Other friction surfaces exist, some of these are global and the majority of these are at a coarser (broader) spatial resolution, for example 1km x 1km or 5km x 5km, whereas our surfaces are generated at a 30m x 30m resolution.
Each existing surface has it's own assumptions which might not be entirely transparent - for example, what speeds were assigned for off-road travel, what speeds for on-road etc? As other sources have provided only the friction surface and not the underlying data allowing construction (i.e. roads, ndvi, buildings etc) there is no flexibility to modify parameters.
We're interested in testing:
- The effect of increasing spatial resolution on the results gained - is there a benefit of using 30m vs 1km?
- Do our assumptions hold when compared with other existing datasets (i.e. Malaria Atlas Project)?
- Are our predictions more accurate (need to discuss how to test this) when compared with other sources?
Associated tasks to follow - this is not an immediate priority, but will involve importing friction surfaces for the same area of interest (malariaAtlas R package and API) and comparing predictions.
Other friction surfaces exist, some of these are global and the majority of these are at a coarser (broader) spatial resolution, for example 1km x 1km or 5km x 5km, whereas our surfaces are generated at a 30m x 30m resolution.
Each existing surface has it's own assumptions which might not be entirely transparent - for example, what speeds were assigned for off-road travel, what speeds for on-road etc? As other sources have provided only the friction surface and not the underlying data allowing construction (i.e. roads, ndvi, buildings etc) there is no flexibility to modify parameters.
We're interested in testing:
Associated tasks to follow - this is not an immediate priority, but will involve importing friction surfaces for the same area of interest (malariaAtlas R package and API) and comparing predictions.