Satellite Detection of Inter-Row Management Practices in a North-Italy Vineyard: Preliminary Results

Independent, large-scale assessment of agricultural practices can be useful for inventorying and certification purposes. Satellite Earth observation technology may represent a powerful tool to implement this application. In this paper we analyze the problem of identifying from NDVI sequences the type of inter-row management in vineyards, i.e. whether vegetation is allowed to develop in the inter-row space or this latter is tilled and soil remains exposed. Typical sequences are analyzed from Sentinel-2 acquisitions over a test site in northern Italy, and both supervised and unsupervised classification is applied to features extracted from time sequences. The accuracy levels reach high values only in some cases, but joint analysis of features and resulting accuracy levels is providing significant clues laying the basis for future improvement.