Linking Net Assimilation with Multispectral Vegetation Classification to Understand Mesquite-Grass Response to Fire
AuthorSutter, Leland Frederic
support vector machine
unmanned aerial vehicle
AdvisorBarron-Gafford, Greg A.
MetadataShow full item record
PublisherThe University of Arizona.
RightsCopyright © is held by the author. Digital access to this material is made possible by the University Libraries, University of Arizona. Further transmission, reproduction or presentation (such as public display or performance) of protected items is prohibited except with permission of the author.
AbstractUnderstanding vegetation dynamics across space and time has been a grand challenge in Earth sciences, but the induction of remote sensing products has made large-scale mapping of vegetation possible. We initially used Landsat satellites (30 m; eight-day return interval) to assess the Sawmill Fire of 2017 within the Santa Rita Experimental Range. Because of the spatial and temporal decoupling associated with this remote sensing product, important, but smaller-scale disturbances may not be properly captured; this prompted the use of finer scaled data. As such, we used an unmanned aerial vehicle (UAV) equipped with a five band Micasense RedEdge camera for derived land classification and scaling. Additionally, we measured leaf level net assimilated photosynthesis (ANET) to quantify plant function. We repeated the measurements at three points in time at a control and burned site. Spectrally, changes in the Relative Normalized Burn Ratio (RNBR) using Landsat images from directly before the fire and then after the growing season showed minimal evidence of the fire because of its spatial scale, though there were significant impacts from the fire on vegetative physiognomy and ecosystem function. Classifications built from the multispectral camera showed an overall accuracy of 0.89. This study shows the need for fine-resolution data from newly available UAV systems for practical land management practices. Low altitude, fine resolution data, combined with ecophysiological datasets, can be used to quantify and follow tractable land cover changes not captured by our traditional, lower resolution remote sensing sensors and derived products.
Degree ProgramGraduate College