This paper discusses scalar behaviour within the model canopy with emphasis on the interactions between flow conditions and scalar source distribution. In particular, vertical scalar fluxes within the canopy are highest for scalars emitted at ground level into the non-wake region of the flow. These scalars are emitted into a region of the flow characterised by diffusion and then highly intermittent but energetic sweeps, so that in non-wake regions of the flow, ground dominated scalar sources label sweeps. On the other hand, scalar concentrations are generally higher in the wake/transitional parts of the canopy, horizontal fluxes larger, but vertical fluxes are smaller than for non-wake regions. The scalar field is also highly intermittent in the wake compared with the non-wake region of the flow. This is because below 0.5 h in wake zones, scalars emitted from ground dominated sources diffuse into a region of highly intermittent, slow moving ejections that periodically transport the scalar vertically or horizontally. These scalars label ejections. In contrast, canopy scalar sources in the wake region of the flow are co-located with the momentum sink and the scalar near-field is displaced into highly turbulent flow. These scalars are rapidly mixed throughout the canopy. They label sweeps and vorticies shed by the dense canopy surface.
The data also allow direct calculation of dispersive fluxes demonstrating that, for this model canopy, dispersive momentum fluxes are of the same order of magnitude as turbulent momentum fluxes in the lower portion of the canopy and that dispersive flux divergence is nearly (minus) one half of the turbulence flux divergence in the upper canopy. These outcomes have important implications for the estimate of canopy drag coefficients with a projected 30 % underestimation in the lower canopy and up to 50 % overestimation in the upper canopy. Dispersive scalar fluxes are large and positive in the mid-canopy with higher values for ground compared with canopy scalar source distributions. Vertical scalar flux divergences are large compared with those for momentum, especially for the ground dominated scalar source distributions.
The data allow more complete determination of terms in the conservation equations, allowing us to investigate the role of scalar source distribution on scalar variance and flux budgets. The dispersive transport term in the scalar variance budget is small for canopy dominated sources but cannot be ignored in the lower canopy for ground dominated scalar source distributions. The gradient production term in this budget is largest for the canopy and ground only scalar source distributions, but the effect of adding a ground component is to broaden the peak so that it extends further into the lower canopy. In contrast, turbulent transport of scalar variance in the upper canopy is enhanced when the scalar source is predominantly located in the canopy region. Dispersive transport terms do not play a large part in the vertical turbulent scalar flux budget. The effect of the gradient production term is similar to that for the scalar variance budget. Introduction of even a small ground scalar source swings the wake production term from being is positive in the canopy to negative, albeit small. The turbulent transport term increases as scalar source partitions more towards the ground. There is no evidence of negative gradient production in the lower portion of the canopy. Counter gradient fluxes only occurred in a few profiles taken in the wake regions with the canopy only scalar source distribution. The mean scalar concentration profiles did not show counter-gradient behaviour within regions of the canopy or for the canopy as a whole.