We introduce a new polarized radiative transfer model able to compute the polarization measured by a virtual instrument in a given nocturnal environment recreating real world conditions (1-dimensional atmospheric and aerosol profiles, 3-dimensional light sources with complex and widespread geometries, terrain obstructions). Initially developed to address the issue of aurorae and nightglow polarization, the model has potential applications in the context of light pollution, or aerosols and air pollution measurements in night time conditions. We provide the physical assumptions behind the model together with the main points regarding its numerical implementation, together with the inherent constraints and liberties it brings. The model, based on single scattering equations in the atmosphere, is first tested on a few simple configurations to assess the effect of several key parameters in controlled environments. The model outputs are then compared to field measurements obtained in four wavelengths at mid-latitude in a dark valley of the French Alps, 20 km away from the closest city. In this context where the nightglow emissions are supposedly stationary and widespread, a convincing fit between the model predictions and observations is found in three wavelengths. This confrontation of ground-based records with our modeling constitutes a proof of concept for the investigation of our polarized environment in nocturnal conditions, in the presence of localized and/or extended sources. It calls for further investigations. In particular we discuss the future need for inter-calibrating the sources and the polarimeter in order to optimally extract the information contained in such measurements, and how multiple-scattering (not implemented in the present study) could impact our observations and their interpretation.