Skip to content

Releases: Bigtapha/samareew

Release list

SamaReew 1.0.0

SamaReew 1.0.0 Latest
Latest

Choose a tag to compare

View all tags
@Bigtapha Bigtapha released this 13 Oct 09:31
v1.0.0
2ff6f88

First public release of SamaReew, an open-data atlas of urban expansion in four Senegalese cities: Dakar, Touba, Thiès and Mbour. Built-up areas are mapped every year from 2016 to 2025 at 10 m, from Sentinel-2 imagery, with one national Random Forest model and the same rules for every city.

Website: https://bigtapha.github.io/samareew/

Downloads

File For whom Content
samareew-data.zip (6 MB) Data users Indicator tables, rasters, boundaries, validation sample, cross-checks, method. Start with its README.md.
site-assets.tar.gz (23 MB) Site deployment only Map tiles and basemap downloaded by the deployment workflow. You do not need it to use the data.
SHA256SUMS Everyone Checksums of the two archives.

What the data show

Dakar Touba Thiès Mbour
Mapped built-up area, 2016 → 2025 (ha) 21,134 → 31,775 8,542 → 14,578 3,721 → 6,813 4,330 → 7,813
Growth +50 % +71 % +83 % +80 %
Built-up m² per inhabitant 60 → 76 88 → 109 84 → 124 88 → 123
Accuracy, built / not built in 2025 0.87 ± 0.06 0.83 ± 0.07 0.86 ± 0.09 0.82 ± 0.09

In all four cities, built-up land grew faster than population.

Read this before using the figures

The map was checked against 480 random points, photo-interpreted without knowing the map class. The tables and rasters keep the mapped values; the corrected areas and their 95 % margins are published next to them (validation/accuracy.json).

  • Total built-up area in 2025 is reliable: 61,000 ha mapped against 60,400 ± 7,400 ha after correction.
  • New built-up area runs ahead of the roofs: about half of the mapped new built-up area is truly built in 2025. Another quarter is unbuilt subdivision or construction site, which Sentinel-2 sees as built-up.
  • Growth types (densification, extension, leapfrog) are indicative. The ranking of cities is stable, the percentages are not.
  • 2025 is provisional, and the year of change is right within one year for two points out of three.
  • Thiès is the least reliable city. Its largest known false detection, a 75 ha irrigated farm at Fandène, is left in the map and outlined on the site and in known_errors/map_notes.geojson. It accounts for 2 points of the city's growth (+83 % with it, +81 % without).

The map is never corrected by hand: known errors stay in the data and are documented. The full method, settings and limits are in METHODOLOGIE.md (French).

Licence and citation

  • Code: MIT.
  • Data: CC BY 4.0, except boundaries/communes.geojson (ODbL 1.0, © OpenStreetMap contributors).
  • How to cite: see CITATION.cff.

Built from commit 2ff6f88.

Assets 5
Loading