The effect of large roughness elements on sand transport efficiency was evaluated on a coastal sand sheet by measuring sand flux with two types of sand traps [Big Spring Number Eight (BSNE) and the Cox Sand Catcher (CSC)] at 30 positions through a 100 m-long × 50 m-wide roughness array comprised of 210 elements each with the dimensions 1·17 m long × 0·4 m high × 0·6 m wide. The 210 elements were used to create a roughness density (λ) of 0·022 (λ = n bh/S, where n is the number of elements, b the element breadth, h the element height, and S is the area of the surface that contains all the elements) in an area of 5000 m2. The mean normalized saltation flux (NSF) values (NSF = outgoing sand flux/incoming sand flux) at the furthest downwind distance for the two trap types were 0·44 and 0·41, respectively. This is in excellent agreement with an empirical model prediction of 0·5. The reduction in saltation flux is similar to an earlier separate study for an equivalent λ composed of elements of similar height (0·36 m), even though the roughness element forms were different (rectangular in this study as opposed to circular) as were the horizontal porosity of the arrays (49% versus 16%). This corroborates earlier results that roughness element height is a critical parameter that enhances reduction in sand transport by wind for similar λ configurations. The available data suggest the form of the relationship between transport reduction efficiency and height is likely a power relationship with two limiting conditions: (1) for elements ≤ 0·1 m high the effect is minimized, and (2) as element height matches and then exceeds the maximum height of the saltation layer (≥ 1 m), the effect will stabilize near a maximum of NSF ≈ 0·32. Copyright © 2012 John Wiley & Sons, Ltd.