Support Vector Machine Combined With Distance Correlation Learning For Dst Forecasting During Intense Geomagnetic Storms

The Earth’s magnetic field is dipoles field that extends tens of thousands of kilometers into space. It forms a protective layer of the Earth. The magnetic field can protect the Earth from the effect of the solar wind and cosmic rays directly. In the period of strong solar activity, the Solar flare will be more intense, especially during the sunspots maximum. CME and solar wind impact the Earth’s magnetosphere will lead to drastic changes of magnetic field in a short time, and it will affect human activities, change the space electric field and produce abnormal voltage that have a great impact on power systems and oil pipelines.In this paper we apply the Support Vector Machine (SVM) combined together with distance correlation (DC) to the forecasting of Dst index by using 80 intense geomagnetic storms (Dst ≤-100 nT) from 1995 to 2014. We also train the Neural Network (NN) and the Linear Machine (LM) to verify the effectiveness of SVM. The purpose for us to introduce DC is to make feature screening in input datasets that can effectively improve the forecasting performance of the SVM. For comparison, we estimate the correlation coefficients (CC), the Root Mean Square (RMS) errors, the absolute value of difference in minimum Dst (ΔDstmin) and the absolute value of difference in minimum time (ΔtDst) between observed Dst and predicted one. K fold Cross Validation is used to improve the reliability of the results. It is shown that DC-SVM model exhibits the best forecasting performance for all parameters when all 80 events are considered. The CC, the RMS error, the ΔDstmin, and theAtDst of DC-SVM are 0.95,16.8 nT,9.7 nT and 1.7 hour, respectively. We also can find that the Distance Correlation learning significantly improves the performance of all the models we built. For further comparison, we divide the 80 storm events into two groups depending on minimum value of Dst (group 1:-200< Dstmin ≤-100 nT; group2:). It is also found that the DC-SVM is better than other models in the two groups.