Ground-based technological systems, such as power grids, can be affected by geomagnetically induced currents (GIC) during geomagnetic storms and magnetospheric substorms. This motivates the necessity to numerically simulate and, ultimately, forecast GIC. The prerequisite for the GIC modeling in the region of interest is the simulation of the ground geoelectric field (GEF) in the same region. The modeling of the GEF in its turn requires spatiotemporal specification of the source which generates the GEF, as well as an adequate regional model of the Earth?s electrical conductivity. In this paper, we compare results of the GEF (and ground magnetic field) simulations using three different source models. Two models represent the source as a laterally varying sheet current flowing above the Earth. The first model is constructed using the results of a physics-based 3-D magnetohydrodynamic (MHD) simulation of near-Earth space, the second one uses ground-based magnetometers? data and the Spherical Elementary Current Systems (SECS) method. The third model is based on a ?plane wave? approximation which assumes that the source is locally laterally uniform. Fennoscandia is chosen as a study region and the simulations are performed for the September 7?8, 2017 geomagnetic storm. We conclude that ground magnetic field perturbations are reproduced more accurately using the source constructed via the SECS method compared to the source obtained on the basis of MHD simulation outputs. We also show that the difference between the GEF modeled using laterally nonuniform source and plane wave approximation is substantial in Fennoscandia.