Variability of boreal forest reflectances as measured from a helicopter platform
Journal Articles
Overview
Research
Identity
Additional Document Info
View All
Overview
abstract
Helicopter and surface‐based radiometric measurements of forested sites acquired during the Boreal Ecosystem‐Atmosphere Study (BOREAS) were analyzed to assess reflectance variability and to address the relationships of reflectances and vegetation indices (VIs) with overstory leaf area index (LAI). Statistical measures of central tendency and linear regression analysis, including a multivariate regression which introduced the effects of understory reflectance, were used. The sensors were deployed during all three intensive field campaigns (IFCs) of 1994 and consisted of (1) a helicopter‐based eight‐channel Barnes modular multiband radiometer (MMR) and a ground‐based Spectron Engineering SE‐590 field spectroradiometer to measure spectral reflectance factors, (2) a ground‐based Sun photometer used for atmospheric correction, and (3) a LI‐COR LAI‐2000 plant canopy analyzer and the tracing radiation and architecture of canopies (TRAC) sensor for retrieval of canopy LAI. Of the eight MMR bands, those most strongly correlated to variations in surface LAI were the second and third middle infrared (IR) (1.57–1.80 μm and 2.08–2.37 μm, respectively) with r2=0.45 in both cases, and the red (0.63–0.68 μm) with r2=0.34. In general, linear regression analyses between the canopy LAI measurements and both the helicopter reflectances and VIs resulted in low r2 values commonly less than 0.3. When stratified by season and species, only among sites that were vegetated by aspen (Populus tremuloides) was a stronger relationship observed. In the aspen stands, LAI displayed a traditional trend that maximizes the utility of the VIs: a negative relationship with red reflectance (r2=0.62) and a positive relationship with the near IR (NIR) of r2 = 0.21. The absence of a positive LAI‐NIR reflectance trend degraded the information content of the VIs for the conifers. A multivariate regression confirmed the influence of the understory vegetation on the sparse canopies of the boreal forest and suggested that a more complex model‐based approach is necessary to better characterize variations in canopy reflectance. Atmospheric contamination of the radiometric signal by smoke from forest fires may also have contributed to the initial unsatisfactory results.
