Modelling of rough frictional interfaces is often based on asperity models, in which the individual behaviour of individual microjunctions is assumed. In the absence of local measurements at the microjunction scale, quantitative comparison of such models with experiments is usually based only on macroscopic quantities, like the total tangential load resisted by the interface. Recently however, a new experimental dataset was presented on the onset of sliding of rough elastomeric interfaces, which includes local measurements of the contact area of the individual microjunctions. Here, we use this more comprehensive dataset to test the possibility of quantitatively matching the measurements with a model of independent asperities, enriched with experimental information about the area of microjunctions and its evolution under shear. We show that, despite using parameter values and behaviour laws constrained and inspired by experiments, our model does not quantitatively match the macroscopic measurements. We discuss the possible origins of this failure.