Table 1 Comparison of network- and raster-based methods with open data
| Â | Network-based | Raster-based |
|---|---|---|
Critical datasets | OSM network | Friction layer |
Dataset update frequency | Daily The scope of a daily dataset update is not known as it is based on VGI | No info available Last update: 2020 |
Data incompleteness exposure | Very high The calculation is only based on the road network, thus the analysis is limited by the completeness of the mapped network | Medium Friction layer construction mitigates the lack of road network completeness |
Spatial resolution | 100Â m Area extent is a vector geometry and thus does not have resolution limitations. Population datasets were available at a 100Â m resolution | 1Â km Friction layer was available at a 1Â km resolution |
Time resolution | Time intervals The loss calculation cannot retrieve loss of access time within the same time range. Time intervals may be reduced to increase the number of time ranges but this would increase the computation time | Limited by cell size According to Nelson [71], travel speed on tracks for motorized vehicles is 10Â km/h and walking speed on bare ground is 2.5Â km/h. Considering the granularity of the cell the time resolution would be 6Â min for motorized vehicles and 24Â min for walking |
Computation time Calculated for the specific analysis of Cyclone Idai case study with the computational environment described in Additional file 3 |  ~ 3.5 h The main time consuming step was the isochrones processing |  ~ 1 h The main time consuming step was downloading the friction layer |