Time Series Modeling and Short-Time Forecasting for DST Index of Geomagnetic Storm

Magnetic storm is a significant magnetic disturbance, which has some influence in communication system and power system. Its intensity is always measured by DST and it is often predicted by the application of neural network nonlinear simulation and differential equation so on. Most of these methods need the data collected several hours before magnetic storm besides DST data. Here we proposed a new method which only depends on the information of DST index. Based on the time-serial theory and the index character, an ARIMA model was established. The model used on predicting magnetic storm evolution in several hours fit well and the relative error in shorter time is small, which can be used to predict the size of the DST index in the next few hours, and then measure the intensity of storms in the short term.