Regional Scale Dryland Vegetation Classification with an Integrated Lidar-Hyperspectral Approach

摘要: Author(s): Dashti, H; Poley, A; Glenn, NF; Ilangakoon, N; Spaete, L; Roberts, D; Enterkine, J; Flores, AN; Ustin, SL; Mitchell, JJ | Abstract: © 2019 by the authors. The sparse canopy cover and large contribution of bright background soil, along with heterogeneous vegetation types in close proximity, are common challenges for mapping dryland remote sensing. Consequently, results a single classification algorithm or one type sensor to characterize typically show low accuracy lack robustness. In our study, we improved semi-arid ecosystem based on use optical (hyperspectral) structural (lidar) information combined environmental characteristics landscape. To accomplish this goal, used both spectral angle mapper (SAM) multiple endmember mixture analysis (MESMA) classification. Lidar-derived maximum height delineated riparian zones were then modify Incorporating lidar into scheme increased overall from 60% 89%. Canopy structure can have strong influence variability provided complementary SAM's sensitivity shape but not magnitude spectra. Similar approaches map regions drylands uncertainty may be readily implemented unmixing algorithms applied upcoming space-based imaging spectroscopy lidar. This study advances understanding nuances associated xeric mesic regions, highlights importance incorporating sensors accurately heterogeneity ecosystems.