Travelling ionospheric disturbances detection: A statistical study of detrending techniques, induced period error and near real-time observables

Abstract

Due to advances in remote sensing of the Earth's Ionosphere through Total Electron Content (TEC) estimates by Global Navigation Satellite System (GNSS) receivers, it is possible to detect and characterize Travelling Ionospheric Disturbances (TIDs) in both post-processing and, to some extent, in near real-time (NRT). A reliable and precise TEC filtering technique must be adopted to characterize waves accurately. Specifically, TEC detrending is widely adopted to extract the amplitude and period of the detected ionospheric waves from the background ionospheric conditions. Therefore, this study aims to understand and compare how different TEC detrending techniques and their settings impact the ability to extract such parameters. We highlight that the novel Fast Iterative Filtering (FIF) and the Savitzky-Golay filter (SGOLAY) techniques are the most reliable overall compared with moving average (MA), multi-order numerical difference (DD), polynomial detrending (POLY) and Finite Impulse response (FIR) band-pass filter (BUTF). Moreover, the impact of general algorithm settings on the extracted TID period is investigated, such as the Ionospheric Piercing Point (IPP) height and elevation cut-off angle, showing that such parameters drastically impact the retrieved period, especially for slower TIDs. Finally, due to the growing interest in real-time (RT) detection and classification of TIDs, the study proposes techniques for accurately estimating the TID amplitude in an NRT scenario. Such NRT techniques are then compared with the widely used post-processing products, such as the calibrated vertical TEC (vTEC), showing a difference that is mostly lower than the typical noise level of GNSS receivers (0.05 TECu).