NACIS 2013: Cartography for a Changing Map

Presented at NACIS 2013 in Greenville, SC.

Transcript

1. Cartography for a Changing Map AJ Ashton, MapBox
3. Global • All the data OpenStreetMap has to offer, and more • Multiscale: 20+ Web Mercator zoom levels • Multiple label language options
5. Scalable • Cloud servers that get added and dropped constantly can’t be too “heavy” • Tiles should load just as fast with 10 users as with 10,000 users • Apply different styles to the same underlying data
7. Up to date • Currently sync with OSM every 5 minutes • We want this to be even faster • The world is always changing
9. Global + Scaleable + Up to date A challenging combination
10. Global + Scaleable + Up to date + Make it look good
11. Keeping up with OSM
12. OSM Changes 1.5 million changes per day over 1000 changes per minute http://osmstats.altogetherlost.com
13. Most changes are small, but some have big consequences
16. Cartography for a changing map: Simple things
17. Generalize
18. Normalization Cleaning OSM data for consistency
19. Cartography for a changing map: Not-so simple things
20. aka. Automatic manual tweaks
21. City labels
22. City labels • Custom list of scaleranks (based on Natural Earth) and label direction hints • Fuzzy spatial & fuzzy text join function between incoming OSM data + our list
23. Airport labels/icons
24. Cartography for a changing map: Working with tiles
25. Tiles are small & static 256x256 pixels not a lot to work with
27. Tiles don’t know about other tiles
30. The terrifying chaos label problem
34. No perfect solutions Only mitigation strategies
35. Consistent ordering ORDER BY scalerank, population DESC, osm_id
36. De-duplicate at the data level GROUP BY ST_SnapToGrid(geom)