[HOT] PHL: making use of building damage data

[Banick, Robert]

Hi All,

I'm gearing up to do a considerable amount of fieldwork validating and
adding to the damage assessments conducted for Typhoon Haiyan to date.
I'll be working with a team of enumerators conducting standard shelter
assessments to incorporate relevant data from those assessments into OSM.
As such I've been taking a hard look at the OSM data model for damages and
thinking about how best this can be done.

My comments are those of Nick McWilliam from MapAction: if buildings and
damages aren't separated then it's very difficult to tell the two apart.
While semicolons *can* be used, its more work than necessary and
needlessly complicated for the many new or occasional tracers we get
during these events.

I'm developing a schema that we can use to pair Shelter Cluster assessment
data with OSM. The goal is a data model that  matches with what responders
are using and is easy to query, update and clean as recovery work
progresses.

There's been some great discussion here so far that I'm trying to pull
together into something we can use. I wanted to propose the following four
tags and get feedback. Note that this scheme is meant as much for field
teams as for remote tracers.

damage=minor/limited/moderate/extensive/destroyed
disaster=typhoon;Yolanda
validated=yes
assessment=XYZ_Organization

Thoughts on the scheme:

I would propose that guidelines/example for remote tracers using these
tags be established (e.g. X photo is an example of moderate damage) and
only a limited number are used by them.

The damages tag was modified to reflect the 5-level scale I've seen in use
by shelter assessors  in previous disasters in the Philippines. I welcome
other input.

I decided against having a typhoon-specific damages tag because
realistically multiple sources of damage don't matter and are hard to
disentangle besides.

Validation should be used by field teams to mark that they've verified or
modified remotely provided data. The assessment tag should mark which
organizations did that validation.

I think the disaster tag could be done better and would welcome thoughts
there. I can theoretically see multiple disasters needing to be recored
but can't think of a more elegant way to do so.

Any and all thoughts are most appreciated.

Best,
Robert

Robert Banick | Field GIS Coordinator | International Services | Ì
American Red Cross <http://www.redcross.org/>
2025 E Street NW, Washington, DC 20006





On 11/22/13 11:30 AM, "Nick McWilliam (MapAction)"
<nmcwilliam at mapaction.org> wrote:

>Hi Andrew -
>Much appreciate the quick response - will check the tag attribute; and
>understand your comments about the design of the tagging scheme.
>Best wishes -
>Nick.
>
>-----Original Message-----
>From: Andrew Buck [mailto:andrew.r.buck at gmail.com]
>Sent: 22 November 2013 16:15
>To: hot at openstreetmap.org
>Subject: Re: [HOT] PHL: making use of building damage data
>
>-----BEGIN PGP SIGNED MESSAGE-----
>Hash: SHA1
>
>Nick, see my responses inline below:
>
>> 1. Having referred to the OSM Damaged buildings crisis mapping
>> <http://wiki.openstreetmap.org/wiki/Damaged_buildings_crisis_mapping>
>> page, I'm not sure I'm making the best use of the data that's there,
>> e.g. how to access typhoon:reviewed and typhoon:damage ?
>> Would we be better not using shapefile downloads but another format
>> that preserves more of the source data structure?
>
>Have a look at the last column in the shapefiles which should be called
>'tags'.  In that you will find all the other tags that are not part of the
>other columns so you can query against that for
>typhoon:reviewed=* and so forth.
>
>> 2. Perhaps it's just the way that the shapefile attributes are derived
>> from the source OSM data, but how could we separate the building type
>> (house, office etc) from the damage status?
>
>Using the current tagging shceme we really cant since a damaged church for
>example would need building=damaged and building=church at the same time
>which isn't allowed.  This is a known problem with this scheme and it will
>likely be fixed after the public is mostly done mapping, we will just go
>through and change the tags accordingly to the new schema.
>
>> 3. Finally (and apologies for touching on what I expect is a
>> long-running question), how are such data time-stamped? I'm thinking
>> for example of buildings that were already tagged as damaged
>> pre-typhoon, or that become repaired.
>
>We have the time stamps for when the objects were added to the DB and/or
>tagged as damaged so in theory you can work out what disaster everything
>relates to.  In practice though we also plan to work out a better scheme
>for
>tagging this explicitely in the coming weeks/months after the disaster
>response has settled down a bit.  Right now we are just running with the
>current scheme since a lot of people are already using it, we don't want
>to
>pull the rug out from under them during the response, even if we know our
>current schema is sub-optimal and will be changed in the future.
>
>Hope this addresses your questions.
>
>- -AndrewBuck
>
>
>_______________________________________________
>HOT mailing list
>HOT at openstreetmap.org
>https://lists.openstreetmap.org/listinfo/hot
>