Integrating liDAR and optical remote sensing for forest inventory in the NWT

Dr R.J. Hall, Natural Resources Canada, Canadian Forest Service, Edmonton, presented a seminar on 5 April 2012 at the University of Alberta.

The forested area of the Northwest Territories (NWT) in Canada is a widely distributed land mass 33 million ha in size much of which is inaccessible, costly to survey, and containing inventory information for less than 10 percent of its forests. As a result, being able to report on whether forests in the NWT are a carbon sink or source, and reporting on the NWT state of the environment become challenging tasks. A study area 200,000 km2 in size comprising much of the most timber productive forests within the southern Taiga Plains Ecozone was identified as an area for which more current forest inventory information was needed to support these federal and territorial government reporting objectives.

The only spatially contiguous inventory available was a recent satellite land cover map. To improve its informational value, particularly for forest inventory, could more detailed information regarding forest structure, age class distribution, volume and aboveground biomass be generated as value added attributes from which a forest inventory map could be generated?

Considerable research has appeared in the literature on deriving estimates of forest structure from airborne and to a smaller degree, spaceborne LiDAR data. Acquiring spatially contiguous airborne LiDAR data over large remote regions, however, is cost prohibitive. An alternative approach would be imaging small areas within a sampling, data integration framework. ICESat, a spaceborne satellite that carries the Geoscience Laser Altimeter System (GLAS), is a large footprint waveform LiDAR sensor. The density of satellite LiDAR footprints increases at more northerly latitudes and its potential for characterizing canopy height profiles was of particular interest for its potential application over forested areas in the NWT. A contiguous sample of airborne LiDAR was used to relate field plot sample estimates of forest structure to airborne LiDAR metrics which in turn, was used to calibrate metrics derived from ICESat-GLAS waveform data. Application of these models resulted in point estimates of stand attributes that are being used in a k-Nearest Neighbour spatial imputation algorithm. This process results in spatial stand attribute values that are being stratified with a land cover map to generate a satellite-based forest inventory over the study area. The integration of field, airborne and satellite LiDAR is enabling a characterization and mapping of forest structure, volume and aboveground biomass that has not previously been possible in this region of the northern boreal forest.