CLIENT: Mosaic Forest Management
PROJECT DESCRIPTION:
Mosaic Forest Management contracted Ecora to complete landcover classification and terrain indices for 700,000 ha on their private and crown chart areas Vancouver Island and coastal mainland BC.
Land cover prediction modeling identified treed/non-treed, roads, and more detailed non-treed land cover/land use classes (snow, avalanche chutes, shrub, transmission lines, roads, wetlands, rock, etc.). The was done in a condensed timeframe, in a semi-automated fashion, with machine learning and then reviewed by photo-interpreters. This information was used in a subsequent remote sensing-based forest inventory.
APPROACH & OUTCOME:
A complex set of terrain indices, examining terrain and slope position at a range of scales (5m to 50m kernels) and neighborhood sizes (25m to 2000m neighbourhoods). These were used to make better predictions of site productivity and species compositions.
We also reviewed sample ground plot locations in a 3D environment to ensure that the calls made on the ground were consistent with what we could see in the lidar and imagery – this was done to confirm the GPS positional accuracy of the ground plots and move the plot to a representative location if it looked like there were issues with the GPS coordinates.