Скачать или смотреть 3D model of a real forest from airborne LiDAR

This is a fly through 3D digital model of a real forest (Bois noir forest Barcelonnette, France) from LiDAR derived forest inventory parameters and aerial ortho-image. The model was developed as part of my research on 'Forest inventory parameters and Carbon mapping from airborne LiDAR'. High density, small footprint discrete airborne LiDAR is capable of capturing rich structural information of ground features. Trees were identified in the point cloud with high accuracy of 96%. It was quite challenging to develop algorithms to extract tree parameters like height, canopy projection area (CPA) , canopy volume, canopy diameter, canopy base height, canopy density, canopy tilt and orientation, local tree density and gaps from LiDAR point cloud. Carbon/biomass was obtained from regression model by trying out different combination of inventory parameters. CPA, height and local tree density were found to be significant estimator of Carbon when taken in combination. Canopy volume was equally significant when replaced with CPA in Carbon modelling.

The model gave me opportunity to hover around the forest and explore seen/unseen areas and take snapshots which otherwise were inaccessible. In the model, tree locations, species, tilt and orientation are based on LiDAR derived inventory data from my research. While the ground surface in the model has been prepared from aerial ortho-photo with elevation from LiDAR derived DTM. With the detailed LiDAR derived individual tree based geo-spatial inventory having more than two dozen bio-physical parameters, it is possible to model trees close to reality. In the model, it is possible to inquire any tree for its LiDAR derived parameters by simply clicking it. The model has in total 128, 000 trees. This includes trees as small as 2m height and 1m canopy diameter. The main species in the area were Pinus uncinata, Pinus sylvestris, Larix decidua and some broadleaved.

I found 3D visualization, a powerful decision support tool in forest management. As the model is driven by inventory data, it is possible to select any desired trees based on the query on inventory database for visualization. For example, one can study regeneration by selecting trees less than 4-6m height, may visualize forest after digital thinning, select trees for felling based on maturity criteria from database or look for forest disturbance by selecting tilted trees. A time series LiDAR data, if analysed may potentially give valuable information on species and area wise growth statistics. One can easily see the difference from 2D geo-spatial maps as 3D models are close to reality and easy on human perception. The future forest managers will hopefully have such models in their laptops to study forests in leisure.

Your suggestions and comments are eagerly invited.

Vinod Kumar
[email protected]