Sao Tome and Principe Walking Travel Time to nearest Level IV health centre

The least cost path method was used here which is broken down into two steps: (1) the creation of a ‘cost’ allocation surface which can also be referred to as an effort or friction surface and represents the effort to travel across a particular pixel (2) uses the cost allocation (or effort surface) in a least cost path analysis to estimate travel time from every pixel to the nearest destination location (in this case health centres). This was done using Dijkstra’s algorithm to create Dijkstra trees which find the shortest path from one point to another. Full details of the method can be found in Watmough et al. (2022) and the code available in Zenodo (Hagdorn 2021)

The cost allocation surface used three primary input datasets: (1) land cover (2) roads (3) topography. Roads were converted to a 100 m resolution grid with the fastest road being given preference when they overlap. Any pixels with no road get given the land cover. Each pixel is then given a value to represent the speed in which an individual can travel across that pixel considering the land cover or road type. Each of these is then weighted depending on the elevation from the DEM, with pixels that have a slope of more than 45 degrees being masked from the analysis (ie they are too steep for travel). The road types vary for each country depending on the Open Street Map and MapwithAI roads so the speeds are provided in the data download.

Travel was assumed to be walking on all road types and land cover. For motorised transport see the other download for each country.

Watmough, et al. (2022) Using open-source data to construct 20 metre resolution maps of children’s travel time to the nearest health facility, Scientific Data, 9(217)

Hagdorn (2021)

The United Nations Office for the Coordination of Humanitarian Affairs (OCHA) Common Operational Datasets (COD) Administration Level 0 boundary polygons were used in instances where geoBoundaries simplified polygons were not available.

Where countries were not included in the health facility data published by Maina et al. (2019) we used data from the Global Health sites Mapping Project published on Humanitarian Data Exchange (this included: Egypt, Libya, Tunisia, Algeria, Morocco).

For each country we removed health sites that were unlikely to offer child focused services and vaccinations. Facilities that were removed included: pharmacy, dentist, veterinary, café/pharmacy, social facility.

The accuracy of the road data sets has not been validated,

Maina et al. (2019)