Remote sensing of forest condition in tolerant hardwoods: An examination of spatial scale, structure and function

Sustainable forest management requires objective measures of forest condition and productivity. To help develop these measures, this stud, uses a multispectral remote sensing approach to explore the structural characteristics and physiological condition of vegetation following forest disturbance. The disturbance examined was three silvicultural practices (selection, shelter wood and clearcut) in an over-mature maple forest in the Algoma Highlands near Sault Ste. Marie, Ontario. Since the quality of remote sensing data is scale dependent, studies must be designed with some consideration of the spatial characteristics of the variable of interest. Variables investigated here included aboveground organization of vegetation (structure), and physiological condition of this vegetation following different silvicultural practices. Data were acquired at three altitudes (1000, 2000, and 3000 m), using a Compact Airborne Spectrographic Imager (casi). To uncover the scale and pattern of variation or each altitude, semivariogram analysis was applied Range and sill parameters were derived from semivariogram analysis of overall scene variance using the first principal component (PCA). Variance in physiological condition was analyzed from spectral red-edge inflection point wavelength (AP) data. The resulting variogram analysis indicated spatial resolutions of approximately 9.0 m and 5.0 m for the PCA and AP data, respectively. The sill and range parameters were also successfully correlated to critical forest parameters such as biomass, stocking and crown area. Obtaining a measure of the scale and pattern of variation was a precursor to the use of spectral information for analyzing the relationship of canopy reflectance to leaf-based measures (e.g., chlorophyll and other pigment concentrations) and biophysical features (e.g., crown cover and biomass). Relationships were examined using an ordination technique known as redundancy analysis (RDA). The RDA biplots uncovered a significant relationship between canopy opening and spectral indices. An improved understanding of resolution and condition related to forest disturbance was achieved.