New Findings From Explainable SYM‐H Forecasting Using Gradient Boosting Machines. Issue 8 (23rd August 2022)

• Record Type: Journal Article
• Authors: Iong, Daniel; Chen, Yang; Toth, Gabor; Zou, Shasha; Pulkkinen, Tuija; Ren, Jiaen; Camporeale, Enrico; Gombosi, Tamas
• Is Part Of: Space weather. Volume 20:Issue 8(2022)
• Page Start: n/a
• Page End: n/a
• Publication Date: 2022-08-23
• Subjects: Space environment -- Periodicals; 551.509992
• DOI: 10.1029/2021SW002928
• Languages: English
• ISSNs: 1542-7390

Abstract

Abstract: In this work, we develop gradient boosting machines (GBMs) for forecasting the SYM‐H index multiple hours ahead using different combinations of solar wind and interplanetary magnetic field (IMF) parameters, derived parameters, and past SYM‐H values. Using Shapley Additive Explanation values to quantify the contributions from each input to predictions of the SYM‐H index from GBMs, we show that our predictions are consistent with physical understanding while also providing insight into the complex relationship between the solar wind and Earth's ring current. In particular, we found that feature contributions vary depending on the storm phase. We also perform a direct comparison between GBMs and neural networks presented in prior publications for forecasting the SYM‐H index by training, validating, and testing them on the same data. We find that the GBMs yield a statistically significant improvement in root mean squared error over the best published black‐box neural network schemes and the Burton equation. Plain Language Summary: Forecasting geomagnetic indices is crucial for mitigating potential effects of severe geomagnetic storms on critical infrastructures such as power grids. In this work, we adopt a machine learning method for SYM‐H prediction hours ahead with various combinations of solar wind & interplanetary magnetic field parameters, past SYM‐H values, and other derived parameters. The feature importance quantification that we derive provides important, new insight into the complex relationship between the solar wind and the Earth's ring current. Key Points: We adapt gradient boosting machines for forecasting the SYM‐H index multiple hours ahead We quantify feature contributions using Shapley additive explanation values to explain model predictions Our proposed method has similar accuracy to existing methods, while being more interpretable