Predicting geophysical measurements: Testing a combined empirical and model-based approach using surface waves

There are several approaches commonly used to provide spatial corrections of geophysical data. One end-member approach is to develop model-based corrections in which the variations are based either on a priori knowledge or on geophysical inversions, such as tomography. While this approach would have the advantage of full spatial coverage, in general one would be unable to recover the full station corrections due to the inherent averaging of models. Another end-member approach, common in seismically active regions, would be to develop empirically based corrections based directly on measurements. An example of this approach would be kriging or other interpolation techniques. The advantage of this approach is a true fit to the most applicable measurements, but at the cost of severe spatial limitations. Ideally, one would like to combine the two approaches. This short note demonstrates the value of a combined empirical and model-based approach to generating accurate spatial correction surfaces using surface wave measurements and a tomography model from the Middle East and North Africa. By kriging residual values from a tomographic model, there was a significant reduction in misfit over either approach alone. In principle, kriging measurement residuals from models could improve many geophysical applications from travel times to amplitudes.