|
Shrub Abundance in Alaskan Arctic Tundra from MODIS Reflectance Anisotropy
Mark
James
Chopping, Montclair State University, chopping@pegasus.montclair.edu
(Presenter)
Increasing shrub abundance has been observed in Arctic tundra over the last 60 years and this is important because shrubs are present over a very large area and are able to expand rapidly, with consequences for ecosystem structure and function, albedo, and feedbacks to climate. Although the warming of the climate may result in rapid expansion, currently fractional tall shrub cover is very low at the scale of NASA's moderate resolution imagers (MODIS, MISR) and is rarely >0.1. The difficulty of mapping vegetation at high latitudes using passive solar wavelength remote sensing is well known, particularly for pixels with < 0.2 fractional cover. Together with dark and heterogeneous tundra surfaces (rarely >0.06 in the red wavelengths except where bright lichen species are abundant) and high solar zenith angles (>50 degrees, even in summer, resulting in a relatively large diffuse:direct irradiance ratio), this low shrub cover makes the detection and measurement of shrubs from space extremely challenging. Here we exploit measures of surface physical structure encapsulated in the anisotropic scattering kernel weights of linear, semi-empirical kernel-driven models of the surface bidirectional reflectance distribution function (BRDF). MODIS Terra/Aqua red band BRFs accumulated over DOY 169 to 184 2010 were used to invert the RossThick-LiSparseMODIS model using the AMBRALS algorithm. Reference data were obtained for fourteen sites along the Chandler River on the North Slope of Alaska, based on field belt transects, high resolution imagery, and ground photography data. Multiangle field radiometry provided estimates of typical shrub-free tundra BDRFs. Relationships between the per-site estimates of fractional cover, mean shrub height, and the product of these (a surrogate for woody biomass) and BRDF model kernel weights are not generally linear, with the exception of the geometric kernel weight that has a reasonably strong and quasi-linear relationship to measured tall shrub cover, height, and cover x height. This is likely to reflect the strong signal from canopy-scale shadowing relative to the relatively minor increase (decrease) in volume scattering (isotropic scattering). These results indicate that the RTLS BRDF model geometric kernel weight has potential for mapping shrub abundance even when both cover and contrast between shrub crowns and the background are very low. Presentation: 2011_Poster_Chopping_312_359.pdf (3760k) Presentation Type: Poster Session: Other (Wed 10:00 AM) Associated Project(s):
Poster Location ID: 312
|