Global 30 meters spatiotemporal 3D urban expansion dataset from 1990 to 2010

Abstract

Understanding the spatiotemporal dynamics of global 3D urban expansion over time is becoming increasingly crucial for achieving long-term development goals. In this study, we generated a global dataset of annual urban 3D expansion (1990–2010) using World Settlement Footprint 2015 data, GAIA data, and ALOS AW3D30 data with a three-step technical framework: (1) extracting the global constructed land to generate the research area, (2) neighborhood analysis to calculate the original normalized DSM and slope height of each pixel in the study area, and (3) slope correction for areas with a slope greater than 10° to improve the accuracy of estimated building heights. The cross-validation results indicate that our dataset is reliable in the United States(R2 = 0.821), Europe(R2 = 0.863), China(R2 = 0.796), and across the world(R2 = 0.811). As we know, this is the first 30-meter 3D urban expansion dataset across the globe, which can give unique information to understand and address the implications of urbanization on food security, biodiversity, climate change, and public well-being and health.

“Matthew Effect” in Global Major Cities Over Decades: In the context of the spatiotemporal 3D urban expansion

“…This study employed machine learning techniques to construct a global three-dimensional expansion dataset for major cities from 1990 to 2020, and conducted further analysis by integrating socioeconomic data, including population and GDP. In obtaining building heights, we used two datasets as labels for model training, including a global building height dataset from 1990 to 2010 (He et al, 2023) and GEDI height data (Potapov et al, 2021). We adopted an innovative three-step sampling method, which involves "CCDC detection," "pixel selection," and "strati ed sampling" to ensures the scienti c rigor and reliability of the machine learning model.…”

GLAMOUR: GLobAl building MOrphology dataset for URban hydroclimate modelling

“…Since most global DEM datasets fall towards the DSM side which detects the elevation of the surface canopy composed of vegetation and man-made structures 9 , the key challenge here is to produce an accurate representation of the terrain ground. While some algorithms have been developed to discern the top and bottom sections of buildings through morphological operations on global DSM datasets (e.g., ALOS AW3D30 10,11 ), the limited spatial resolution of publicly accessible topographical data often conflates building height with ground elevation in its measurements and thus introduces significant uncertainty when attempting to deduce building heights from these amalgamated figures using straightforward mathematical transformations. Esch et al 12 improved nDSM-based approaches by local height variation analysis aiming to find vertical edges in 12 m TanDEM-X DEM as building outlines and finally generated the World Settlement Footprint 3D (WSF3D), which is the first globally consistent three-dimensional building morphology dataset.…”

GLAMOUR: GLobAl building MOrphology dataset for URban hydroclimate modelling

“…Since most global digital elevation models (DEMs) fall towards the DSM side which detects the elevation of the surface canopy composed of vegetation and man-made structures 9 , the key challenge here is to produce an accurate representation of the terrain ground. While some algorithms have been developed to discern the top and bottom sections of buildings through morphological operations on global DSM datasets (e.g., ALOS AW3D30 10 , 11 ), the limited spatial resolution of publicly accessible topographical data often conflates building height with ground elevation in its measurements and thus introduces significant uncertainty when attempting to deduce building heights from these amalgamated figures using straightforward mathematical transformations. Esch et al .…”