Automated Detection of Changes in Built-Up Areas for Map Updating: A Case Study in Northern Italy

Keeping track of changes in urban areas on a large scale may be challenging due to fragmentation of information. Even more so when changes are unrecorded and sparse across a region, like in the case of long-disused production sites that may be engulfed in vegetation or partly collapse when no-one is witnessing. In Belgium the Walloon Region is leveraging Earth observation satellites to constantly monitor more than 2200 redevelopment sites. Changes are automatically detected by jointly analysing time series of Sentinel-1 and Sentinel-2 acquisitions with a technique developed on Copernicus data, based on ad-hoc filtering of temporal series of both multi-spectral and radar data. Despite different sampling times, availability (due to cloud cover, for multispectral data) and data parameters (incidence angle, for radar data), the algorithm performs well in detecting changes. In this work, we assess how such technique, developed on a Belgian context, with its own construction practices, urban patterns, and atmospheric characteristics, is effectively reusable in a different context, in Northern Italy, where we studied the case of Pavia.