Global urban and rural settlements (GURS) and global urban envelopes (GUE) dataset

Update Notes for the Global Urban and Rural Settlements (GURS) Dataset (1990-2025)

The Global Urban and Rural Settlements (GURS) dataset has been updated to cover the period from 1990 to 2025. It is provided as 100m-resolution raster data in the Mollweide projection, which can be directly used for area statistics.

This update introduces an optimized algorithm that utilizes global administrative boundaries (GADM) to conduct local spatial evaluations. The algorithm groups mega villages by their corresponding local administrative units and applies a step-by-step analysis to reclassify them into either an urban context or a rural context. The sequential classification process is as follows:

1.     Step 1: If an administrative unit contains only one mega village, this patch is directly classified as an urban context.

2.     Step 2: If multiple mega villages exist within an administrative unit, the algorithm identifies patches with exceptionally large areas. A mega village is classified as an urban context if its area strictly exceeds the local mean area plus one standard deviation.

3.     Step 3: If multiple mega villages exist within an administrative unit but none meet the criteria in Step 2, the algorithm selects the single largest mega village within that unit and classifies it as an urban context. All remaining mega villages that do not meet the above sequential criteria are classified as rural contexts.

By applying this local evaluation method, the optimized dataset better identifies urban settlements in early historical years. Furthermore, it significantly improves the recognition capacity of urban contexts in underdeveloped regions, such as Africa.

Release Notes for the Global Urban Envelopes (GUE) Dataset (1990-2025)

The Global Urban Envelopes (GUE) dataset provides a seamless and highly refined delineation of physical urban boundaries across the globe from 1990 to 2025. It is provided as Shapefiles utilizing the Mollweide projection. This dataset builds upon the extracted urban settlements from our GURS product and applies a hybrid raster-vector morphological approach to overcome the fragmented and pixelated nature of raw grid data.

Methodology overview:

1.     Morphological connection: We first applied binary closing and hole-filling algorithms on the raster matrices. This step spatially connected fragmented urban pixels within a 200-meter proximity and eliminated internal voids. To prevent edge artifacts during this massive computation, an overlapping window-reading strategy was strictly employed.

2.     Seamless dissolution: The contiguous urban pixels were then vectorized and physically dissolved into global seamless entities, ensuring absolute topological integrity across regions.

3.     Envelope smoothing: Isolated patches smaller than 1 square kilometer were filtered out. We subsequently applied a vector-based buffer-debuffer smoothing technique (expanding and contracting by 250 meters) to the remaining entities. This critical step eliminated pixelated jagged edges, finalizing the naturally continuous and smoothed urban envelopes.

The GUE dataset demonstrates enhanced capability in identifying urban contexts in underdeveloped regions, such as Africa, thereby providing a solid foundation for unbiased global urban research.

Visual Comparisons:

Four schematic figures are provided to visually demonstrate the optimizations of the updated GURS dataset compared to its previous version, as well as the advantages of the GUE dataset over the GUB dataset in identifying urban contexts in the African region.