SPATIAL LINKAGE AND URBAN EXPANSION: AN URBAN AGGLOMERATION VIEW

Given the limited coverage or spatial resolution of existing datasets, we develop the first global building height map (3D-GloBFP) at the building footprint scale using Earth observation datasets and advanced machine learning techniques. Our map reveals the complex 3-D morphology of the world's building heights at a finer scale and provides reliable results (i.e., R²: 0.66–0.96, RMSEs: 1.9 m–14.6 m) over global regions 3D-GloBFP has great potential to support both macro- and micro-urban analysis

Most existing global urban products with future projections were developed in urban and non-urban categories, which ignores the gradual change of urban development at the local scale. Using annual global urban extent data from 1985 to 2015, we forecasted global urban fractional changes under eight scenarios throughout 2100. The developed dataset can provide spatially explicit information on urban fractions at 1 km resolution, which helps support various urban studies (e.g., urban heat island).

A global land cover map with fine spatial resolution is important for climate and environmental studies, food security, or biodiversity conservation. In this study, we developed an improved global land cover map in 2015 with 30 m resolution (GLC-2015) by fusing the existing land cover products based on the Dempster–Shafer theory of evidence on the Google Earth Engine platform. The GLC-2015 performed well, with an OA of 79.5 % (83.6 %) assessed with the global point-based (patch-based) samples.

A large-scale and high-resolution urban green space (UGS) product with 1 m of 31 major cities in China (UGS-1m) is generated based on a deep learning framework to provide basic UGS information for relevant UGS research, such as distribution, area, and UGS rate. Moreover, an urban green space dataset (UGSet) with a total of 4454 samples of 512 × 512 in size are also supplied as the benchmark to support model training and algorithm comparison.