Modelling Geomagnetic Activity Data

Strong geomagnetic activity is a hazard to electronics and electric power facilities. Assessment of the actual geomagnetic activity level from local magnetometer monitoring therefore is of importance for risk assessment but also in earth sciences and exploration. Wavelet based signal processing methods are applied to extract meaningful information from magnetic field time series in a noisy environment. Using a proper feature vector a local geomagnetic activity index can be derived under not ideal circumstances using computer intelligence methods. Locally linear radial basis function nets and self organizing maps are discussed in this context as data based process models. Key–Words: geomagnetism, signal processing, wavelets, neuro fuzzy modelling, self organizing map