Evaluating free and simple remote sensing methods for mapping Chinese privet (Ligustrum sinense) invasions in hardwood forests

Abstract

Chinese privet (Ligustrum sinense) is a common invasive shrub in hardwood forests of the southeastern US and has been shown to negatively affect native herbaceous and woody plants. The ability to map the distribution of L. sinense on a property could help land managers plan and budget for control operations. We evaluated whether freely available moderate resolution multispectral imageries (Landsat 8 and Sentinel 2) and open-source GIS software (QGIS with the Semi-Automatic Classification Plugin) could be effective tools for this application. These tools are widely used by remote sensing and mapping professionals; however their adoption by field-level land managers appears limited, and their utility for mapping L. sinense invasions is untested. We evaluated how satellite type, image acquisition date, classification algorithm, and L. sinense cover affected detection accuracy. We found that Sentinel 2 imagery from March tended to produce good results, especially when analyzed using the maximum likelihood algorithm. Our best classifier obtained an overall accuracy of 92.3% for areas with ≥ 40% L. sinense cover. We recommend that land managers interested in applying this tool use an adaptive process for developing training polygons and test multiple images and classification algorithms in order to achieve optimal results.