Mapping invasive species with a new, user-friendly iterative method.
Kenneth
McGwire, Desert Research Institute, kenm@dri.edu
(Presenting)
Timothy
Minor, Desert Research Institute, tminor@dri.edu
Brad
Shultz, University of Nevada Cooperative Extension, schultzb@unr.edu
Chris
Kratt, Desert Research Institute, chris.kratt@dri.edu
High spatial resolution hyperspectral imagery was collected over study sites in Northern Nevada to test methods of mapping invasive plant species in natural and agricultural settings. The primary targets of this activity were Lepidium latifolium (tall whitetop, perennial pepperweed), Acroptilon repens (Russian knapweed), and Tamarix spp. (saltcedar). A new iterative method of implementing statistical identification was compared to matched-filter and multiple endmember spectral mixture analysis (MESMA). The new method greatly simplified the user’s approach to target identification and provided equivalent or better results to the other methods that were tested. This new method automated much of the sample collection process and allowed the user to easily interact with intermediate results to minimize false-positives. In addition, the new method provided better correlation to percent cover of the target species. The availability of airborne hyperspectral data is increasing and vendors are now able to provide highly refined geo-corrected, reflectance products. High spatial resolution hyperspectral data is allowing species-level identification of some plants and the detailed imagery is of a scale that allows improved image interpretation by non-specialists. This new data analysis approach is intuitive, much like a painting program, and provides a practical way of moving hyperspectral data analysis out of the lab and into the hands of field personnel and stakeholders.
NASA Carbon Cycle & Ecosystems Active Awards Represented by this Poster: