A New European Settlement Map From Optical Remotely Sensed Data

Abstract

An application of a general methodology for processing very high-resolution imagery to produce a European Settlement Map (ESM) in support of policy-makers is presented. The process mapped around 10 million ${\text{km}}^{2}$ of the European continent. The input image data are satellite SPOT-5/6 pan-sharpened multispectral images of 2.5- and 1.5-m spatial resolution, respectively. This is the first time that remote sensing technology demonstrates capability to produce a continental information layer using 2.5-m input images. Moreover, it is the highest resolution continental map produced so far. The presented workflow is data-driven and consists in fully automatic image information extraction based on textural and morphological image analysis. The learning method allows the processing of high-resolution image data using coarse resolution thematic layers as reference. Validation shows an overall accuracy of 96% with omission and commission errors less than 4% and 1%, respectively.