Natural self-potential (SP) fields observed in the vicinity of conductive orebodies can be explained in terms of electrochemical reactions in which the conductors participate. Battery-like models assume that a buried conductor creates an anode-cathode pair by conveying a flow of electrons to oxidizing areas in the shallow subsurface from reducing areas at depth. For conductors with invariant composition (behaving as an inert electrode), a quantitative formulation is already available. Numerical Green's functions are used to allow one-step evaluation of SP fields from an inert electrode model. The model is used to simulate geoelectric targets in mineral exploration and to test a procedure to obtain current source terms by inverting an SP data set. Data inversion is constrained by charge conservation and prescribes source terms at the target surface. A background resistivity model is assumed to be known and is used to recognize interfaces and evaluate numerical Green's functions in forward and inverse modeling. The inversion procedure is applied to interpret 2D data from two gold deposits of the Yanacocha district, Peru.