Questions & Answers - Measuring OpenStreetMap building footprint completeness using human settlement layers

Disclaimer: This is an archived version of the once interactive session pad with content from conference attendees. Please note that some information was lost during the transformation (i.e. the edit history, user colours, the chat and some formatting).

Back to session page

https://gisco-services.ec.europa.eu/distribution/v2/countries/distribution/TZ-region-01m-4326-2020.geojson

DOI: https://doi.org/10.5281/zenodo.3923033 Slides: https://bit.ly/sotm2020-osm-completeness Code: https://github.com/thinkingmachines/osm-completeness

Science Mailing List: https://lists.openstreetmap.org/listinfo/science

(It is considered polite to sign your name to questions.)

Questions

  1. [DONE] How long were you able to process the whole data in the Philippines? -Feye
  2. [DONE] You assume that the Facebook settlement layer is accurately depicting where built-up areas are. Did you verify that assumption against some known data in the Philippines? – Frederik (this seems to somewhat conflict with question 4 - i was assuming that Facebook was AI’ing aerial imagery?)
  3. [DONE] You said that building data are required to “know where people live”. But if you were e.g. an aid organisation wanting to find out where people live, could you not also look at landuse=residential or highway=residential objects in OSM - since these are easier to add for mappers than tracing buildings? – Frederik +1
  4. [DONE] The settlement layer focuses on mapping population, i.e. populated areas. Some areas have buildings but only non-residential ones, with zero population. Is that an issue for your method? +1
  5. [DONE] Does your method consider the different population density? Some buildings may have 1 storey, some may be skyscrapers, so a pixel in the settlement layer with a certain value for the population may mean different things - the same population can fit in one building or lots of them, depending on the number of floors.
  6. How difficult is to run your method for all countries?
  7. [DONE] (if time allows) Are you interpolating over the settlement data or assigning pixel-to-area? have you tried interpolation?
  8. [DONE] How do you evaluate Frederik Ramm’s statement on volunteer vs. paid mapping, giving you’ve seen the talk yesterday. With sponsoring stopping, consequently the mapping stopping. - YaguraStation
  9. [DONE] Amazing work! Is it also possible to download the completeness datasets you generated for Phillipines and Madagascar? Would like to see it without going to run your wonderful code. Felix D. And me too please Janet
  10. WorldPop takes into consideration the real population data based on the census (+ projections) and then assigns the 100m pixel value across those admin polygons. Did you consider the same?
  11. [DONE] With the results for the Philippines, what are your recommendations for the local OSM community in terms of focus in mapping areas?

Comments

  • Things on Facebook are never free! They are only useable for those who are its customers and are tracked by it. Since OSM is a free project, Facebook is not a suitable tool for organizing things. Moreover, Facebook breaks the OSM rule and does not mention that its maps are from OSM.
    • It is released under cc by 4.0
  • This is absolutely amazing! This can help our communities identify priority areas as well! No building left behind!
  • Thanks for the talk. There’s Audio-Feedback during the Q&A from somewhere.
  • For those who are still interested in an intro to OSM and QGIS (like my lightning talk showed), see this free self-learning material: OpenSchoolMaps - www.openschoolmaps.org - Download PDFs  https://tinyurl.com/openschoolmaps-zip-download . – Stefan Keller
    • thank you! - indiebio
  • Authors are already aware of this, but for the benefit of the audience - another great dataset that can help in such analysis is the Global Human Settlement Layer (GHSL), produced by the European Commission - JRC: https://ghsl.jrc.ec.europa.eu – Marco Minghini