With the extensive adoption of transgenic crops, an understanding of transgene flow is essential to manage gene flow to non-GM crops. Thus, a flexible and accurate numerical model is required to assess gene flow through pollen dispersal. A three-dimensional atmospheric model combined with a diffusion transport model would be a useful tool for predicting pollen dispersal since it would be flexible enough to incorporate the effects of factors such as the spatial arrangement of crop combinations, land use, topography, windbreaks, and buildings. We applied such a model to field measurements of gene flow between two adjacent maize (Zea mays) cultivars, with suppression effects due to windbreaks, in an experimental cornfield in Japan. This combined model reproduced the measured cross-pollination distribution quite well in the case of maize plots with plant windbreaks slightly taller than the maize and without windbreaks, but the model underestimated the effect of a 6-m-tall windbreak net beyond 25 m from the donor pollen source on cross-pollination. The underestimation was most probably due to the problem of assimilated wind data. The model showed that the 6-m-tall windbreak and the plant wind break suppressed average cross-pollination rate by about 60% and 30%, respectively. Halftall and coarser mesh windbreak net suppressed cross-pollination rates by 40% by reducing the swirl of donor pollen by reduced wind speed.