8. Summary and Conclusion
Ionospheric perturbations induced by tsunamis and earthquakes are obtained from GPS-TEC sampled at uniform time intervals along the satellite tracks. However, such samplings will be non-uniform in space. Not accounting such non-uniform spatial sampling while computing the ionospheric perturbations introduces signal aliasing, predominantly in the amplitude. All the methods hitherto used to detect the co-seismic and tsunamigenic ionospheric perturbations did not account the non-uniform spatial sampling while computing the perturbations. Further, the residual method introduces artifacts as selection of the order of polynomials in this method is subjective. Recently, Shimna and Vijayan (2020) proposed an algorithm called Spatio-Periodic Leveling Algorithm (SPLA) to remove such aliasing from ionospheric irregularities induced by geomagnetic storms.
In this study, we showed that adopting SPLA to compute tsunami and earthquake induced ionospheric perturbations are efficient in removing aliasing and artifacts. Further, we showed by carrying out efficiency tests under two simulated scenarios and using GPS observations carried out during the 26th December 2004 Indian Ocean tsunami, and 25th April 2015 Nepal-Gorkha earthquake that SPLA can resolve the perturbations from sharp static variations. Observational validation (Fig. 13) show that the perturbations obtained using SPLA are within the expected values, whereas, dTEC (differential method) and rTEC (residual method) show clear deviation. The maximum deviation (δr ) of rTEC and dTEC in the observational data set are 1.08 and 0.69 (Fig. 13). The uncorrected inter-IPP distances cause the magnitude of aliasing up to 2 times (Fig. 13). This emphasizes the importance of correcting the influence of inter-IPP distance while computing the ionospheric perturbations.
Uncorrected aliasing and artifacts severely impact the characteristics of the ionospheric perturbations. An assessment of the impact of aliasing and artifacts showed that SNR of the aliasing and artifact free ionospheric perturbations computed using SPLA is ~39% and ~149% higher compared to the perturbations obtained using differential and residual methods (Fig. 18). Wavelet and cross-correlation analyses carried out on TIPs and CIPs reveal that the time of occurrence and frequency of the perturbations differ significantly between SPLA and residual method (Fig. 14 and Fig. 16). Besides, the residual method fails to detect 25% of TIPs and 50% of CIPs which were detected by both differential method and SPLA (Tables 1 and Table 2). Further explorations showed that misfits of uniform high order polynomial representing the trend of GPS-TEC caused the failure in the detection of the perturbations in the case of residual method (Fig. 12 and 17). Above all, the results shown in section 6.4 reveals that SPLA is a very good candidate to obtain ionospheric perturbations at low elevation angles and employing SPLA will increase the area of ionospheric exploration by a GPS receiver.
Overall, this study shows that residual method performs poorly compare to other methods in resolving sharp static variations from signals and misses to detect ionospheric perturbations. Hence, caution needs to be exercised while adopting residual methods in real-time detection for earthquake or tsunami early warning, particularly, the one like VARION – Variometric Approach for Real-Time Ionosphere Observation (Savastano et al., 2017) which uses both differential and residual approach to obtain TIPs in real-time.
Despite the advantages, the perturbations obtained using SPLA bound to vary with the selection of ionospheric shell height (hmax). Hence, a careful selection of appropriate ionospheric shell height specific to the region and time of the ionospheric monitoring is essential while adopting SPLA to obtain ionospheric perturbations using GPS or any other Global Navigational Satellite Systems.