Prediction Methods For Solar Flare And Relativistic Electron On Geostationary Orbit

Geosynchronous Earth Orbit （GEO） is in the earth’s outer radiation belts,where energetic electrons enriched, more than400satellites are orbiting in thisarea, hence it is of vital importance to alert and predict GEO space environment.The flux of energetic electron will surge after the geomagnetic disturbance，which is one of the causes for the failure of satellites. So，it is important to predictthe electron flux on the geosynchronous orbit in order to prevent the satellite frombeing harmed. As the enhancement of electron flux is controlled by manyconditions as well as complex temporal and spatial evolution, statistical analysis tovarious data is needed to make the prediction model. In this work, energetic fluxdata from GOES^-12was statistically analyzed together with Geomagnetic Apindex and solar wind velocity and combination of input parameters was determined.An artificial neural network based model was built and trained with the datamentioned above. We tried to make the prediction of electron flux of E>2Mev oneday ahead and the result was good that the prediction efficiencies were0.766、0.808and0.882for2008^-2010.We also tried to make the prediction of electron flux of FY^-2 Satellites using theGOES^-12prediction result by analyzing their flux related characteristics. Thestatistics show that the E>2Mev electron flux measurement of the two kinds ofsatellite become similar when the flux was above10(8cm2day^-1 sr1). Because thedeep charge effect become significant when E>2Mev electron fluxexceed10(8cm2day^-1 sr1).Based on the GOES prediction, it’s reasonable to set itas an alert line when electron flux reaches10(8cm2day^-1 sr1)to predict thepossible harms that satellites can get.Using data from FY^-2D to predict E>2Mev electron flux is another work. Bycomparison the measurement of electron with E>2MeV by FY^-2D and GOES^-13on different location at geosynchronous orbit, energetic electrons showed thesimilar character of dawn-dusk asymmetry and the27.3-day periodicity, In thiswork, energetic electron flux data of E>2MeV from FY^-2D was statisticallyanalyzed together with Geomagnetic Ap index and solar wind velocity, andcombination of input parameters was determined. We constructed a predictionmodel based on RBF Neural Network, trained it with historical data via thecombination mention above. The prediction of electron daily fluence of E>2MeV at the orbit of FY^-2D one day ahead was made and the result show goodcoordination with the test data that the prediction efficiencies were0.73and0.75,the mean errors were0.024and0.019for2010^-2011respectively. Theevidence indicates that the FY^-2D energetic electron data could be applied in boththeoretical and practical work and the prediction algorithm of electron is alsopractical.Two main works have been conducted to predict X-ray solar flares, one is theprediction of X-Ray flux, the other is the prediction of lasting time of descendingphase of flares. We adopt the23rd cycle’s data of soft X-ray flux from Goes-8toforcast the size of X-Class flares’ peak and the end time of X-Class flares in themethod of Numerical Fitting. The result of analyzing the X-Class flares show thatthe X-Class flares’ size of peak can be predicted17minutes ahead and the end timecan be forcasted60minutes ahead. Judging from the forecast results, the predictionmethod has a certain validity and practicability.