[Talk-ug] Import of Buildings in Districts Bordering DRC

Tue Jul 14 06:29:06 UTC 2020

Hello Bert,

Thanks for your insightful feedback and comments about the process.

   1.  Regarding the offsets, you are right, the team will be updating that
   section of the process so that mappers rely both on the buildings and road
   features when making the offset to align features.
   2. For the imagery, the team will be using a purchased updated maxar
   Imagery. Maxar imagery is now back on JOSM and ID editor (premium and
   standard). The imagery purchased covers the bordering districts in Uganda
   as well as DRC and is of a 2019 date. The import team will also be trained
   to use buildings under construction and other features to determine the
   most recent imagery. From our initial assessment, the imagery purchased is
   newer than inbuilt maxar and bing imageries and clearer than the Esri World
   Imagery.
   3. Replacing geometry is definitely a tricky situation, the team plans
   to use the conflation tool mainly for the buildings be able to replace
   geometry and reduce on the number of deletions in the process. Through this
   process, the tags already existing on a building feature will be preserved
   and only new tags will be added. The import team will also be trained to be
   able to align prereferral features like gates and boundary walls but I do
   agree that will be a sticky situation that I will discuss with the team to
   see how we can navigate it smoothly
   4.  Yes, we are counterparts in DRC are in touch with the UN Mappers
   that are currently mapping in the areas of Ituri province. The discussions
   on ways how we could combine forces as well as share responsibilities are
   still ongoing but that's for highlighting this situation.

Thanks again Bert for the good comments I will be discussing to my team to
make some of the edits you sighted as well as acquire more concrete
information on our collaboration with the UN mappers and revert to you with
more information.

Best
Deogratius

*Humanitarian OpenStreetMap Team
*Project Manager, Uganda

Email: kigguddedeogratias at hotosm.org
Skype: deogratius kiggudde
Twitter: @KigguddeD
UG Mobile: +256 703 143 225

http://donate.hotosm.org/

*Using OpenStreetMap for Humanitarian Response & Economic Development*
web <http://hotosm.org/> | twitter <https://twitter.com/hotosm> |
facebook <https://www.facebook.com/hotosm> | donate
<https://donate.hotosm.org/>

On Thu, Jul 9, 2020 at 10:08 PM Bert -Araali- Van Opstal <
bert.araali.afritastic at gmail.com> wrote:

