Using canopy reflectance models and spectral angles to assess potential of remote sensing to detect invasive weeds

One of the goals of applied remote sensing is to map locations of invasive weeds. However, differences in plant cover and leaf area index (LAI) alter canopy reflectance, making detection of a single species difficult. Variation in canopy reflectance may be simulated using the Scattering by Arbitrarily Inclined Leaves ( SAIL) model. Simulated reflectances are used to calculate spectral angles to determine the separability of an invasive weed from co-occurring vegetation. Leafy spurge is a noxious invasive weed with yellow-green flower-bracts. Spectral angles from SAIL model simulations show that flowering leafy spurge may be detected when LAI is greater than 1.0 and flower-bract cover is greater than 10%. A threshold of 3.5 degrees (0.061 radians) was determined to provide the best separation between leafy spurge and co-occurring vegetation. To test this prediction, the Spectral Angle Mapper was used to classify leafy spurge using AVIRIS, Landsat ETM+ and SPOT data. Classification accuracy was inversely related to simulated spectral angles from the SAIL model analyses. Using canopy reflectance models and spectral angles may help identify those invasive species that are potentially detectable by remote sensing, and may indicate the conditions where detection will be problematic based on variation of LAI, cover and other variables.