| Annual invasive grasses increases in sagebrush steppe have degraded vast areas in the Western US reducing biodiversity and production of rangelands. Discriminating this plant functional type (PFT) from other vegetation types such as shrubs and perennial grasses at the landscape scale has been challenging because of background (soil and litter) reflectance and effects of shadows using nadir-only view angle passive remote sensors. However, changes in vegetation structure and anisotropic behavior in the middle and late stages of the growing season have enabled us to differentiate some of these PFTs using remote sensing measurements collected at different spatial scales within a shrub-steppe rangeland ecosystem in west-central Idaho. Changes from erectophile to planophile leaf orientation of annual invasive grasses, comprised in this area mostly by medusahead (Taeniatherum caput-medusae [L.] Nevski) produced a distinctive set of vegetation reflectance values for this PFT at ground level during the transition from green to senesce stages. However, measuring plant canopy reflectance depends on the geometry between sensor view angle and solar position. We therefore further demonstrated that medusahead can be detected from native perennial vegetation near solar noon, which could be useful when scheduling multispectral or hyperspectral aerial image surveys. Medusahead did not exhibit strong anisotropic reflectance behavior, quantified through the measurement of bidirectional reflectance factors (BRF) and anisotropy index (ANIX), however the shrub PFT did exhibit high anisotropy in field measurements. We therefore tested whether we could classify areas dominated by shrubs, non-shrubs, and crops at the landscape scale to generate PFT-based productivity estimates using the MODIS Gross Primary Productivity product. Although accuracy classification at moderate scales was high (Khat 88-98%), GPP shrub partitioning was very poor (R2<0.06). |
