Abstract
Urban boundaries are essential indicators for understanding spatial structure and dynamic changes in human settlements. Most existing high-resolution urban boundaries datasets performed poorly on distinguishing low-density built-up areas from non-urban spaces and overlooking the role of population distribution in defining urban extents. In this study, we developed a dual-threshold method by integrating impervious surface density and population data to map annual city and town boundaries from GISD30 (global 30 m impervious-surface dynamic dataset). Specifically, we combined kernel density estimation and cellular automata algorithms to generate global urban boundaries, and then differentiated urban settlement types (e.g., cities and towns) based on population thresholds, thereby producing the Global City and Town Boundaries (GCTB) dataset at 30 m resolution for the period 2000-2022. The GCTB dataset achieves strong agreement with the high-resolution urban boundary interpretation dataset-Atlas of Urban Expansion (R² > 0.88). Using OpenStreetMap place tags, the City/Town split reaches precision 0.80 (City) and 0.65 (Town) with overall accuracy 0.75. Therefore, GCTB provides essential spatial information for global urbanization and sustainable-development planning.