When relatively large areas of the surface are analysed using surface-sensitive compositional techniques like spatially-resolved X-ray Photoelectron Spectroscopy (XPS), the large number of spectra makes traditional quantitative analysis a complex and time-consuming procedure. In turn, “mapping“ a heterogeneous surface using optical microscopy can be very helpful in rapid localization of different surface structures. Thus, developing a procedure to obtain reliable information on the composition of heterogeneous samples that combines the XPS signals with image data is an interesting option in surface analysis. This paper proposes the use of multivariate statistical analysis (MVA) to provide deeper understanding of the observed behaviour and to investigate the relationship between two independent data sets. As a feasibility study, the oxide formation at the polycrystalline rhodium surface, which is a complex system, has been analysed. Unlike existing local models, the pattern of the crystallographically different domains for their role in the oxidation process has been examined on the large scale. The MVA results of this study are consistent with previous research on this topic and may open a supporting approach towards heterogeneous samples analysis.