Petrophysical and geophysical constrained inversion of gravity data based on starting and referenced models

Abstract To reduce the inherent ambiguity of gravity-data modeling in mineral resources exploration, it is important to incorporate available a priori information into the inversion process. In this paper, we study the use of starting and reference density models, based on two kinds of a priori information: (i) by interpolating the physical properties available from petrophysical or direct (well logs) knowledge; (ii) converting into densities by another geophysical property, such as resistivity or seismic velocity, previously estimated from geophysical modeling. We use both kinds of model as either a starting or reference model to optimally solve the regularized inverse problem using the preconditioned conjugate gradient method. We validate both the procedures using synthetic datasets produced by a high-density ore-layer model with drill-hole log constraints, and then apply them to the field data of an iron-ore deposit and a polymetallic ore region, in southern China. Both the procedures produce an improved and detailed density model; the use of a priori information as a reference model is, however, preferred, giving indeed a more compact and detailed solution, which seems more suitable to mining exploration.