Comparing source inversion techniques for GPS‐based local tsunami forecasting: A case study for the April 2014 M8.1 Iquique, Chile, earthquake

Abstract

Real‐time GPS is nowadays considered as a valuable component of next generation near‐field tsunami early warning systems able to provide fast and reliable source parameters. Looking for optimal methodologies and assessing corresponding uncertainties becomes an important task. We take the opportunity and consider the 2014 Pisagua event as a case study to explore tsunami forecast uncertainty related to the GPS‐based source inversion. We intentionally neglect all other sources of uncertainty (observation set, signal processing, wave simulation, etc.) and exclusively assess the effect of inversion technique. In particular, we compare three end‐member methods: (1) point‐source fastCMT (centroid moment tensor), (2) distributed slip along predefined plate interface, and (3) unconstrained inversion into a single uniform slip finite fault. The three methods provide significantly different far‐field tsunami forecast but show surprisingly similar tsunami predictions in the near field.

What Can We Do to Forecast Tsunami Hazards in the Near Field Given Large Epistemic Uncertainty in Rapid Seismic Source Inversions?

“…The coefficient of variation yields 1.2, 1.0, and 0.9, respectively (see Figure ), indicating that variability range is as large as the median value. It is important to notice that, despite this large variability, most of the values concentrate well above the 1‐m threshold used to trigger evacuations; hence, it would not affect the operational outcome of TEWS (Chen et al, ; Melgar, Fan, et al, ) for this large event.…”

“…However, most of the assessments have been carried out using tsunami amplitudes in coastal waters or runup at the coast (e.g., Babeyko et al, ; Chen et al, ; Hoechner et al, ; Melgar & Bock, ; Melgar, Fan et al, ; Ohta et al, ; Riquelme et al, ; Tang et al, ; Tsushima et al, ). Melgar, Fan, et al () and Chen et al () compared the capabilities within the context of TEWS, by estimating hazard and categorize it in accordance to tsunami hazard thresholds as usually done by operational TEWS, with good results. However, the use of thresholds allows for significantly different solutions to yield similar hazard assessments (Chen et al, ) and does not take full advantage of the wealth of data that finite fault solutions could provide.…”

“…Melgar, Fan, et al () and Chen et al () compared the capabilities within the context of TEWS, by estimating hazard and categorize it in accordance to tsunami hazard thresholds as usually done by operational TEWS, with good results. However, the use of thresholds allows for significantly different solutions to yield similar hazard assessments (Chen et al, ) and does not take full advantage of the wealth of data that finite fault solutions could provide. It would be appealing to use that information to obtain NRTTHA in terms of tsunami inundation in coastal areas, which is more demanding computationally and may require a higher level of accuracy.…”

Toward Near‐Field Tsunami Forecasting Along the Cascadia Subduction Zone Using Rapid GNSS Source Models

“…Ohta et al (2012) also tested the utility of real-time GPS in detecting coseismic displacements, linking them to tsunami forecasts. Chen et al (2016) tested different rapid GPS-based source inversions for their utility in tsunami early warning for the 2014 Iquique earthquake, finding similarities in local tsunami amplitudes. The use of GNSS positions of cargo ships has been proposed as a useful data set for measuring tsunamis (Inazu et al, 2016).…”

Automated GNSS and Teleseismic Earthquake Inversion (AutoQuake Inversion) for Tsunami Early Warning: Retrospective and Real-Time Results

“…Furthermore, Song (2007) demonstrated that coastal GNSS stations can be used to infer seafloor displacements and estimate earthquake-induced oceanic energy and tsunami scales within a few minutes after the quake for early warning. The GNSS approaches have been further demonstrated and improved for more earthquakes (e.g., Ohta et al 2012;Song et al 2012;Yang et al 2014;Chen et al 2016a;Fu et al 2017).…”