DETECTION OF INTERPLANETARY ACTIVITY USING ARTIFICIAL NEURAL NETWORKS

Early detection of interplanetary activity is important when attemptin g to associate, with better accuracy, interplanetary phenomena with so lar activity and geomagnetic disturbances. However, for a large number of interplanetary observations to be done every day, extensive data a nalysis is required, leading to a delay in the detection of transient interplanetary activity. In particular, the interplanetary scintillati on (LPS) observations done with Ooty Radio Telescope (ORT) need extens ive human effort to reduce the data and to model, often subjectively, the scintillation power spectra. We have implemented an artificial neu ral network (ANN) to detect interplanetary activity using the power sp ectrum scintillation. The ANN was trained to detect the disturbed powe r spectra, used as an indicator of the interplanetary activity, and to recognize normal and strong scattering spectra from a large data base of IFS spectra. The coincidence efficiency of classification by the n etwork compared with the experts' judgement to detect the normal, dist urbed and strong scattering spectra was found to be greater than 80 pe r cent. The neural network, when applied during the LPS mapping progra mme to provide early indication of interplanetary activity, would sign ificantly help the ongoing efforts to predict geomagnetic disturbances .