UNGIWG, meat on the bones. OpenStreetMap and the UNSDI.

There is very much an emerging just $*!(% do it attitude. SDI East Africa has gone ahead within a specific geography and user base, started setting up the infrastructure of GeoNetwork and GeoServer and just put it into action; there are wins and there are glaring examples of why SDI is needed. Andrew Turner and I had the pleasure of meeting many of these folks in Nairobi, including a session with the Somalia Interagency Mapping and Coordination group (their walls were plastered with real pirate maps). Elsewhere in the UN, UNOSAT and GeoNetwork have supported GeoRSS for a while now .. using the simplest thing that works and not reinventing the wheel. There’s many encouraging signs.

The remarkable thing about the opening panel at UNGIWG 9, “SDI vs. Neogeography” was how everyone dropped the versus. Everyone in the room uses the web, uses google maps, facebook, posts cat photos. They get that web and bottom up standards work for easy interoperability, get the user focus, and some “excessive creativity” like maps of UFO sightings is a good thing. And the neogeographers also get that formal SDI and GIS have requirements of precision, complexity, and cartographic elegance that aren’t always met. So the two approaches are moving to adapt and hybridize and meet halfway. For forward thinkers in either camp, there isn’t an antagonism here, only an opportunity to improve things for everyone. Earlier this year, Chris Holmes gave a great presentation on this theme, the GeoWeb and SDI.

Nicolas Chevant and I presented the vision going forward for the UNSDI-t and how our work with OpenStreetMap supports that effort with real tangible results today. Nicolas has developed a widely accepted schema for the transport layer of the UNSDI, no small task when every agency had a different method of recording road data. The neat thing is that this top down process generated schema closely resembles the key/value tags in OpenStreetMap. Perhaps not so suprising since we’re talking about the same underlying reality, but instructive that two very different processes can get similar results .. the informal, bottom up crowd does work. And OSM can learn from the formal schema. In the UNSDI-t, the emphasis is on the practical passability of roads, rather than official classification. That’s important in many places that the UN operates, and in many active OSM communities. With OSM’s open tagging system, it’s easy to build equivalencies to other schemas and retain the original attiributes as well. So OSM has been able to bring in UN road data, and other thematic layers should pose no problem.

OSM has imported many data sets (AND, TIGER, GeoBase soon..), as well as UN data sets. Neil Penman, Brett Henderson, and Ian Checkley of IBM volunteered to start with the UNJLC data for Sudan, built shapefile imports, and tag equivalencies. Issues of data quality and provenance reared early, in an emergency, you scramble to grab any available data source you can find, copyright or not, but that won’t fly for OSM. The larger looming issue is data synchronization. OSM is a single node in a wide UNSDI, and the question is how do these multiple updating databases provide the opportunity for choice of best dataset and synchronization. It’s a huge issue generally for the GeoWeb, and OSM API v0.6 provides a little bit of the scaffolding OSM will need to accomodate multi-master sync. Overall Sudan proved the OSM and UN were entirely compatible mapping partners and set the stage for more.

After Cyclone Nargis, I met Christopher Tun at WhereCamp, and quickly got working to provide support for web mapping infrastructure in the relief effort. Unlike most every other disaster, the UN was not free to operate on the ground in Myanmar. They UNJLC did what they could with existing data and satellite imagery, and in a crucial change, the maps themselves started asking for contribution. Fortunately through Chris, we were connected to local engineers who could operate the server and survey the disaster zone. The team from IBM joined in again, and started working directly with Sahana, to deploy on a single server inside Myanmar, where it would be more accessible to locals. We set up a full OSM server and mapnik tile server running locally inside Myanmar, along with osmosis to help facilitate diffs and synchronization. Sahana integrated their excellent excellent GIS module fully with OpenStreetMap. We had great remote training sessions with the Myanmar team, in Bangkok, via the surprisingly useful webex, and there were plans for us to visit directly. However things have gone quiet since the flury of activity, and I only hope it was of some enduring use in the relief effort. In any case, there was good technical development and increased capacity on the tech support side of things.

Our last interaction was the response to Hurricane Hannah in Haiti. This time we partnered with CartONG, and specifically focused on getting a routing service set up with OpenRouteService. Chippy provided great assistance with Shapefile wrangling and presence in Geneva for follow up meetings. The OpenRouteService experimented with a simple OpenLayers interface for users to roughly mark blocked roads, and adjust the routing based on this quick feedback. The only downside was that our best data source was provided in a not entirely open form. We couldn’t import it directly in OSM, only into a special instance of ORS, and so we don’t have something of last use .. open data is always where we want to be.

All three of these emergencies provided opportunity for OSM and the UN to work closer together. My effort now is to establish links beyond the emergencies, in preperation, in the normal course of development work. The nascent idea is to set up OSM Corp, a group of volunteers working to build infrastructure, relationships, and processes for using OpenStreetMap in humanitarian work. The next time there’s an opportunity to respond, we can get rolling rapidly, and maybe even work towards deployment in the field with MapAction or CartONG. And wider, we’re open to collaborations with UNGIWG and its member agencies; this was my offer to the assembled…

Myself and others on the openstreetmap foundation board, and in the large global volunteer community, are keenly interested in the activities of UNGIWG and the development of the UNSDI. Data sharing among agencies, governments, and ngos for humanitarian purposes is totally complementary to the OSM mission, to create a free and open of the entire world. I whole heartedly offer our support and willingness to work with you all as an engaged and practical partner to build nodes in an SDI that works for the world.

1 Comment

  1. Stuart Thomson said,

    November 26, 2008 @ 1:57 am

    Hi Mikel,

    My name is Stuart and I work with World Vision Australia in our African team. I am very interested in seeing how the concept of open street mapping can work at a community level and have many different applications using simple technology such as mobile phones. I work mostly with West and East Africa and would be interested to see if you have any examples of your work in those countries. Otherwise examples from other developing countries would suffice. Do you ever pass by Australia? If you do it would be great to see if you could come here and talk with us. We have been trying to develop simple but effective approaches togehter with IBM. Would be great to see what you have to offer to.

    Kind regards,

RSS feed for comments on this post · TrackBack URI

Comments (1)