Forecasting the Probability of Large Rates of Change of the Geomagnetic Field in the UK: Timescales, Horizons, and Thresholds. Issue 9 (22nd September 2021)

• Record Type: Journal Article
• Authors: Smith, A. W.; Forsyth, C.; Rae, I. J.; Garton, T. M.; Bloch, T.; Jackman, C. M.; Bakrania, M.
• Is Part Of: Space weather. Volume 19:Issue 9(2021)
• Page Start: n/a
• Page End: n/a
• Publication Date: 2021-09-22
• Subjects: GICs -- space weather -- forecasting -- machine learning -- neural networks -- model validation; Space environment -- Periodicals; 551.509992
• DOI: 10.1029/2021SW002788
• Languages: English
• ISSNs: 1542-7390

Abstract

Abstract: 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. Plain Language Summary: Geomagnetically induced currents (GICs) are external electrical currents that can be created in power lines and pipe networks as a result of variability in near‐Earth space. GICs are likely to be created when the ground magnetic field is changing rapidly. We test three different machine learning models to see whether they can predict if the ground magnetic field at a site in the UK is going to be highly variable in the future. We find that the models all perform excellently, providing useful forecasts as to when the rate of change of the magnetic field will be elevated. Key Points: Three neural network variants can use solar wind inputs to provide skillful and reliable probabilistic forecasts of large dB/dt in the UK The forecast skill/reliability increases with forecast horizon, maximizing at a horizon of 180 min Increasing the volume of input solar wind input data without increasing the model complexity does not boost performance