Strong heterogeneities in the composition of the volatile species have been detected in the coma of comet 67P/Churyumov-Gerasimenko (67P/C-G) by the ROSINA instrument onboard the ESA’s Rosetta spacecraft. However, it is not clear if these heterogeneities are indicative of heterogeneities in the near-surface nucleus composition or if the coma composition is mainly insolation-driven. In order to clarify the link between the composition of the nucleus and the composition of the coma we have performed numerical simulations and compare our results with measurements acquired by ROSINA/DFMS for three major volatile species namely, H2O, CO2, and CO. We use a previously published thermo-physical numerical model designed to study cometary nucleus evolution, including volatile outgassing and internal stratigraphy, as the comet orbits the Sun. The model follows schemes used for much of the past three decades to model cometary outgassing. Our results match well the experimental volatiles density measurements of ROSINA/DFMS for most of the Rosetta mission. They suggest that the outgassing pattern is mainly insolation-driven and the variations are caused by the tilt of rotation axis and eccentricity of the nucleus. The nucleus shows to 1st order a homogeneous composition and therefore we can provide constraints on the bulk volatiles composition of 67P/C-G nucleus which is dominated by H2O (91.4% ± 4.5%), then CO2 (6.7% ± 3.5%) and CO in small amount (1.9% ± 1.2%). However, in details, a dichotomy in composition between the northern and southern hemispheres of the comet is revealed. A CO/CO2 bulk composition ratio of about 0.6 ± 0.1 is required to reproduce the measurements from the northern hemisphere and about 0.2 ± 0.1 for the southern hemisphere. To match the data, the thermal properties of the nucleus surface must be modified by adding a thin desiccated dust mantle (~5 mm) for northern latitudes while this appears not to be necessary for southern latitudes. This may be related to the observed dichotomy in putative airfall deposits. We suspect that, because of thermal inertia, seasonally non-illuminated areas continue to outgas and influence the ROSINA measurements. This effect cannot be reproduced with the model. Therefore during some periods of the mission, the fits are not ideal. Finally, the outgassing of the different ices leads to a layered internal structure defined by the sublimation front of each ice and formation of harder layers close to the surface due to sublimation/condensation processes. However, the thermo-physical model overestimates the absolute volatiles production (mainly in the southern hemisphere) leading to an overestimation of the erosion rates. Further investigations will be performed to improve the thermo-physical model and the sensitivity analysis.