Gravity-driven flows in porous media occur in a wide range of industrial and environmental contexts including hydrology, carbon dioxide storage, geothermal power generation, and contaminant leaks (Bear, 1971;Guo et al., 2016;MacFarlane et al., 1983). These sub-surface flows are difficult to monitor directly and therefore it is common to drill wells to sample the flow, for example in CO 2 storage projects to monitor breakthrough times (Hannis et al., 2015;Mathieson et al., 2011). Wells are also used for extracting fresh water from aquifers and for monitoring salt water intrusion (Barlow & Reichard, 2010;Dagan & Zeitoun, 1998;Masterson et al., 1998). Accurately determining the flow velocity from a single borehole presents difficulties, however, because its presence may significantly alter the flow field even in the case that no fluid is extracted because it alters the structure and permeability of the constituent porous media (Sekhar & Sano, 2001). It is vital to understand this effect in order to interpret the data acquired correctly.It is well-established that pressure-driven flows are focused by inclusions that are of higher permeability than the surroundings whilst lower permeability inclusions divert the flow, and the volume flux of fluid that passes through the inclusion has been calculated in each case (Hinch & Bhatt, 1990;Phillips, 1991). However, in many contexts, the flow forms a free-surface and is primarily driven by buoyancy. We develop a physical model to explore the interaction of free-surface flows driven by gravity (rather than by imposed pressure gradients) on an inclined plane with a cylindrical inclusion of different permeability (Figure 1). We investigate how the flow thickness and velocity field depend on the permeability ratio and the width of the cylinder relative to the oncoming flow thickness and slope inclination. In the case of a relatively narrow cylinder, we show that the behavior is analogous to the interaction of two-dimensional pressure-driven flows with inclusions. The interaction with relatively wide cylinders is substantially different, however. For example, we will show that upstream of a relatively wide impermeable inclusion, deep ponds of fluid will form. Similar behavior occurs on the downstream side of the interior of a high permeability inclusion. We are particularly interested in the flux of fluid into the cylinder from upstream and determining controls on whether the particle paths are predominantly diverted around, focused into or not significantly influenced