Talk:Humanitarian OSM Tags
Satellite image - Interpretation ground control
We need some ground truth/link/images to asses image interpretation (like Map Features), especially for collapse building. I suspect that many POI tag as building=collapse are actually building abandonned long before the earthquake.
- In the rural areas I've also seen many cases where I, or others, have been mislead to see a walking track (or even marginally wide enough to be a highway=track, even if only for light 4x4's), but which only further down the trail have turned out to be a dry stream bed. Guess such can be good for walking, but it's something to watch out for. Alv 08:02, 18 January 2010 (UTC)
I agree - there needs to be a way to discriminate what's been tagged via aerial photos vs. what's been ground truthed -DruidSmith 00:32, 19 January 2010 (UTC)
- That should be recorded in the source information, ideally. --Ceyockey 02:54, 19 January 2010 (UTC)
One thing we are missing is a before-quake set of imagery with high resolution to compare to the voluminous after-quake imagery. This would help greatly in the assessment of whether man-made structures have collapsed due to the quake or not. --Ceyockey 02:56, 19 January 2010 (UTC)
- And here it might be ... Talk:WikiProject Haiti#Before and After Quake --Ceyockey 02:58, 19 January 2010 (UTC)
Is it possible/necessary to settle tags for risked zones : mine fields, battle fields ? FrViPofm 23:52, 22 January 2010 (UTC)
earthquake:damage vs damage:earthquake?
Will each of the catastrophe categories (e.g. fire, flood, volcano) be getting their own tagging namespace for use with :damage? Is categorizing damage causes useful? Is the earthquake namespace used with anything other than damage? --goldfndr 16:27, 30 January 2010 (UTC)
+1 for the syntax ! --FrViPofm 22:00, 30 January 2010 (UTC)
- damage:earthquake is better. So we could map any other cause of damage (e.g. damage:tornado, damage:flood). Or simply mark "damage" if a cause of the damage is unknown or does not matter. --Surly 04:54, 4 February 2010 (UTC)
Thoughts on the scope of disaster mapping. osmapb1 16:28, 28 January 2010 (UTC)
Will tidy up when I learn wiki-ness. osmapb1
Start thinking about all of this now. Most will already be in use by humanitarian/relief/emergency services/military so adopt their best practices if that does not conflict with osm database requirements/best practices.
Disaster - Biological, Chemical, Disease, Drought, Earthquake, Explosion, Fire, Flood, Famine, Nuclear, Tsunami, Volcano, Weather (to cover hurricane, cyclone, typhoon, tropical storm, storm)
Does the UN already have codes for these?
Data confirmation/quality - unknown, reported, confirmed.
Data source - id and date/time.
Functional - 0=fully, 1=partially, 2=no. - Does the UN already have codes for these? Can be used for anything with a function from radio station to water well to bridge.
Damage - current agencies already have codes for this, use them if poss.
Repaired - 0=fully, 1=partially, 2=no. With start_date/time etc.
Passability of ways - 0=yes, 1=obstructed, 2=no.
Timespan - use temporary=yes with start_date, end_date, start_time, end_time to indicate time. Can apply to any kind of disaster, any structure, any POI, way or area.
Agency - civil, military, NGO.
NGO - three letter acronym plus numeric eg WFP, MSF. Numeric identifiers nnn. Does the UN already have codes for these? See "Managing_authority=" below.
Start using UN Placecodes everywhere they exist from now on, starting with areas most likely to experience major disasters in the near future. When and who issues these? In advance of disaster or at the time?
Casualties - 0=none, 1=suspected, 2=yes Casualty status triage - already done by emergency services, copy their codes/practice.
Utility supply - 0=yes, 1=partial, 2=no. Utilities - electricity, potable water, non-potable water, sewage, gas.
Communications supply - telephone landline, mobile/cell, radio telephone, dial-up, broadband
Fuel availability - petrol, diesel, jet, wood, coal, etc
Emergency accommodation - type, temporary, permanent, start and end dates, controlling organisation, camp facilities
Develop specific styles of map presentation (CloudMade where are you?) which highlight particular map features eg power lines, pumping stations, hospitals, petrol stations, emergency camps, police stations etc Produce these as ready-made images eg on Flickr for disaster prone areas as they are mapped.
