Land cover and impervious surface extraction using parametric and non-parametric algorithms from the open-source software R: an application to sustainable urban planning in Sicily

Abstract

Detailed urban land-cover maps are essential information for sustainable planning. Land-cover maps assist planners in designing strategies for the optimisation of urban ecosystem services and climate change adaptation. In this study, the statistical software R was applied to land cover analysis for the Catania metropolitan area in Sicily, Italy. Six land cover classes were extracted from high-resolution orthophotos. Five different classification algorithms were compared. Texture and contextual layers were tested in different combinations as ancillary data. Classification accuracies of 89% were achieved for two of the tested algorithms.