Forest Investment Account (FIA) - Forest Science Program
FIA Project Y093062

    Integration of airborne LiDAR and hyperspectral remote sensing data to support the Vegetation Resources Inventory and sustainable forest management
Project lead: Niemann, Olaf (University of Victoria)
Author: Niemann, K. Olaf
Subject: Forest Investment Account (FIA), British Columbia
Series: Forest Investment Account (FIA) - Forest Science Program
Research Question: Can spatially integrated LiDAR and hyperspectral remote sensing data be used to support the BC Vegetation Resources Inventory (VRI)Programme? What kinds of inventory attributes can be accurately derived from these RS data, and how do the results compare to a traditional photo-based VRI?

Research Purpose: To evaluate the potential capabilities of airborne LiDAR/hyperspectral data for forest inventory and ecological monitoring applications.

Project Description (from original Full Proposal Y07-1062):

In 1995, BC Ministry of Forests introduced the Vegetation Resources Inventory (VRI) as an improved and enhanced set of methodological standards and procedures to assess the quantity, quality, and distribution of BC’s timber and non-timber (shrubs, herbs, and bryoids) vegetation resources. VRI has since played an important role in BC’s Timber Supply Review, Annual Allowable Cut, Provincial and National Forest Inventories, strategic business and operational planning at the management unit level, and more recently for sustainable forest management (Parminter, 2000). VRI was designed as a photo-based, two-phased vegetation inventory process (also known to biometricians as ‘double sampling for stratification’; Cochrane, 1977). Phase I involves (i) manual delineation of the vegetated land base into discrete, homogeneous units (polygons) of distinct land cover types using mid-scale (1:10,000 to 1:30,000) aerial photographs, and (ii) estimation of a broad range of terrain, biophysical, and ecological attributes for each of these polygons using traditional photo-interpretative techniques and other existing ancillary data sources. Phase II focuses on the acquisition of ground-reference data from a stratified and randomly selected set of polygons to help reduce any measurement error or bias found in the initial Phase I estimates. For the treed component of the vegetated land base, Phase I polygons represent forest stands that are relatively (spatially) homogeneous with respect to disturbance history, species composition, and edaphic factors. Phase I and II attributes of interest are live tree cover pattern, crown closure, canopy layering (single or multilayered), vertical and horizontal complexity, species composition, age, height, basal area, volume, and stem density.

Recent advances in airborne remote sensing (RS) sensors, applications, and data processing indicate that the combined use of two commercially available technologies may further improve the quality, timeliness, and cost-effectiveness of both phases of the VRI. First, airborne laser scanning or LiDAR (light detection and ranging) is an active RS technology that utilizes high-frequency, pulsed laser light to measure the location and 3-D geometry of objects on the ground. Numerous studies published over the past decade demonstrate repeatedly that LiDAR is capable of accurately measuring ground-surface elevations, individual tree and stand heights, stem density, volume, basal area, and aboveground biomass/carbon with high precision (many now argue better than ground-reference measurements) (Lefsky et al., 2002; Lim et al., 2003; Reutebuch et al., 2005). Second, hyperspectral RS is an emerging and complementary technology that captures a nearly continuous reflected shortwave energy spectrum ranging from the visible to shortwave infrared (400 – 2500 nm) using an airborne imaging spectrometer. Hyperspectral sensors have the unique ability to acquire detailed spectral information related to species composition (Franklin et al., 2000; Leckie et al., 2003a; Roberts et al., 2004), nutrient and moisture status (Niemann et al., 2002), chlorophyll content, productivity, environmental stress, and natural disturbance (Treitz and Howarth, 1999; Ustin et al., 2004). Integration of LiDAR and hyperspectral data therefore provides both a spatially and spectrally rich data set, with LiDAR contributing a third spatial dimension (height) to the horizontally and spectrally continuous imagery generated by hyperspectral imaging sensors. So far, these integrated data sets have been successfully used to map (i) individual tree stems and crowns (Leckie et al., 2003b; Coops et al., 2004) for plot- and stand-level estimates of height, density, basal area, volume, and biomass (McCombs et al., 2003; Popescu et al., 2004); (ii) stand-level forest canopy surface albedo and rugosity (Ogunjemiyo et al., 2005); chlorophyll content (Blackburn, 2002); wildlife habitat (Hill and Thomson, 2005), and forest structure and seral stage (Tiede et al., 2004).

We anticipate that the continued development and application of integrated airborne LiDAR and hyperspectral RS datasets in the area of operational forest management may lead to substantial advancements and efficiencies in forest inventories in the following ways:
1) Integrated LiDAR and hyperspectral RS datasets could be used to derive a broad range of Phase I attributes (i.e., cover pattern, crown closure, vertical complexity, height, volume, basal area, stem density, species composition, diversity, and vertical stratification) to populate an existing manual photo-based or semi-automated RS-based Phase I polygonal coverage.
2) Well calibrated RS datasets could be used to expand the number and spatial distribution of Phase II ground-reference plots to more remote and inaccessible regions of the vegetated land base, or other areas where there is substantial uncertainty in the quality of photo-interpreted Phase I attributes.
3) Estimates of vegetation height, canopy cover, and surface texture derived from LiDAR sensors could be combined with spectral data (ratios, indices, derivatives, or other synthetic spectral components) as inputs to semi-automated stand delineation and classification procedures.
4) A quantitative RS approach to VRI would allow for rapid measurement of continuous or categorical inventory variables that are precise, replicable, and lend themselves well to statistical analyses.
5) Integrated LiDAR and hyperspectral RS datasets may lead to more meaningful and ecologically relevant measures of ecosystem composition, structure, and function that would support the criteria and indicators (C&I) approach of SFM, effectiveness monitoring, adaptive forest management, and forest certification.
Related projects:  FSP_Y071062FSP_Y082062


Executive summary (75Kb)
Final report (1.4Mb)

To view PDF documents you need Adobe Acrobat Reader, available free from the Adobe Web Site.

Updated August 16, 2010 

Search for other  FIA reports or other Ministry of Forests and Range publications.

Please direct questions or comments regarding publications to