[S21-P] Poster

[S21-P-13] Estimation of global crustal model uncertainty using geostatistical analysis

Wolfgang Szwillus¹, Walter D. Mooney², ^○Jorg Ebbing¹ (1.Kiel University, Kiel, Germany, 2.USGS, Menlo Park, USA)

Active seismology provides the best tool to estimate the thickness and seismic velocity of the crust. However, seismic experiments are mostly constrained to profiles and hence do not provide spatially complete coverage. Passive seismology can complement active seismology, but is often adversely affected by sparse station coverage, especially on a global scale. As a result, models of crustal properties are based on a form of interpolation, which leads to smoothing and might not reproduce the true crustal structure.
This has for example consequences if such models are used to estimate upper mantle densities by stripping of a crustal density structure from the gravity field. The residual gravity field reflects both possible density anomalies in the mantle and the errors in the crustal density model.
Here, we make use of a global database of active seismological data acquired over the last 60 years and explore how interpolation is affecting the attainable accuracy of crustal models and compare our results with previously published global crustal models. Our interpolation method is based on Kriging and provides uncertainties in addition to the gridded values. This uncertainty translates differently to near surface gravity and satellite gravity gradients, which allows to evaluate the uncertainties compared to uncertainties by velocity-density conversion and to define strategies for global inversion.