[HOT] HOT Digest, Vol 50, Issue 5

Andrew Buck andrew.r.buck at gmail.com
Thu Apr 3 15:23:54 UTC 2014

Hash: SHA1

Hey, thanks for your email; very interesting to hear from someone who
has worked for this agency.  We make heavy use of the GNS data around
the world.  I will explain what we are doing in this particular case
in Guinea.

We are currently start by importing the GNS data into our database
first.  Next I go through an area covering 1 or 2 topo sheets and
create a blank place node with a fixme tag on it for every town that
appears on the topo maps, whether they have a name there or not
(except I skip creating a node for any that already are in OSM from
the previous GNS import for the area).  Then a task manager job is
created to have volunteers go over the area and type in any names that
appear on the topo maps (and also enter alternate spellings if the two
sources don't agree).  Finally, after all the name data is in the
database, we go back over the area one last time and move every place
node to its correct position using the false color landsat 8 imagery
(villages are bright pink dots so this is very easy to do).  During
this stage we often flip back and forth between the imagery and the
topo maps to determine what the local offset if due to inveteracies in
the original datasets and/or inaccuracy of my othro-rectification of
the scanned maps.

In regard to your second question about landscan and Afripop:
The Red Cross and MSF who are doing work with our data are aware of
afripop and may end up using it in conjunction with OSM data to do
their own analysis, but the format of the afripop data makes it not
particularly suited for import into OSM directly.

The same thing is basically true of the landscan data.  It is 1km
gridded raster data so it would be useful for background rendering on
our lower zoom levels, but again it can't really be used to improve
the OSM database directly.

Hope this answers your questions.  It would be good to speak with you
on Mumble if you are interested in helping with merging in this name
data.  We need experienced people to handle some of the more
complicated bits of the process.  Here is a link to instructions on
how to set up mumble if you want to talk about this further.


Thanks again,
- -AndrewBuck

On 04/02/2014 10:21 AM, John Gates wrote:
> re: Ebola Outbreak Guinea, West Africa -- Mapping to support
> Apologies from a HOT newbie, but as retired NGAer (National 
> Geospatial-Intelligence Agency--successor to NIMA, DMA & Army Map
> Service) had some basic observations/questions concerning base
> mapping:
> 1. GEOnet Names Server (GNS) is good standardized names source,
> but probably not more accurate or dense than the 1:250K JOGs that
> were compiled in the 1950's from 1930's base mapping. Are there
> other sources that the HOT relies on? Although I can't peruse NGA
> holdings anymore, the US Library of Congress here in Washington has
> a very good collection of maps from around the world. Unfortunately
> very few have been scanned and placed on-line. I can check holdings
> in the stacks if there is interest. They allow maps to be
> photographed or scanned but copyright must be respected.
> 2. Does the HOT use standardized raster population datasets like
> Landscan or Afripop to augment the vector mapping? It might be
> helpful in this case.
> Thanks,
> John Gates
Version: GnuPG v1.4.11 (GNU/Linux)
Comment: Using GnuPG with Thunderbird - http://www.enigmail.net/


More information about the HOT mailing list