We present a new technique for reducing the uncertainties inherent in the interpretation of lithospheric magnetic field observations over the Earth. This technique, without involving iterations, provides an improved estimate of the depth-integrated magnetic susceptibility of the crust as compared to previous approaches. Departing from the normal practice of using observations at specific locations, we model directly Gauss coefficients of the lithospheric magnetic field and use an a-priori initial lithospheric thickness model to circumvent magnetic annihilators (i.e. magnetisation distributions that cannot be determined from magnetic data since they have no impact on the lithospheric magnetic field). Of the several initial magnetic layer models tested, we prefer the model where magnetic thickness is based on the Moho or the regional estimate of the Curie depth when it is shallower than the Moho because it is physically reasonable and produces the fewest artefacts. The method is applied to a recent high-degree lithospheric magnetic field model called LCS-1 derived from CHAMP and Swarm magnetic satellite data. The technique is appropriate for regions where induced magnetisation dominates over remanent magnetisation. We show that high degrees of the final depth-integrated magnetic susceptibility variation are dependent only on the corresponding high degrees of the LCS-1 magnetic field model. Thus, the depth-integrated magnetic susceptibility variation is an important quantity derived in the study which enables readily qualitative interpretation of regional geology.