The point cloud is initially produced using the latest boresight values for the sensor. Preliminary quality assurance steps are taken to ensure data integrity. A series of off the shelf software and proprietary tools are utilized throughout the LiDAR data processing procedures.
The calibrated point cloud strips are then processed into 1km tiles in the New Brunswick Double Strereographic projection. Next, a combination of automatic and manual classification is done separating points into ground (class 2), non ground (class 1), low noise (class 7) and high noise (class 18). The manual and visual inspection plays a major role in improving the classification accuracy.
Once the classification of the ground class is deemed final, automatic classification is run on all non-ground points. Project specifications for vegetation classification is: above ground points to 50cm - low vegetation (class 3), above 50cm to 2m - medium vegetation (class 4), above 2m - high vegetation (class 5). A manual/visual QC pass is made to fine tune the classification of points, including the manual classification of buildings (class 6) and bridges (class 17). Ground above culverts is left in the ground class. Further classification of water and water bodies is done with the following parameters: water bodies and water courses meeting specifications of greater than 3600 m2 (water bodies), greater than 10 m nominal width (water courses) and at least 220 m in length are delineated. LiDAR points falling within these areas are classified the water class (class 9). Islands greater than 100m2 are delineated within any water feature and not classified as water.
The final 1m Bare Earth DEMs are produced from the TIN of the "ground" and "model keypoint" classes (2 & 8). Similarly, Full Feature grids are produced by using all non-noise points.
