[HOT] OSM Damage Assessment
bgirardot at gmail.com
Fri Jul 3 13:26:25 UTC 2015
This is a somewhat complicated topic.
Just for clarity: Humanitarian OpenStreetMap Team (HOT) and
OpenStreetMap (OSM) are closely related, but separate organizations.
HOT would be the group actually doing any damage assessments. We
probably would be using OpenStreetMap (OSM) to store that damage
assessment data or we might find another data store for that
information, but for Nepal, the mapping related to damage we did, we
stored in OSM.
But HOT and OSM are separate organizations, even though we (HOT) rely on
OSM to do our work and follow OSM conventions and guidelines for all the
mapping we do.
So talking about OSM damage assessments would not be the best way to
refer to them.
They would be HOT damage assessments stored in OSM. I only mention this
very subtle distinction because as a researcher I want you to have the
most accurate understanding possible and people often confuse the
relationship between HOT and OSM.
We did not do detailed damage assessments for Nepal. What we did map
were areas that appear to have been completely destroyed by the earthquake.
This was at a very macro scale, so the small settlement, neighborhood or
We did not do any specific damage grading or building level analysis.
Basically we compared pre- and post-event imagery and tried to determine
if settlements and buildings that appeared in the pre-event imagery were
still there in the post-event imagery.
If it looked like buildings have been totally destroyed and no longer
appeared as buildings in the post-event imagery, we drew a polygon
around that area and tagged it as a "landuse=brownfield" and
You can read the exact methodology we used in one of the Projects we
created to do the mapping in the "Instructions" tab of this project:
All of the data we have generated should be in the 30 mins updates and
you can find those polys by looking for the two key=value pairs I
Another way to get just the areas marked as I described is to use the
overpass-turbo.eu export tool.
Use the query wizard to generate a query like this:
landuse=brownfield and "damage:event"="nepal_earthquake_2015" in nepal
It is a lot of data for a web browser to render so expect your browser
to be non responsive at times. But you can then use the "Export" button
to get the data in several formats.
Here is a link to that query: http://overpass-turbo.eu/s/afm
I would very roughly estimate that we have completed about 50% of the
areas we have targeted to do those assessments on.
As to the usability of the data generated, I think it served its
intended purpose very well, which was to give a rough idea of where
damage is likely to have occurred when very little other information was
available. I think it might be used in the extended recovery phase as
well, but I am less familiar with how it will be used in that phase.
We are in the process of working with our partners on the ground to
determine if we need to continue the process of identifying these areas
and what the priority is.
Any feedback, correlations, or on the ground evaluations of the accuracy
of damage related data we generated would be most welcome and we will do
what we can to help you do any evaluations of our assessments.
Thank you very much for writing us, please let me know if you have any
more questions or we can help you with your investigations any further.
On 7/2/2015 7:55 PM, Michal Bodnár wrote:
> Hello HOT community,
> I am currently doing a research where I am validating and comparing
> different types of damage assessments of Nepal EQ - UNOSAT/Copernicus,
> Tomnod, Ushahidi (a special case) and I would like OSM damage assessment
> to be researched as well.
> I would have therefore some questions:
> 1. What everything is being mapped in terms of damage assessment? I have
> noticed landslides and damaged buildings.
> 2. In case of buildings, is there any classification of the level of
> damage being made (e.g. EMS-98 scale)?
> 3. How is the progress of the damage assessment so far? Do you think
> that the current status of DA is already enough "good" to be used?
> 4. How can I download all the damage assessment data in one package? I
> noticed the links for 30 minutes updates of the data, but that does not
> (as far as I know) include DA data.
> I would like to thank you in advance for any answer.
> Wish you all a great day,
> Michal Bodnár, M.Eng.
> PhD Researcher at Beihang University/NDRCC
> China +8613031164554| Czech republic +420 607957528
> Find me on: LinkedIn
> <http://cn.linkedin.com/pub/michal-bodn%C3%A1r/42/aa/24a> Facebook
> <https://www.facebook.com/michal.bodnar> Twitter
> Alumni at CTU - Czech Technical University, Prague www.cvut.cz
> <http://www.cvut.cz/>Geomatics @CTU <http://geomatics.fsv.cvut.cz/en/>
> Alumni at BEST (Board of European Students of Technology)
> www.best.eu.org <http://www.best.eu.org>
> HOT mailing list
> HOT at openstreetmap.org
More information about the HOT