Eulerian modelling of pollen dispersal over heterogeneous vegetation canopies
Résumé
With the introduction of genetically modified (GM) crops, predictive tools modelling wind velocity and pollen concentration fields as well as pollen deposition rate over heterogeneous canopies are required to assess the cross-pollination rates between GM and conventional crops. Over the last decade several numerical flow models have been developed to simulate dynamic mean and turbulent fields within and above the vegetation layer. In this paper, an Eulerian advection–diffusion conservation equation for pollen particles has been incorporated into one of these flow models, Aquilon. The relative velocity between air parcels and particles is simply represented through the addition of a particle settling velocity, i.e. the particle fall velocity in still fluid. The dynamic part of this model has been previously validated in two-dimensional heterogeneous cases (roughness change, forest edge flow) and tested in a more complex three-dimensional heterogeneous case (urban forested park). In order to test the ability of this Eulerian approach to simulate accurately airborne pollen concentration and pollen deposition rate within and above heterogeneous vegetation canopies, the model is validated against two field experiments where the airborne concentration and the deposition rate of maize pollen (Zea mays) were measured downwind from source plots.We also compare the outputs of Aquilon with those of a Lagrangian model previously tested against the same dataset. Generally speaking the model performs well, with a similar accuracy to the Lagrangian model. However, both models underestimate the measured maximum in ground pollen deposition just downwind from the maize plot. It is shown that this discrepancy may be due to an underestimation of the pollen settling velocity in this region. As the Stokes number, defined as the ratio between the maize pollen response time and the characteristic time of the turbulent structures at dissipation-range scale, is about 1 in the wake region behind the source plot, it is likely that turbulence leads to an increase in the apparent settling velocity there.