This is a compilations of what I sent to other people.

I have taken it from the cadastre at (https://seduvi.cdmx.gob.mx/programas-delegacionales-de-desarrollo-urbano), I was extremely surprised that CDMX’s cadastre has not being used because it is in the public domain (“Este programa es de carácter público, no es patrocinado ni promovido por partido político alguno y sus recursos provienen de los impuestos que pagan todos los contribuyentes.”) You only need to credit the source: “Agencia Digital de Innovación Pública “Sistema Abierto de Información Geográfica (SIGCDMX)” Disponible en: https://sig.cdmx.gob.mx/”.

After reading osm.wiki/Import/Guidelines, I will make a import page for CDMX data to serve as a documentation for where the data comes from.

Step 1

I download the data from (https://sig.cdmx.gob.mx/datos/descarga). There are two types of data there, the actual cadastry data “Descarga de datos del catastro” and landuse data “Descarga de datos SEDUVI”. The cadastry data is basically a bunch of points and I find that the only useful data there is the address data, which is fairly inaccurate.

It is better to download the landuse data since it has the polygons to work with. I download the shapefile from the “Descarga de datos SEDUVI” section. I’m working at Cuauhtémoc alcadia because it is the central alcadia of CDMX

Generally, since houses in Mexico are connected to each other, often the landuse is the building footprint itself. But, for more complex places like schools, churches, hospital, government areas, etc. it only covers the surrounding of the complex.

Step 2

I then use an application called QGIS to process the shapefile data, by open the .shp file. There are 4 files in that folder that are meant for different functions, fortunately QGIS is smart enough to parse them all. There are a couple of things that I need to do: - I changed the projection from default to EPSG:4326 (osm.wiki/Converting_to_WGS84#QGIS) - Using the Processing toolbox, I use k-means clustering and split vector layer tools to split the alcadia data to 30 files, which is an arbitrary number. This make the files much easier to manage and review in JOSM. - I export it to geojson format, which is readable by JOSM and OSM.

Step 3

I open one file to JOSM and then on another layer, I download the OSM data around the area. We now have two layers, our SIGCDMX data and OSM data. This is the hardest step since I need to figure out how to make the SIGCDMX data live peacefully with OSM data. This is how I do it, in order:

1. Use Ctrl+F, search “type:way” to select all ways. Then I remove all the junk tags and add “building=yes”, since 90% of the polygons reflect the actual building footprints in Mexico because all the houses are connected.
2. Simplify the areas with a JOSM plugin named “SimplifyArea”. With all ways selected, I click a button that activate the plugin, and a lot of useless nodes are removed! Then, I go to the Validation tool, click verify, then it would spit out an error of “duplicated nodes”. I then press the wrench to auto-fix it, and it works like a charm!
3. For places with existing OSM data, I either spend my time adjusting the data node-by-node, occasionally splitting the data or such. Or, I use “Ctrl+C” and “Ctrl+Shift+V” to copy the tags from the old OSM shapes to the new geometries, then delete the old ways. Sometimes I saw nodes with the address, to which I usually merge them to the building or remove them.
4. Reclassify landuses for the remaining 10%. A lot of them are basically “landuse=construction” (people building something), “landuse=brownfield” (dirt left unused), or “amenity=parking” (mexicans park their cars on empty spots). But some of them are also schools, commercial areas, industry, parks, etc. During this time I also be mindful to align the roads to the new buildings, or the service roads to the parking lots.
5. We still have two layers: SIGCDMX data and OSM data. After doing most of the processing work, I merge the SIGCDMX layer to OSM layer, by right click on it, merge, and click through all the warnings since we have done the processing work.
6. I then continue to look for problematic areas, sometimes drawing new buildings or edit the geometries. Afterwards, I use the verification tool to look for common errors such as “highways overlap buildings” or “building inside buildings” or “building overlaps buildings” or “self-intersect ways”, etc. You get the idea.
7. Before actually uploading, I then press the upload button to check for any additional errors, and scan the area for any additional adjustments. I often use the satellite layer for this purpose. Finally, I upload the data!

I estimate that for each time I did this, I uploaded at 1000 buildings, since each geojson files contains around a thousand buildings. It is frankly not as quick as you might think, given that I often further split the geojson file to a half and work on 500 buildings at a time for hours. A rough calculation that if I work though 1 geojson file per day, I will finish an alcaldia every 2 months, and finish mapping CDMX in 2 years, ignoring adding addresses or additional verification checks. Mexico City is huge (;𖦹ㅁ𖦹).

Step 4

The next step is to redraw building geometries to match with the satellite data and doing more verifications to check the OSM data is accurate. I then use this page (https://seduvi.cdmx.gob.mx/programas-delegacionales-de-desarrollo-urbano) to map actual landuse data (residential, commercial, retail, etc.) following the example of Quixlian in the Zona Rosa/Colonia Juaréz. As for adding addresses, currently I only use Maxillary and street view images, but I think that in the future we could merge the cadastre data with the building geometries for addresses, but I don’t know how to do that yet and I’m not sure whether the city government did a good job or not.

And here you go, a “short” guide on importing data! Time to write the actual import documentation.

Viva México ٩(^ᗜ^ )و ´-