An assessment of the impact of Next Generation Gravity Missions on earthquake signal retrieval. Constructing a database of time-varying co-seismic and post-seismic gravity change and a detectability assessment strategy

Advancements in mission concepts and instrumentation, such as those investigated under the Mass change And Geosciences International Constellation (MAGIC), could enable significant enhancements in the spatial and temporal resolution of gravity field products. Closed-loop simulations of these new constellations of instruments estimate radical improvements in the error budget in the retrieval of geophysical signals, including those arising from mass movement in the solid Earth – earthquakes included. The displacements caused by co-seismic dislocation and post-seismic relaxation are sensed by a broad array of seismological and geodetic techniques. Gravity has the potential of providing additional information, especially when the mass movement is a-seismic and when its surface expression occurs mostly in areas that are difficult or impossible to sense with other remote-sensing techniques (GNSS, dInSAR), such as off shore. In order to assess by how much the improvements in MAGIC would lower the detectability threshold, we modelled a database of synthetic earthquake gravity signal, including the effect of post-seismic viscoelastic relaxation. We computed the gravity change in time using the QSSPSTATIC [1] code, set up in a way to obtain the spherical harmonics (SH) coefficients of the geopotential change through time. This data, which we then used in a detectability assessment, also allow comparing different modelling strategies and signal retrieval methods. We devised its data structure, by design, to be easily included as part of time-varying signals used in simulations, enhancing the solid-Earth component of models such as AOHIS [2]. We test detectability in terms of the SH-domain SNR between the earthquake signal and the gravity model errors. The SH coefficients of both quantities undergo a spatio-spectral localization procedure [3] and are compared in terms of their localized degree variances. We show how a spatial-scale dependent analysis, such as the one that a spectral-domain method allows, is needed to fully exploit the signal in the optimal range of spatial wavelengths owing to the coloured spectra of signal and noise. We perform a parametric study of the effect on detectability of moment magnitude, source parameters (focal mechanism, depth), and rheological profiles – with magnitude being the first-order predictor of detectability in co- and post-seismic signals. As a methodological test, we also present an experiment on the signal omission arising from approximating an earthquake dislocation as a point-source, comparing its signal to the one we can obtain using a finite fault solution of a real event instead. We assess and discuss the impact of a simpler model on the trade-space between the precision of a detectability assessment and the added computational effort. References [1] Wang et al., 2017 DOI:10.1093/gji/ggx259 [3] Dobslaw et al., 2015 DOI:10.1007/s00190-014-0787-8 [2] Wieczorek and Simons, 2005 DOI:10.1111/j.1365-246X.2005.02687.x