A LiDAR method of canopy structure retrieval for wind modeling of heterogeneous forests

The difficulty of obtaining accurate information about the canopy structure is a current limitation towards higher accuracy in numerical predictions of the wind field in forested terrain. The canopy structure in computational fluid dynamics is specified through the frontal area density and this information is required for each grid point in the three-dimensional computational domain. By using raw data from aerial LiDAR scans together with the Beer–Lambert law, we propose and test a method to calculate and grid highly variable and realistic frontal area density input. An extensive comparison with ground-based measurements of the vertically summed frontal area density (or plant area index) and tree height was used to optimize the method, both in terms of plant area index magnitude and spatial variability. The resolution of the scans was in general low (<2.5reflectionsm−2). A decrease of the resolution produced an increasing systematic underestimation of the spatially averaged tree height, whereas the mean plant area index remained insensitive. The gridded frontal area density and terrain elevation were used at the lower boundary of wind simulations in a 5km×5km area of a forested site. The results of the flow simulations were compared to wind measurements using a vertical array of sonic anemometers. A good correlation was found for the mean wind speed of two contrasting wind directions with different influences from the upstream forest. The results also predicted a high variability on the horizontal and vertical mean wind speed, in close correlation with the canopy structure. The method is a promising tool for several computational fluid dynamics applications requiring accurate predictions of the near-surface wind field.