Wavelet Filtering of GNSS Network Data for the Detection and Identification of Ionospheric Disturbances Caused by Tsunamis

Disturbances Caused by Tsunamis Yu-Ming Yang, James L. Garrison, See-Chen Lee Purdue University 1 Problem Statement Acoustic-Gravity Waves (AGWs) in the neutral atmosphere can induce disturbances in the ionosphere that are subsequently observable in trans-ionospheric Global Navigation Satellite System (GNSS) measurements of Total Electron Content (TEC). Disruptive events on the Earth’s surface, such as earthquakes, tsunamis [1] and large explosions are known to be one source of these disturbances. Through observing the disturbances induced by a tsunami, a better capability to detect and provide warning of the imminent occurrence of the tsunami wave may be possible. Coherent structure in the ionosphere, indicative of a propagating disturbance, can be detected by crosscorrelating pairs of filtered TEC time series [2], [3]. Cross-correlating every pair of stations in a large (100’s of stations) network, or sub-areas of a very large (1000’s of stations) network [4] produces a greatly overdetermined system that can be inverted to estimate the horizontal components of the disturbance velocity. Cross-correlation has a severe limitation, however, in that it assumes that only a single wave train is observable within the correlation time window. Prior studies have shown [5] that the occurrence rate of traveling ionospheric disturbances (TID’s) under quiescent conditions can be quite high, with a strong seasonal dependence. Furthermore, a single event on the surface of the Earth could give rise to multiple disturbances with different arrival times [6]. It is thus quite possible that multiple disturbances could be present within the same time and space window. This could reduce the correlation magnitude, causing a disturbance to be missed, obscure the structure of the disturbance (such as multiple wave trains from different paths), or lead to incorrect estimates of the speed and direction [7]. In this paper, we investigate the use of wavelet analysis to detect these disturbances under more general conditions, and design a filter to isolate each identified wave train. Filtered time series could then be processed with pair-wise cross-correlation as before, to estimate the parameters of a wave model.