Gravity and gravity gradiometry measurements are commonly used to map density variations in the subsurface. Gravity measurements can characterize gravitational anomalies at long and short wavelengths effectively, but the cost of collecting a sufficiently spatially dense survey to characterize the short wavelengths can be prohibitive. Gravity gradient data can be quickly collected with short-wavelength information at a low noise level, but they have decreasing sensitivity to longer wavelengths. We have described a method to jointly invert gravity and gravity gradient data that takes advantage of the differing frequency contents and noise levels of the two methods to create an improved image of the subsurface. Previous work treated the inversion as a multiple component gravity inversion; however, this can cause unintended errors in the recovered models because each data set is not guaranteed to be fit within its noise level. Our joint inversion methodology ensures that the gravity and gravity gradient data sets are fit within their individual noise levels by incorporating a relative weighting parameter, and we describe how to find that parameter. This method can also be used to create an improved broadband gravity anomaly map that has a reduced noise level at long wavelengths using a joint equivalent source reconstruction. We first build a synthetic model using a Minecraft world editor that has different wavelength anomalies, and find the improvement with joint inversion. These results are also confirmed using a real-world example at the R. J. Smith test range in Kauring, Australia.
Read full abstract