A hybrid multi-step approach for urban area mapping in the Province of Milan, Italy.

Remote sensing products have proven an effective tool for the study of urban areas, from city management to environmental monitoring. This work focuses on the mapping of urban areas in the Province of Milan, Northern Italy, using mid-resolution remote sensing data covering the last 20 years. The methodology consists of three main steps: (i) a pixel-based classification tree, (ii) object-based filtering of agricultural terrain, and (iii) joining of land cover classes into two (urbanized and non-urbanized) ,adding post-classification editing. The final derived urban maps were validated and demonstrated to reach very good accuracy (error: 7-13%), thus providing reliable thematic information for urban planning of local and regional authorities. [ABSTRACT FROM AUTHOR]