The decontamination of materials is a challenging problem in a wide range of industrial, medical, civilian, and disaster-response applications. Attempts to decontaminate porous surfaces (e.g., concrete, tarmac or wood) can lead to a partial redistribution of the contaminant within the porous medium during the decontamination process rather than a complete removal. This contaminant redistribution while attempting to clean the material is of crucial importance in cases of extremely harmful contaminants. In this work, we present experiments modelling the surface-washing of pre-wet porous substrates. Model porous surfaces, produced by sintering packed glass beads onto solid glass, are integrated into a surface-washing apparatus. The contaminant agent is simulated by a passive tracer (disodium-fluorescein aqueous solutions) introduced in the form of droplets, while the surface-washing is simulated by a thin gravity-driven water film flowing over an inclined porous plane. The contaminant mass transfer due to the cleansing flow is tracked by direct image analysis based on dye attenuation (using three different illumination wavelengths), enabling study of the space-time evolution of the contaminant field within the porous medium. Additionally, an inline UV-Vis spectrometer is used to monitor in real-time the contaminant concentration in the washing effluent. Our experiments provide insights on the fundamental physics governing the cleaning process, such as the role of initial conditions (e.g., ingress of contaminant) and the impact of process parameters on the decontamination efficiency. Importantly, they demonstrate a decontamination-induced redistribution of the contaminant within the porous matrix that, in particular for short and incomplete washing processes, could increase the risk posed by the agent.