In-Canopy Differences in Source Distributions Affect the Degree of Segregation between Scalars — a Large Eddy Simulation Study

From the perspective of the atmosphere, vegetated surfaces can often be considered as a single surface as long as one is only interested in the resulting surface fluxes of passive scalars. However, in order to understand the behavior of scalars (both passive and reactive) in or just above the canopy, knowledge of the combined effect of canopy turbulence and scalar source distributions is necessary. This is of particular importance in the case of reactive species in which the intensity of segregation (directly related to correlation between both scalars) affects the mean reaction rate.

To enable controlled experiments, we have simulated the turbulence under neutral atmospheric conditions, in and above a canopy with height h_c using the Dutch Atmospheric Large Eddy Simulation (DALES) model (domain size length×width×height: 72h_c×36h_c×32h_c, grid size 0.1h_c). This has been done for two plant area densities: a uniform distribution and a forest-like distribution (both with an LAI of 2). The two distributions give rise to different turbulence in and above the canopy: e.g. vorticity thickness are 1.56h_c and 1.73h_c, respectively. In each simulation 11 separate scalars were emitted: one at the soil surface and ten at different levels in the canopy. Using these 11 scalars, scalar fields with a wide variety of source distributions could be constructed.

The analysis focuses on three aspects:

• The scalar displacement height is both determined from the mean scalar profiles above the roughness sublayer and as the mean height of the source distribution. We show that the displacement height determined from the mean profiles does vary with the vertical source distribution, but less than the variations in mean source height suggest. The discrepancy between both estimates is largest for the forest-like plant area density.
• Next we consider the correlation between two scalars, one with a source at the soil surface and the other with a source at the top of vegetation. For this scalar combination the correlation is minimum half-way the canopy height (h_c), with a local maximum close to the ground. Throughout the canopy the correlation is positive, despite the fact that in the lower half of the canopy the sign of the vertical gradients of both scalars is opposite. Strong indications of counter-gradient transport are found in the plots showing the covariance of both scalars. The scalar-scalar correlation first exceeds 0.9 well above the canopy (z/h_c > 2.5). This level of near-perfect correlation is only marginally sensitive to the height of the canopy source as long as the latter is located at z/h_c > 0.6. For sources below mid-canopy level the correlation between the soil scalar and canopy scalar reaches 0.9 in or just above the canopy.
• Finally, the correlation between two scalars that have a composite source distribution (both soil and canopy) is considered. We constructed scalar pairs with all positive canopy sources (concentrated in the upper 1/3 of the canopy), but with variations in the sign and strength of the soil sources (+/+, +/− and −/−). In particular inside the canopy the canopy source dominates the correlation signal: the soil source (or sink) is only able to affect the correlation significantly when it is about twice as strong as the canopy source (or stronger). This asymmetry is stronger for the forest-like canopy than for the uniform plant area distribution.

Summarizing:

• Scalar-scalar decorrelation is affected both by the vertical source distribution of both scalars, and by the architecture of the canopy. With our setup we were able to study both effects separately.
• Decorrelation between scalars, as far as it is caused by in-canopy source differences, is restricted to the roughness sublayer.