Resolution and uncertainty in lithospheric 3-D geological models

As three-dimensional (3-D) modelling of the subcontinental mantle lithosphere is increasingly performed with ever more data and better methods, the robustness of such models is increasingly questioned. Resolution thresholds and uncertainty within deep multidisciplinary 3-D models based on geophysical observations exist at a minimum of three levels. Seismic waves and potential field measurements have inherent limitations in resolution related to their dominant wavelengths. Formal uncertainties can be assigned to grid-search type forward or inverse models of observable parameter sets. Both of these uncertainties are typically minor when compared to resolution limitations related to the density and shape of a specific observation array used in seismological or potential field surveys. Seismic wave source distribution additionally applies in seismology. A fourth, more complex level of uncertainty relates to joint inversions of multiple data sets. Using independent seismic wave phases or combining diverse methods provides another measure of uncertainty of particular physical properties. Extremely sparse xenolith suites provide the only direct correlation of rock type with observed or modelled physical properties at depths greater than a few kilometers. Here we present one case study of the Canadian Mohorovičić (Moho) discontinuity using only two data sets. Refracted and converted seismic waves form the primary determinations of the Moho depth, gravity field modeling provide a secondary constraint on lateral variations, the slope of the Moho, between the sparse seismic estimates. Although statistically marginal, the resulting co-kriged Moho surface correlates better with surface geology and is thus deemed superior.