On the roburtness of GNSS-based tsunami early warning systems

Abstract: Recent devastating tsunamis (e.g., the 2004 Sumatra and 2011 Tohoku events) have repeatedly reminded us of the imperfection of existing tsunami early warning systems (TEWS) in such a harsh way and motivated researchers to make them more trustworthy. Realizing the fact that these tragedies were partly caused by the underestimation of earthquake magnitudes in the first few minutes after earthquakes because of the saturation of traditional seismic sensors, Global Navigation Satellites Systems (GNSS), which can measure co-seismic displacements directly, have found increasing applications in TEWS. Generally, there exist two major approaches that project GNSS displacements into tsunami scenario predictions. The conventional one adopts GNSS displacements for seismic source inversion (see, e.g., Ohta et al., 2012; Melgar and Yehuda 2013) while the other one interpolates seafloor deformation based on coastal GNSS station displacements (see, e.g., Song 2007; Xu and Song 2013; Liu et al., 2014). Despite both methods are being embedded into operational systems, no systematic comparisons between these two have been carried out to assess their robustness towards GNSS-based TEWS. To evaluate the reliability of the two methods in TEWS, we employed both approaches and applied them to three selected historical tsunamis caused by 2010 Mw7.8 Mentawai earthquake, the 2011 Mw9.1 Tohoku earthquake and the 2015 Mw8.2 Illapel earthquake. The three earthquakes represent typical tsunamigenic scenarios seen in practice: rare tsunami earthquake (2010 Mentawai case) without dense GNSS network, mega-thrust earthquake (2011 Tohoku case) with very dense GNSS network and medium-size earthquake with relatively dense GNSS network (2015 Illapel case). Through detailed analysis and comparison with tsunami observations, we address the pros and cons of the two methods and potential ways to combine them in GNSS-based tsunami early warning system. Key words: GNSS, tsunami early warning, seismic source inversion, sea-floor deformation interpolation