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

Abstract

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.

Analysis of Large Geomagnetically Induced Currents During the 7–8 September 2017 Storm: Geoelectric Field Mapping

“…GIC values during the geomagnetic storms are usually in the order of tens of amperes (e.g., Švanda et al., 2021; Torta et al., 2021) and our results confirms these observation (we do not observe values higher than 100 A). Moreover, GICs computations correspond to values mentioned for mid‐latitudes by (e.g., Albert et al., 2022; Bailey et al., 2022). For comparison, for Simpevarp‐2 power substation on 6 April 2000, Wik et al.…”

“…It would be interesting to apply GEDMap in this context. Moreover, the GIC problems at middle-and low-latitude transmission lines have increasingly gained attention (e.g., Albert et al, 2022;Bailey et al, 2022;Barbosa et al, 2015;Švanda et al, 2021;Torta et al, 2012;Tozzi et al, 2019;Zois, 2013). In the light of this task proposed procedure of the geoelectric field mapping, not based on SECS methods (appropriate for high latitudes) can be tested and applied.…”

Analysis of Long‐Term GIC Measurements in Transformers in Austria

“…Nevertheless, with the recent storms we now have sufficient data to reliably calculate the GICs in the Austrian power grid, which is unfortunately low for studying larger GIC events, however due to the relaxation of maximum measurement limits in the past year, we now have sufficient data to estimate GIC levels from recent storms. A historical analysis has been carried out in another recent study (Bailey et al., 2022), giving an estimate for the largest GICs during the 2003 Halloween storm that likely ranged from at least 30 A up to 60 A.…”

Space Weather Disturbances in Non‐Stormy Times: Occurrence of dB/dt Spikes During Three Solar Cycles

“…Our choice of threshold for spike detection (100 nT/min) can be motivated by previous investigations, for example, by a recent study of the power grid in Sweden, where a horizontal component dB / dt down to 30 nT/min can cause detectable GICs, reminiscent of grounding faults, and 200–500 nT/min can cause a temporary loss of individual transformers (Rosenqvist et al., 2022). In another recent study relating dB/dt variations to GICs in Austrian infrastructure, it was found that dB / dt of the order of 100 nT/min can be associated with cumulative GICs (over an hour at selected transformer stations) larger than 200 Ah (Bailey et al., 2022). Weygand (2021) listed dB / dt for three well‐studied previous geomagnetic storms, and found that peak dB / dt values of ∼60 nT/min can correspond to GIC magnitudes up to 40 A.…”