We study the inversion of potential fields and evaluate the degree of depth resolution achievable for a given problem. To this end, we introduce a powerful new tool: the depth-resolution plot (DRP). The DRP allows a theoretical study of how much the depth resolution in a potential-field inversion is influenced by the way the problem is discretized and regularized. The DRP also allows a careful study of the influence of various kinds of ambiguities, such as those from data errors or of a purely algebraic nature. The achievable depth resolution is related to the given discretization, regularization, and data noise level. We compute DRP by means of singular-value decomposition (SVD) or its generalization (GSVD), depending on the particular regularization method chosen. To illustrate the use of the DRP, we assume a source volume of specified depth and horizontal extent in which the solution is piecewise constant within a 3D grid of blocks. We consider various linear regularization terms in a Tikhonov (damped least-squares) formulation, some based on using higher-order derivatives in the objective function. DRPs are illustrated for both synthetic and real data. Our analysis shows that if the algebraic ambiguity is not too large and a suitable smoothing norm is used, some depth resolution can be obtained without resorting to any subjective choice of depth weighting.

We present a potential-field-constrained inversion procedure based on a priori information derived exclusively from the analysis of the gravity and magnetic data (self-constrained inversion). The procedure is designed to be applied to underdetermined problems and involves scenarios where the source distribution can be assumed to be of simple character. To set up effective constraints, we first estimate through the analysis of the gravity or magnetic field some or all of the following source parameters: the source depth-to-the-top, the structural index, the horizontal position of the source body edges and their dip. The second step is incorporating the information related to these constraints in the objective function as depth and spatial weighting functions. We show, through 2-D and 3-D synthetic and real data examples, that potential field-based constraints, for example, structural index, source boundaries and others, are usually enough to obtain substantial improvement in the density and magnetization models.

We analyze a wide gravity low in the Campania Active Volcanic Area and interpret it by a large and deep source distribution of partially molten, low-density material from about 8 to 30 km depth. Given the complex spatial-temporal distribution of explosive volcanism in the area, we model the gravity data consistently with several volcanological and petrological constraints. We propose two possible models: one accounts for the coexistence, within the lower/intermediate crust, of large amounts of melts and cumulates besides country rocks. It implies a layered distribution of densities and, thus, a variation with depth of percentages of silicate liquids, cumulates and country rocks. The other reflects a fractal density distribution, based on the scaling exponent estimated from the gravity data. According to this model, the gravity low would be related to a distribution of melt pockets within solid rocks. Both density distributions account for the available volcanological and seismic constraints and can be considered as end-members of possible models compatible with gravity data. Such results agree with the general views about the roots of large areas of ignimbritic volcanism worldwide. Given the prolonged history of magmatism in the Campania area since Pliocene times, we interpret the detected low-density body as a developing batholith.

The gravity gradient tensor (GGT) has been increasingly used in practical applications, but the advantages and the disadvantages of the analysis of GGT components versus the analysis of the vertical component of the gravity field are still debated. We analyse the performance of joint inversion of GGT components versus separate inversion of the gravity field alone, or of one tensor component. We perform our analysis by inspection of the Picard Plot, a Singular Value Decomposition tool, and analyse both synthetic data and gradiometer measurements carried out at the Vredefort structure, South Africa. We show that the main factors controlling the reliability of the inversion are algebraic ambiguity (the difference between the number of unknowns and the number of available data points) and signal-to-noise ratio. Provided that algebraic ambiguity is kept low and the noise level is small enough so that a sufficient number of SVD components can be included in the regularized solution, we find that: (i) the choice of tensor components involved in the inversion is not crucial to the overall reliability of the reconstructions; (ii) GGT inversion can yield the same resolution as inversion with a denser distribution of gravity data points, but with the advantage of using fewer measurement stations.

We describe a multihomogeneity theory for source-parameter estimation of potential fields. Similar to what happens for random source models, where the monofractal scaling-law has been generalized into a multifractal law, we propose to generalize the homogeneity law into a multihomogeneity law. This allows a theoretically correct approach to study real-world potential fields, which are inhomogeneous and so do not show scale invariance, except in the asymptotic regions (very near to or very far from their sources). Since the scaling properties of inhomogeneous fields change with the scale of observation, we show that they may be better studied at a set of scales than at a single scale and that a multihomogeneous model is needed to explain its complex scaling behaviour. In order to perform this task, we first introduce fractional-degree homogeneous fields, to show that: (i) homogeneous potential fields may have fractional or integer degree; (ii) the source-distributions for a fractional-degree are not confined in a bounded region, similarly to some integer-degree models, such as the infinite line mass and (iii) differently from the integer-degree case, the fractional-degree source distributions are no longer uniform density functions. Using this enlarged set of homogeneous fields, real-world anomaly fields are studied at different scales, by a simple search, at any local window W, for the best homogeneous field of either integer or fractional-degree, this yielding a multiscale set of local homogeneity-degrees and depth estimations which we call multihomogeneous model. It is so defined a new technique of source parameter estimation (Multi-HOmogeneity Depth Estimation, MHODE), permitting retrieval of the source parameters of complex sources. We test the method with inhomogeneous fields of finite sources, such as faults or cylinders, and show its effectiveness also in a real-case example. These applications show the usefulness of the new concepts, multihomogeneity and fractional homogeneity-degree, to obtain valid estimates of the source parameters in a consistent theoretical framework, so overcoming the limitations imposed by global-homogeneity to widespread methods, such as Euler deconvolution.