Developing L2 urban classification in python
The L1 urban classification has three levels of urban classification:
- High density urban (HD) - minimum population density 1500/km2, total settlement population: 50000
- Urban clusters (Urban) - minimum population density 300/km2, total settlement population: 5000
This classification is taken further in the DEGURBA classification (L2). Details on the complete methodology can be found here:
| Class | Name | Pop Density | Total Pop | Distance | Notes |
|---|---|---|---|---|---|
| 30 | Urban Centre | >1500 | >50000 | hole-smoothed | |
| 23 | Urban cluster, town, dense urban cluster | >1500 | >5000 | not type 30 | |
| 22 | Urban cluster, town, semidense urban cluster | >300 | >5000 | farther than 3km from 23 or another 22 | |
| 21 | Urban cluster, suburb | >300 | >5000 | within 3km of 23 or another 22 | |
| 13 | Rural cluster | 300 | >500, <5000 | ||
| 12 | Low density | 50 | |||
| 11 | Very low density | remainder |
The DEGURBA code has been replicated in python; the results are very similar, although not perfect. Note that this analysis relies solely on the population grid, it does not consider the built-area dataset. This is a decision made both to simplify the code, and to simplify replication.
When exploring the results and the original data, there are a number of questions that arise, see the comparison of three maps below
- The comparison between the datasets at 1km are very similar. There are a couple issues, but before I go digging into the reasons, I want to make sure I am working on the correct data. Could you let me know where the authoritarian population and GHS-SMOD data can be downloaded?
- The issues in the python code seem to mostly be related to the distance calculations for classes 21 and 22. As I play around with the code, I wonder how you deal with queen's case contiguity when calculating distances.
