Dellwik et al. (2013) presented data from a forest edge experiment based on two meteorological towers instrumented with sonic anemometers. The experiment was performed at a dense edge of the Tromnaes Forest, which is a 24 m tall mature beech stand on the island Falster, Denmark. The topography at the site is flat. The towers were placed approximately 1.5 canopy heights upwind and downwind of the edge, respectively, and were two canopy heights tall. For near-neutral, near-perpendicular flow towards the edge, one finding concerned that although the wind speed gradients were similar before and after the edge, the momentum flux was strongly reduced above the canopy. This is contrary to the results by standard Reynolds' averaged Navier-Stokes models that predict an overshoot of the momentum flux. Further, a reduction of the vertical variance of the flow was largely compensated by an increase in the lateral variance, whereas the streamwise variance remained approximately constant. This result is in contrast to the predictions by homogeneous rapid distortion theory. We apply and develop an alternative framework based on inhomogeneous rapid distortion theory, also called blocking, in combination with the turbulence model by Mann (1994), and investigate whether this model can predict the observed changes of the flow. The presented results are relevant for understanding the rapid changes of turbulence in the heterogeneous landscape.