Identifying optimal remotely-sensed variables for ecosystem monitoring in Colorado Plateau drylands. (June 2018)

• Record Type: Journal Article
• Authors: Poitras, Travis B.; Villarreal, Miguel L.; Waller, Eric K.; Nauman, Travis W.; Miller, Mark E.; Duniway, Michael C.
• Is Part Of: Journal of arid environments. Volume 153(2018)
• Page Start: 76
• Page End: 87
• Publication Date: 2018-06
• Subjects: Landsat -- Land cover change -- Ecological state mapping -- Biological soil crust -- Grasslands -- Vegetation index; Arid regions ecology -- Periodicals; Arid regions -- Periodicals; Écologie des régions arides -- Périodiques; Régions arides -- Périodiques; 577.54
• DOI: 10.1016/j.jaridenv.2017.12.008
• Languages: English
• ISSNs: 0140-1963

Abstract

Abstract: Water-limited ecosystems often recover slowly following anthropogenic or natural disturbance. Multitemporal remote sensing can be used to monitor ecosystem recovery after disturbance; however, dryland vegetation cover can be challenging to accurately measure due to sparse cover and spectral confusion between soils and non-photosynthetic vegetation. With the goal of optimizing a monitoring approach for identifying both abrupt and gradual vegetation changes, we evaluated the ability of Landsat-derived spectral variables to characterize surface variability of vegetation cover and bare ground across a range of vegetation community types. Using three year composites of Landsat data, we modeled relationships between spectral information and field data collected at monitoring sites near Canyonlands National Park, UT. We also developed multiple regression models to assess improvement over single variables. We found that for all vegetation types, percent cover bare ground could be accurately modeled with single indices that included a combination of red and shortwave infrared bands, while near infrared-based vegetation indices like NDVI worked best for quantifying tree cover and total live vegetation cover in woodlands. We applied four models to characterize the spatial distribution of putative grassland ecological states across our study area, illustrating how this approach can be implemented to guide dryland ecosystem management. Highlights: Sparse dryland vegetation can be difficult to measure using remote sensing. We use monitoring plots and coincident Landsat data to assess various spectral models. Percent cover of bare ground was accurately modeled across vegetation communities. Shortwave infrared and red bands were important in most models.