> Hello Deogratius,
>
> First of all, great project, really looking forward to get going.
>
> I went through both wiki pages, and overall very solid and detailed
> approach.
> I do have some concerns, points on which improvements could be made.
>
> 1) Offsets - in many previous HOT projects offsets were provided by the
> project manager, which are easy to store and implement in JOSM. Until
> November 2019 we could use the MAXAR imagery in JOSM (still blocked ? only
> Maxar imagery in ID), and I used it a lot.  My experience was that it has
> significant offsets in comparison with other imagery layers, especially on
> hill sides you see a lot of distortion.  This can even be seen in the
> offset of roofs in regard to the ground level.  An important issue in
> correct alignment of the imagery.
> So when you state in the procedure that we should align the layers by
> looking at the buildings this will not be correct.  On a tile by tile basis
> you should align with features on ground level, most convenient are roads.
> Another way could be we have GSM tracks or reference points from the
> field.  The project manager could make offsets available for every tile
> using these field datasets or you could engage some people in the field to
> get them.  Because even many of the existing buildings could be misaligned.
>
> 2)  How to determine which data is the most recent.  The Maxar imagery is
> dated. Other imagery layers on which existing buildings are mapped is
> mostly not.  SO there is no clear reference of what is most recent.
> Comments and descriptions in changesets can provide some information, as
> many existing mapping is done based on field knowledge. So some improvement
> and more details could be provided in the procedure on how to determine
> which is the most recent and accurate data by referring and evaluating data
> in the changesets.
> Same applies to building relations when 3D mapping was done.  You should
> provide some details there that in these cases only the building outline
> should be replaced and other building:parts should be updated.  There are
> not many but some are there because I added them.  This is a very
> complicated task.
>
> 3) Replacing building geometry.  Will be a much more extensive exercise as
> described.  Many areas are far more detailed then just the buildings.  As
> you probably know many have the plots aligned with landuse tags, other
> features as fences and gates are present at many locations.  If you replace
> the geometry with the new buildings all these "attached" features, some
> even sharing nodes with the building, must be updated.  I have no
> experience with geometry replacement so how will this work ?  If a
> misalignment is seen all the features need to be aligned as a group.
>
> 4) Currently there are two UN mapping teams doing a lot of work in these
> areas, in support of MONUSCO troops. One focuses on roads and buildings,
> the other team on landuse.  How do we avoid conflicts with those projects
> or get them involved ? They contacted me several times so they are open for
> cooperation.
>
>
> Greetings and success with the project.
>
> Bert Araali
> On 02/07/2020 17:46, Deogratias Kiggudde wrote:
>
> Hello community,
> The subject of this message is to inform the OSM community in Uganda of an import prepared by HOT [1] as part of a project to scale up the Missing Maps initiative, combining data extracted in a way automated and information collected in the field [2].
> In the Uganda , this import consists of more than 2 million footprints of buildings located in the districts of Kisoro, Kanungu, Rubirizi, Kasese, Bundibugyo, Ntoroko, Bulisa Hoima, Kikuube, Nebbi, Zombo, Arua, Koboko, Maracha, Kiryandongo, Kamwenge, Kyegegwa and Insigiro. This data was extracted from high-resolution imagery provided by Maxar over the period 2015 to 2019.
>
> A wiki page with details of the proposed import is available here:https://wiki.openstreetmap.org/wiki/FR:Import_B%C3%A2timents_Maxar-Ecopia_Est_RDC_%26_Ouest_Uganda <https://wiki.openstreetmap.org/wiki/FR:Import_B%C3%A2timents_Maxar-Ecopia_Est_RDC_%26_Ouest_Ouganda>
> A second page contains the practical details of the proposed methodology
> to perform this import:https://wiki.openstreetmap.org/wiki/FR:Import_B%C3%A2timents_Maxar-Ecopia_Est_RDC_%26_Ouest_Ouganda/D%C3%A9roulez <https://wiki.openstreetmap.org/wiki/FR:Import_B%C3%A2timents_Maxar-Ecopia_Est_RDC_%26_Ouest_Ouganda/D%C3%A9roulement>
> Experienced OSM contributors who would be interested in participating in
> this import will be welcome.
> A similar message regarding this import will also be shared on the OSM mailing  list of DRC and the HOT mailing list to gather a maximum
> feedback.
> Don't hesitate to share your comments, questions, or problems
> related to this import before it begins.
> Best
>
> Deogratius
>
> [1] https://wiki.openstreetmap.org/wiki/Humanitarian_OSM_Team
> [2]https://www.hotosm.org/projects/leveraging-ml-generated-building-data-and-motorcycle-mapping-in-the-drc-and-uganda-to-enhance-humanitarian-response/
>
> *Humanitarian OpenStreetMap Team
> *Project Manager, Uganda
>
> Email: kigguddedeogratias at hotosm.org
> Skype: deogratius kiggudde
> Twitter: @KigguddeD
> UG Mobile: +256 703 143 225
>
>  http://donate.hotosm.org/
>
> *Using OpenStreetMap for Humanitarian Response & Economic Development*
> web <http://hotosm.org/> | twitter <https://twitter.com/hotosm> | facebook <https://www.facebook.com/hotosm> | donate <https://donate.hotosm.org/>
>
>
> _______________________________________________
> Talk-ug mailing listTalk-ug at openstreetmap.orghttps://lists.openstreetmap.org/listinfo/talk-ug
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstreetmap.org/pipermail/talk-ug/attachments/20200714/671dac9a/attachment-0001.htm>