There isn't one map of the world.

Anyone who has worked with maps for a while knows this -- that the
process of making a map is something that is creative in form, and that
all of the pieces that go into it play a major role in what comes out.

However, I keep hearing people talk about OAM having a single map of the
world, so I think that it's important to realize why this doesn't work.

 - Almost every important use of OpenAerialMap involves some need for
   editorial control over what the resulting product from the map is.
   
   Humanitarian aid, for example, typically centers on disaster hit
   areas -- and in almost all cases, the most important thing for
   imagery is recency. Seeing imagery from before and earthquake, even
   at 5 times the resolution, doesn't solve the problems for disaster
   relief.

   On the other hand, an organization wanting to put an overview of an
   area on their website -- say, for example, the 2012 Olympics wanting
   to put together a high res flyover of their area -- are not going to
   want to have a blurry kite flown image uploaded over the top of their
   data and obliterating it.

   With every use case I can imagine for OAM, what is the 'best' image
   of an area for one user won't be the best for another. With that in
   mind, we should realize that attempting to present one map of the
   world which can meet the needs of users of OAM data is very unlikely.

 - Every piece of imagery tells a different story.

   OSM has the idea of "ground truth wins" -- that is, local signage
   verified by locals wins over anything else. With data, that's a
   reasonable approach. However, aerial imagery is not data; like the
   maps themselves, each aerial image is a creative work, prepared via a
   certain process and attempting to achieve certain goals. 

   One story I heard at FOSS4G was from someone working in Nome, Alaska.
   In 2010, Google had a great image of Nome -- cloud free, high
   resolution and crisp. In late 2010, it was temporarily replaced by a
   more recent image -- entirely of clouds.

   For other purposes, people will want winter imagery, summer imagery,
   leaf on, leaf off, high resolution with high contrast, mid-sized
   resolution with medium contrast, etc.

   Every image that goes into OAM tells a different story. Imagining
   that any one person can put these images together into a cohesive
   whole is ignoring the multivaried uses that OAM can have.

In addition to the more philisophical reasons that this is impractical,
we shouldn't ignore the technical limitations that have led to the
current design as a catalog:

 - Many of the uses of OAM data concentrate on small areas -- disaster
   recovery, environmental impact monitoring, property assessment.
   Recently, a local team flew balloon imagery of a landfill/incinerator
   site: http://mapmill.org/images/2011-10-10-saugus-landfill-a 

   This type of data is highly localized, and the needed output is also
   highly localized.

   Investing massive resources into hosting a worldwide tileset means
   that we have to dedicate fewer resources to meeting the needs of
   highly localized datasets; even though the map of the Saugus landfill
   will have almost no impact for the average user, for the people
   working on the project, it is far more important than presenting an
   accurate landsat image of Madagascar.

   Given the enormous challenges present in presenting a view of the
   world, and given the high likelihood of benefit being seen at a local
   level first, I would highly recommend that people start by thinking
   of OAM as "an Open View of the World Around You"; rather than trying
   to imagine solving the much more difficult problems of worldwide
   imagery, start by helping people build maps of their towns, or their
   houses.

 - Maintaining a cohesive worldwide cache designed to act as an 
   aerial/satellite map is a full time project. (When I say 'full time',
   I mean it -- I think that you would need at least one person working
   full time on it just to keep it functioning as intended, given the
   massive amounts of data that go into it and the potential risks that
   come out of it.) While I understand the base desire -- 'stop
   depending on non-open data' -- I think that given the limited
   developer resources available to OAM, and the technical limitations
   that such a project would have, it really is best for the community
   to try to focus first on meeting the needs of small scale projects.

   At one point Schuyler and I had hoped that we would be able to solve
   this problem trivially using remote data to create maps of the world,
   but having explored these options to some extent, I become more
   convinced every time I try that this is definitely a huge job -- one
   that can be built on top of OAM, but should not *become* OAM. OAM
   should concentrate on cataloging and making imagery discoverable for
   all users, rather than creating a single map of the world to help
   only some.

To that end, I would like to make it very clear: At this point in time,
I feel that OAM is not in a position where building a worldwide cache
designed to replace commercially available tilesets for base ortho
imagery is going to be the most likely thing to help the needs that OAM
can help to solve the most. When the next earthquake, flood, or tsunami
hits, I feel that it would be best for everyone who is participating in
this community to be able to really have tools built around "The World
Around you" -- and if we succeed massively in that regard, I think that
it would be easy to imagine scaling up. 

There are too many different views of the world to make them all work
equally well. Let's get it so that people can build *their* view of the
world arond them first, and then move on, rather than trying to come at
it the other way: anything other than that seems likely to leave us all
stuck in the mud, and no better off than we have been for the past 4
years.

To that end, I am actively putting my money where my mouth is: my hope
is to spend the next couple weeks building a service which allows users
to build a local view of the world around them, combining the imagery
available in OAM in the way that they see fit. (This work has been 
started in https://github.com/crschmidt/mosaic-webui ; it's far from
done, but progress is happening.) I'm also exploring the use of
MapMill[1] and MapKnitter[2], two great tools for helping build better
maps using highly local imagery.

Anyone interested in helping on OAM, I highly recommend you get engaged
by exploring the data and the world around you -- I think that it's the
best possible way to get a better understanding of what is available --
and how it can help.

[1] http://mapmill.org
[2] http://mapknitter.org/

Best Regards,
-- 
Christopher Schmidt
Web Developer

_______________________________________________
talk mailing list
[email protected]
http://host134.hostmonster.com/mailman/listinfo/talk_openaerialmap.org

Reply via email to