Using observed data we can calculate parameter within the Earth crust correlated for each geophysical field. Such as we can calculate magnetization distribution for magnetic field. We have developed an original iterative method of solving magnetic structure inverse problem (on the basis of local correction method). We suppose that magnetic field value at observed point due to nearest part of surface S which is boundary between two layers with different magnetization. This idea let us to reduce nonlinear inverse problem to algebraic equation which can be solved easy. In this paper we describe new results of construction of the Earth crust magnetic models (Polar Ural Mountains).