[HOT] pre-earthquake imagery for Haiti

[Jaakko Helleranta.com]

Hej Sabrina,

I don't have imagery for you but if you're interested I am happy to share
you my thoughts on some of the quality issues related to VTC-based damage
assessment quality.

This is based on my experience living in Haiti since September 2010,
mapping it since right after the quake (as a more or less newbie
OpenStreetMapper, now the most active mapper _on_ the island), but perhaps
most importantly based on my experience with evaluating hundreds and
hundreds of building=collapsed, etc earth quake damage objects in the OSM
database over the time I've done mapping here -- and ending up deleting
practically all of them partly because they (without any additional
information such as image links, names of collapsed buildings or other
building / POI-in-the-building details) are no longer useful in OSM in
general but to good extent because of the problems in data.

Very clearly The Biggest single problem that I found in the building damage
assessment data in Haiti (which was partly traced to the db by
OSM-volunteers, mostly imported -- iirc) was that many who had done the
assessment had clearly never been to Haiti nor any other similar country.
.. I say this because I saw countless buildings that -- for someone who has
been here / a developing country -- were very clearly not collapsed but
marked as such simply because they were (and many still are) under
construction and missing the roof -- possibly only from the top floor. And
as in many developing countries this is completely normal and quite common.
The other problem of especially the imported collapsed buildings data for
keeping in the OSM longer was that since it was nodes only and very dense
at some points it was practically useless combined with the various
absolute and relational inaccuracies that exist in geo data (starting from
misalignments/offsets of various imagery, inaccuracies of various data
added over time, etc). So, I've nuked them from the OSM database. ... Which
also means that if you want to asses the damage assessment data that once
was in OSM you need to look at a data snapshot from, say, from fall of 2010.

Now, there's something to think about -- or perhaps look into a bit more
first -- for someone who's interested related to this.
I don't know if many people on this list have read/heard but there was a
study/evaluation done on the Feb 2010 quake damage perhaps a year ago or
so. In short it concluded that there could not have been more than about
80,000 casualties in the quake (in stead of the official number of closer
to 300,000). .. I haven't digged into this more but I know one of the
people who were part of conducting the study who's a solid guy and can ask
for the report / put anyone who might be interested in contact with him.
The study was obviously not accepted broadly (as the official numbers still
stand "unchallenged") and there are reasons not to flaunt it regardless of
how right the study would be. And some of those reasons might be good.
But for the sake of an academic discussion for this case/topic-at-hand I am
wondering if it _could_ be possible that _*should* the official casualty
numbers of the quake be bloated_:
What might the reason_s_ / enabling factor_s_ of that be?
.. And more so, based on experience with the issues with the quick building
damage assessment data:
Could the bad quality (and bloated) building damage assessment data have
given an impression or base for larger-than-true casualties?
Now, if I remember right the casualty numbers came out so fast in the
beginning that _it seems_ (to someone like me who doesn't know better) that
nobody had data to back it up and I think that not even the
new-imagery-based building damage assessment data was out then so (if this
was the case) one can draw one's own conclusions from that.

But in any case: it's a pretty clear fact to me that there were some real
issues with some of the building damage assessment data and that should
give food for thought in preparing for similar future cases. .. You might
want to look at / do some comparisons of different source of damage data.
.. And you should most certainly also (try to) compare your evaluations
with the official on-the-ground evaluations of MTPTC(/UNOPS).

Cheers from Haiti,
-Jaakko
http://osm.org/user/jaakkoh

--
jaakko at helleranta.com * Skype: jhelleranta * Mobile: +509-37-269154  *
http://go.hel.cc/MyProfile

On Mon, Jul 2, 2012 at 5:06 PM, Sabrina Grimsrud <
sabrina.grimsrud at geodata.no> wrote:

> Dear Humanitarian OSM team, ****
>
> ** **
>
> I am an MSC student writing my thesis about the quality of volunteered
> geographic information in disasters and using Haiti as an example. ****
>
> I would like to try to re-create the disaster mapping but I am having
> difficulties finding pre-earthquake imagery to compare to: there are a lot
> of links for imagery that can be used in one of the OSM editors and images
> for individual damaged buildings, but I haven’t found anything free that I
> can download and use in ArcGIS. I am looking for imagery of Port-au-Prince
> before the earthquake happened.****
>
> Do you have any suggestions? Thanks in advance for your help!****
>
> ** **
>
> Kind regards, ****
>
> * *
>
> *Sabrina Grimsrud*
> MSc GIMA student ****
>
> ** **
>
> _______________________________________________
> HOT mailing list
> HOT at openstreetmap.org
> http://lists.openstreetmap.org/listinfo/hot
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstreetmap.org/pipermail/hot/attachments/20120711/2c9dccd6/attachment.html>