Abstract
Abstract. Image classification stands as an essential tool for automated mapping, that is demanded by agencies and stakeholders dealing with geospatial information. Decreasing costs or UAV-based surveying and open access to high resolution satellite images such as that provided by European Union’s Copernicus programme are the basis for multi-temporal landscape analysis and monitoring. Besides that, invasive alien species are considered a risk for biodiversity and their inventory is needed for further control and eradication. In this work, a methodology for semi-automatic detection of invasive alien species through UAV surveying and Sentinel 2 satellite monitoring is presented and particularized for Acacia dealbata Link species in the province of Pontevedra, in NW Spain. We selected a scenario with notable invasion of Acaciae and performed a UAS surveying to outline feasible training areas. Such areas were used as bounds for obtaining a spectral response of the cover from Sentinel 2 images with a level of processing 2A, that was used for invasive area detection. Sparse detected areas were treated as a seed for a region growing step to obtain the final map of alien species.
Highlights
Current advances in sensors on-board satellite and UAV platforms are the basis for Earth observation and monitoring with an increasing accuracy and resolution
Richardson and Rejmánek assembled a global list of invasive alien trees and shrubs that describes the distribution of taxa and the spatial representation of 622 species that include, as exceptional, 23 species of the genus Acacia and, among them, the Acacia Dealbata Link species (Richardson and Rejmánek, 2011)
The detection of this invasive species through automated procedures and satellite image processing is a subject with a large presence in the literature, with a focus on public platforms such as Sentinel images (Martimort et al, 2012), ASTER or Landsat (Viana and Aranha, 2010)
Summary
Current advances in sensors on-board satellite and UAV platforms are the basis for Earth observation and monitoring with an increasing accuracy and resolution. Existing works comprise the study of biological attributes that are favouring invasion by A. dealbata (Lorenzo et al, 2010) and the production of maps that show how this invasive species is spreading (Martins et al, 2016). The detection of this invasive species through automated procedures and satellite image processing is a subject with a large presence in the literature, with a focus on public platforms such as Sentinel images (Martimort et al, 2012), ASTER or Landsat (Viana and Aranha, 2010). The advantage of these platforms in terms of spatial resolution results in a more detailed cartography, (Mafanya et al, 2017), with a smooth contrast between similar pixels and where the detection is more accurate than with satellite images
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
