approach this problem with machine learning tools, specifically recurrent neural networks or LSTMs by taking solar wind observations as input and training the models to predict two different kinds of output: first, the geoelectric field components Ex and Ey; and second, the GICs in specific substations in Austria. The training is carried out on the geoelectric field and GICs modeled from 26 years of one‐minute geomagnetic field measurements, and results are compared to GIC measurements from recent years. The GICs are generally predicted better by an LSTM trained on values from a specific substation, but only a fraction of the largest GICs are correctly predicted. This model has a correlation with measurements of around 0.6, and a root‐mean‐square error of 0.7 A. The probability of detecting mild activity in GICs is around 50%, and 15% for larger GICs.

show abstract

“…This approach is in contrast to a more direct forecast of the ground magnetic field variability (c.f. Keesee et al., 2020), providing a simpler problem framework for a machine learning model to solve. We showed excellent correspondence between the target and model forecasts in the examples in Figures 3 and 4.…”

Section: Discussionmentioning

confidence: 99%

Forecasting the Probability of Large Rates of Change of the Geomagnetic Field in the UK: Timescales, Horizons, and Thresholds

Smith¹,

Forsyth²,

Rae

et al. 2021

Space Weather

View full text Add to dashboard Cite

Large geomagnetically induced currents (GICs) pose a risk to ground based infrastructure such as power networks. Large GICs may be induced when the rate of change of the ground magnetic field is significantly elevated. We assess the ability of three different machine learning model architectures to process the time history of the incoming solar wind and provide a probabilistic forecast as to whether the rate of change of the ground magnetic field will exceed specific high thresholds at a location in the UK. The three models tested represent feed forward, convolutional and recurrent neural networks. We find all three models are reliable and skillful, with Brier skill scores, receiver‐operating characteristic scores and precision‐recall scores of approximately 0.25, 0.95 and 0.45, respectively. When evaluated during two example magnetospheric storms we find that all scores increase significantly, indicating that the models work better during active intervals. The models perform excellently through the majority of the storms, however they do not fully capture the ground response around the initial sudden commencements. We attribute this to the use of propagated solar wind data not allowing the models notice to forecast impulsive phenomenon. Increasing the volume of solar wind data provided to the models does not produce appreciable increases in model performance, possibly due to the fixed model structures and limited training data. However, increasing the horizon of the forecast from 30 min to 3 h increases the performance of the models, presumably as the models need not be as precise about timing.

show abstract

Comparison of Deep Learning Techniques to Model Connections Between Solar Wind and Ground Magnetic Perturbations

Cited by 30 publications

References 14 publications

The Current State and Future Directions of Modeling Thermosphere Density Enhancements During Extreme Magnetic Storms

The Current State and Future Directions of Modeling Thermosphere Density Enhancements During Extreme Magnetic Storms

Forecasting GICs and Geoelectric Fields From Solar Wind Data Using LSTMs: Application in Austria

Forecasting the Probability of Large Rates of Change of the Geomagnetic Field in the UK: Timescales, Horizons, and Thresholds

Contact Info

Product

Resources

About