Sentinel-2 time series for mapping Ailanthus altissima invasion on Google Earth Engine

Ailanthus altissima, also known as the "tree of heaven" is an early successional tree originating in Asia, belonging to the Simaroubaceae family, which has become invasive worldwide. The dispersion of its winged seeds by wind, water and machinery even at considerable distances, the ability of its root system to generate numerous suckers and progeny plants, as well as its adaptability to diCerent type of soil and water regime, favor the spread of the species, especially in disturbed areas. A. altissima diCusion has aCected diverse environments, with occasional formation of dense and monospecific aggregates. A. altissima threatens biodiversity through competition, population reduction and extinction of native species and habitat modification, causing ecosystem degradation and resulting in biodiversity loss. ECective management and control of invasive alien species is needed to reduce pressure on ecosystems biodiversity. To achieve this goal it is essential to know the occurrence, status and impact of alien species through their spatial distribution. Remote sensing has long been used as a valuable tool for mapping plants alien species, due to the capability to provide an overview over large geographical extents, for identifying alien species locations, coverage, abundance etc. The high spatial resolution (10 meters) and short revisit time (5 days) of the Copernicus Sentinel-2 imagery allows the identification of the single trees' crowns, at least the larger ones. Sentinel-2 images also have a good spectral resolution, oCering three phenology-aimed red edge bands, albeit at a coarser resolution. Unfortunately, the plethora of information contained in Sentinel-2 imagery o^en translates into a slow image search, download and processing time, whenever the processing is carried locally, via traditional GIS and remote sensing algorithms. The Google Earth Engine platform (GEE) is a relatively recent addition to the remote sensing toolbox. GEE cloud computing is a groundbreaking paradigm shi^ which allows quick and easy access to all openly available satellite imagery (Sentinel, Landsat, MODIS, etc), as well as other climate and geomorphology related datasets. Besides, all the calculations can be carried on the cloud, extracting and downloading only the algorithms' output, both as raster and vector GIS-ready layers. We propose a fully automated GEE cloud-based tool for A. altissima spatial distribution reconnaissance. Our procedure exploits spectral vegetation indices time series and the automated classification tools oCered by GEE, outpu3ing a classified georeferenced vector layer. Our algorithm is based on a set of ground truth observations, it needs no user-operated imagery manipulation and in principle it can be extended to other tree species.