Detrend effect on the scalograms of GPS power scintillation

The study of amplitude scintillation on GPS radio links is usually done after detrending the time series of the transmitted power so to define scintillations as the chaotic fluctuation around a unitary value. In a sense, the choice of how to detrend the time series is part of the definition of scintillation. Here we analyse how far the continuous wavelet analysis of the detrended signal is influenced by the choice of detrending. This study is done using amplitude raw data from the GPS receivers held by INGV and the University of Bath in the Northern polar region, with a sampling time of 0.02 s. Three detrending procedures are considered: a fifth degree polynomial detrending, a high-pass filter with detrending period as twice the length of the time series considered, and a high-pass filter with detrending period determined via some statistical criterion. We show that there exists a "threshold time scale" of about half minute under which the differences between the scalograms from the signals detrended in the three ways are very small. This is not changed by applying the same detrending procedures to the segment of length reduced to one-third. Consequences in terms of scintillation definition and practical applications are given.