Modified Seasonal Decomposition Variations of Earth Magnetic Field Induction Module

In this paper, we present a modification of the classic method of seasonal decomposition of the time series, in particular its application for the analysis of geomagnetic data. Seasonal decomposition is a powerful tool for time series analysis, but its classic implementation does not always provide accurate results when the time series contains amplitude outliers and prolonged gaps. We propose a modified approach to solve this task of seasonal decomposition, by applying an average daily profile. This ensures the extraction of various anomalies in the residual component of the decomposition, in particular, global and contextual outliers, as well as disturbances due to magnetic storms in the variations of geomagnetic field induction module. Keywords: geomagnetic field, seasonal decomposition, data gaps, autocorrelation function, residual component, outliers, magnetic storm, DST index