Hi all. Beginning in November of 2023, HOT received a number of requests to support an update of OSM’s building footprints in Gaza. The requests came from INGO, UN, and humanitarian GIS and data analysts.
Background: HOT’s default is to not support mapping in areas of conflict. However, we will reconsider our default stance if there is a direct request to do so from trusted humanitarian partners with accurate, reliable, up to date information, AND we feel comfortable with our assessment of ethically mapping in the area.
Gaza current crisis: For the last couple of months we have carefully considered our data principles of open, useful, inclusive, and ethical data. In light of the real need to better understand the humanitarian situation in Gaza by providing a comprehensive pre-conflict building footprints dataset, HOT will be working directly with close partners and HOT validators to coordinate a closed update through HOT’s Tasking Manager to update pre-conflict building footprints in Gaza. The data will allow for more accurate assessments of the status of critical infrastructure in the area.
At this point, shouldn’t it be a during-conflict dataset? Reports indicate that some buildings are no more, and if they’re completely gone the map should reflect that fact.
Hi Paul. Good question. We are mapping pre-event in line with previous disaster responses. Current status of buildings is not part of the current scope of this project, but there are ongoing discussions on the topic.
I don;t think this is a good idea. Since this “disaster” is not comparable to a flood or earthquake which happens rapidly but over quite some time. It’s understandable to do so in response to earth quakes,… but mapping now pre-event situations feels a bit odd.
OSM maps the current state of the world, not pre-event status. Sometimes we don’t know what the post-even status is and can only map pre-event, but whenever possible we should be using the most current sources.
I think this depends on what you map. We want to avoid edit wars, like changing “names” from one language to the other depending on current front lines, or changing boundaries… Adding buildings should not be problematic from my point of view.
OSM maps world as it exists now, even if current state is outcome of unwanted/sad/bad events.
(if there is no up-to date imagery accessible mapping from outdated one is justifiable, but this sounds like it is deliberate decision to avoid mapping current state including cases where it is known that situation changed)
I am surprised that people are surprised. Areas under conflict and possibly destroyed (but not certain to be destroyed) pre-disaster infrastructure have been mapped by HOT for ages. There were activations for the Beirut Blast, multiple different activations for earthquakes in Turkey, and the Tigray War (this might have been managed by the local community, with the project hosted on their TM) are just a few I can name off the top of my head. These were all well publicized and had community support.
Well, there is a difference in mapping possibly destroyed buildings (due to no better data source being available) and mapping buildings known to be destroyed as existing.
And no matter why it was not spotted before and whether similar mistake was made before: knowingly mapping not existing building as existing now is not helpful. The same goes for setting up project to do this.
Hi @PierZen, agree with you on avoiding mapping in conflict scenarios as a general rule, but this is the result of requests and thorough discussions with humanitarian iNGO and UN staff augmented by HOT’s own analysis.
Based on extensive conversation directly with multiple agencies looking to use OSM data for humanitarian use, the decision to map pre-conflict buildings in Gaza is directly related to the intended use case of the data, combined with what is actually safe (for both mappers, and the population in Gaza) and feasible to do based on available data and imagery sources:
Publicly available imagery available for OSM mapping only captures the situation in Gaza before October 2023. There is no publicly available post-event imagery with high enough resolution to digitize buildings post October 2023.
A complete baseline dataset is needed to identify not just damaged infrastructure, but also identify potentially functional infrastructure for humanitarian response once that becomes operationally feasible, and eventually to support (early) recovery.
The current buildings in Gaza in OSM are pre-conflict, and where it is complete, is the highest quality building data available. Because the existing OSM data is good quality, OSM buildings have been heavily used, but analysts have had to combine with other less complete data (mostly AI-driven) to fill the gaps, which is causing challenges to confidence levels and in methodological comparisons. A single baseline dataset is strongly needed.
Pre-conflict buildings are important to accurately map damage, and to be able to update the map to an up to date representation over the next year. As the most effective systems to map damage remotely identify damage in an area (not by building). it is important to match a damaged area with what buildings were once in that area for programmatic planning, resource allocation, etc.
After consultations with multiple data requestors, HOT feels there is an opportunity for OSM to fill this data need. While this is a very fluid situation, we agree with the requesting organizations that it makes sense to have a complete baseline to understand the changes in the ongoing situation better, and therefore inform more accurate mapping in the future.
If so, then I think there should be no mapping activity carried out, as most likely outdated data is going to be added to OSM. At least based on media pictures in recent month. Most likely it’s known that these bulidings not existing any more and base principle of OSM Ground Truth is violated in such action.
Did you also consider OpenHistoricalMap here? What you are suggesting (mapping things that used to exist at a known point in time, but may not longer do so) sounds like the sort of thing that they may be interested in.
If you did consider OHM, but rejected it, what were the reasons? If it was lack of rendering, integrated tools etc. that might be something that the OSM and OHM communities could help with.
(for completeness I’m writing this just as an OSM and historical OHM contributor, and as someone who is somewhat familiar with some of the tools around OSM)
I’m not directly involved here, but as a conjecture – OSM data is distributed in so many ways that consumption by humanitarian actors is well established. OHM data is not hooked into these data flows.
People are mapping each bombshell location in Gaza—charting the location and buildings leveled. Would HOT add those back?
This (bombshell) level of mapping is does lead to editing conflicts in other parts of the world where there are active military conflicts.
During these proposed Gaza mapping efforts I’d like to see a statement/position on how HOT would deal with editing conflicts from the local community or the wider community especially where pre-conflict building are no longer extant.
Hi @dieterdreist this is a good point and we are looking at what has been removed from OSM in Gaza since October to make sure that we don’t undo work that is ground truthed.
When we do mapping of this nature, by the way, we would intentionally not remove any buildings from OSM, but actually tag them as damaged (or damaged / destroyed). Which is why we would like to update the map from pre-event imagery available (I’ll provide more detail on this in another part of the thread where it is referenced).
If there are people mapping destroyed buildings by removing them, we can reach out and try and align on a process.