Earthquake Magnitude Estimation using Precise Point Positioning

Abstract

An accurate estimation of an earthquake magnitude plays an important role in targeting emergency services towards affected areas. Along with the traditional methods using seismometers, site displacements caused by an earthquake can be monitored by the Global Navigation Satellite Systems (GNSS). GNSS can be used either in real-time for early warning systems or in offline mode for precise monitoring of ground motion. The Precise Point Positioning (PPP) offers an optimal method for such purposes, because data from only one receiver are considered and thus not affected by other potentially not stable stations. Precise external products and empirical models have to be applied, and the initial convergence can be reduced or eliminated by the backward smoothing strategy or integer ambiguity resolution. The product for the magnitude estimation is a peak ground displacement (PGD). PGDs observed at many GNSS stations can be utilized for a robust estimate of an earthquake magnitude. We tested the accuracy of estimated magnitude scaling when using displacement waveforms collected from six selected earthquakes between the years 2016 and 2020 with magnitudes in a range of 7.5–8.2 Moment magnitude MW. We processed GNSS 1Hz and 5Hz data from 182 stations by the PPP method implemented in the G-Nut/Geb software. The precise satellites orbits and clocks corrections were provided by the Center for Orbit Determination in Europe (CODE). PGDs derived on individual GNSS sites formed the basis for ground motion parameters estimation. We processed the GNSS observations by the combination of the Kalman filter (FLT) and the backward smoother (SMT), which significantly enhanced the kinematic solution. The estimated magnitudes of all the included earthquakes were compared to the reference values released by the U. S. Geological Survey (USGS). The moment magnitude based on SMT was improved by 20% compared to the FLT-only solution. An average difference from the comparison was 0.07 MW and 0.09 MW for SMT and FLT solutions, respectively. The corresponding standard deviations were 0.18 MW and 0.22 MW for SMT and FLT solutions, which shows a good consistency of our and the reference estimates.