Hazard areas eg land mines, sinkholes, military ranges, flash flood/lahar/pyroclastic flow, landslide, avalanche
Compile a wiki catalogue of aerial/satellite imagery which can be used by osm now for tracing on osm regardless of emergency, and imagery which would be useful for reference in the event of an emergency.
Political influences - node, way, area and POI controller, contact details, travel restrictions, roadblocks etc.
Disaster related websites, urls - hu:website=*, un:website=* etc.
Good thoughts, osmapb1 - I think one key goal to come out of this will also be to develop a means of semantic interoperability. There is a group working on this, please add your name to this list: http://spreadsheets.google.com/ccc?key=0Agx_eCvNHqGCdHpaZkhFaUtrVnJodEdPc2p6c3M3Z0E if you are interested in participating in calls and further discussion... -DruidSmith 02:01, 29 January 2010 (UTC)
Thoughts on disaster mapping data
osmapb1 22:03, 31 January 2010 (UTC)
From a generalist, not a cartographer and not an NGO person, who has nevertheless much experience with computing, travel, geography and paramedic work in the real commercial world. I specialise in asking the stupid question, to make sure it gets answered, stating the blindingly obvious to make sure it gets stated, being the Devil's Advocate and seeing clearly despite the fog.
More thoughts, here so they can be read by whoever is interested.
Some general ideas for emergency situations :
KISS - keep it simple, assume all those involved are excellent at other things but 'amateurs' in dealing with disaster. A xxxxx_xxx tag is great but who's going to remember or type xxxxx_nn_xxx_xxxxxxxxx_xxxnn_nn_xxxxxxxx successfully, reliably or voluntarily? Not me for one.
Assume urgency but don't stop thinking just because of urgency.
Assess first, then treat.
Beware of excellence - who cares if the system is perfect if it is unusable?
Absence of evidence is not evidence of absence, better no data than bad data believed to be good.
The complexity of the work done so far is to be expected. (The UNSDI-T 2.0 link is a dud) so is there some normalisation of this already possible? eg.
"quality" is an attribute applicable to almost all osm input from whatever source. Is it confirmed, reported, suspect or unknown?
"permanence" is an attribute applicable to almost all osm input from whatever source. Is a piece of data subject to time constraints? Is it permanent or temporary? If temporary is it of fixed duration or unknown duration? And is its end timestamp, if present, in the past?
"validity" - invalid if it has an end timestamp in the past, otherwise valid.
"functionality" is an attribute applicable to (almost?) all physical osm things. To what degree is this thing still performing the function for which it was created? Yes, yes to some arbitrary but not complete degree, no, unknown. Can a house still be lived in, will a bridge still take traffic, can a power line carry its intended electricity, is a piece of road clear, partially blocked, completely blocked or in an unknown state, is a clinic operating, partially operating or out of action?
"language" - can a value be a code mapping to English, mapping to French mapping to Creole etc? Using codes can instantly overcome language difficulties if the next event takes place in Armenia when you had bet it would be in Indonesia. eg : 0=normal/operational/undamaged/fully functional, 1=the opposite of zero, 99=unknown, all other codes arbitrary but (pre)defined.
"owner" - does a piece of data have an owner in the sense of an entity which controls it? Yes, it is an asset belonging to a private person, a council, a company, the military, an NGO; or no, it's a route or relationship of some kind; or unknown/unknowable/indeterminate/(disputed?) For emergency operations this could serve very well as a highest level qualifier eg hu:amenity=*, un:amenity=*, mi:amenity=* for military and no prefix(amenity=*) for everything else.
"contact" - does a piece of data which has an owner/controller have some details which can be used to discuss the data with the most interested party(ies)?
"access" - does this item of data have some sort of restriction on who, why or when it can be accessed/used/travelled over/moved/reported on/quoted/political sensitivity?
Something like this streamlining can be used to cut the complexity, number and ambiguity of tags. Instead of amenity=hospital, amenity=temporary_hospital, amenity=temporary_military_hospital etc it can be boiled down to amenity=hospital and then have its attributes hanging beneath it as appropriate.
What is the current status of these tags? Is there active work?
Sean Horgan 20:57, 20 September 2010 (BST)