Testing the Contribution of WorldView-2 Improved Spectral Resolution for Extracting Vegetation Cover in Urban Environments

Abstract

Detailed spatial data concerning land cover constitutes valuable information not only for urban design and planning but also to support informed decisions toward urban sustainable development. Obtaining such information from satellite imagery, with quality compatible with cartographic and thematic standards, is still a challenging task. Until recently, very high-resolution (VHR) satellites acquired data with high spatial resolution (1 m or less) but with only four spectral bands, typically in the visible and infrared regions. WorldView-2 (WV-2) offers improved spectral capability by acquiring data in an additional set of spectral bands. Using Lisbon as the case study, this research tests the contribution of the new bands of WV-2 sensor for extracting vegetation information in urban landscapes with different levels of plant heterogeneity. The methodology is based on an automated feature extraction procedure, followed by an object-based accuracy assessment. Results show that including the new WV-2 bands coastal and yellow (bands 1 and 8), along with the standard four-band set, increases the overall accuracy by 1–6%. However, the magnitude of improvement depends on the homogeneity of plants species present in the site.