In polymer electrolyte fuel cells of the types PEFC, DMFC and HT-PEFC, the gas diffusion layer (GDL) connects the electrodes with the feeding channels of the bipolar plate. The GDL is typically composed of materials based on carbon fibers, e.g., paper, woven or non-woven textiles. Efficient fuel cell operation requires that the electrodes are sufficiently supplied by gaseous fluids from the channels. Also, reaction products must be transported away from the electrodes. The GDL also has to provide electronic contact to the bipolar plates, but its major task is the mass transport of fluids. The gas transport in through-plane direction is simulated in the porous structure of the GDL, represented by stochastic geometries equivalent to the real structure. In order to support multi-scale simulation, effective properties can be calculated from these mesoscale simulation results to provide model parameters for continuum approaches in cell-scale simulations. In this paper, the resulting gas flow is analyzed with statistical methods with the focus on the interface between GDL and electrode. This approach provides the opportunity to detect quantitative relationships between functionality and microstructure and to design virtual GDL materials with improved transport properties. The evaluation of the interface with stochastic methods provides substantiated properties suitable for connecting regions representing fuel cell components of different spatial scales.