Humanitarian OSM Team: Haiti Strategy and Proposal

Who, what, where, and how are all open questions. Why is simple … OpenStreetMap has demonstrated incredible value in Haiti and we need to make sure we are prepared for the long run there, and for future disasters. A couple weeks ago, Nicolas and I started digesting the Haiti response, and years of thoughts and discussions in OSM, into something like a plan.

The aim of this evolving document is to start to gather and prioritize major themes of needs and activities, for HOT organizationally, in documentation and processes, in technical developments, in relationships. As we are building the vehicle while already hurtling down the road, parts of this are already in play, parts are falling through the gaps, and overall coordination is a need.

Your review and comments on any facet of this are most welcome. Ideas to make this happen especially!

Haiti OpenStreetMap Response

The have been at least 400 OpenStreetMap editing sessions in Haiti since the quake hit. Mostly tracing Yahoo imagery, and gleaning information from old CIA maps. We also just received permission to use GeoEye imagery acquired post-event … that will allow us to tag collapsed buildings. Many relief groups are deploying now, many checking in with the CrisisMappers list (the main locus of the wider humanitarian tech community), and they are making inquiries into OSM data and requests for particular features. Dozens of mappers and developers are lending a hand, coordinating on the OSM Haiti WikiProject and IRC and the OSM talk list … standing up services, including 5 minute extracts in Shapefile and Garmin formats, and maps with hill-shading. Just the start to relief and reconstruction effort we hope to contribute to.

Two images to show how we’ve progressed … the first OSM Port au Prince just now, the second OSM before the earthquake.



I’ll be on twitter with updates … though I’m due to fly tonight to Ireland.

MapAction uses OpenStreetMap for Philippines Response

MapAction has deployed to the Philippines to support the United Nations response to the Tropical Storm Ondoy disaster in the Philippines. They’re producing many map products, distributed through ReliefWeb and are using OpenStreetMap data collected by the incredible and resilient OpenStreetMap Philippines community.

I hope this makes a small difference to the work there, to help everyone there affected by the disaster.

This is a big result of several years work by the OpenStreetMap community to make open data comprehensive and usable enough for places where it’s needed most .. the places most vulnerable, and also least mapped. It’s taken a lot of effort to listen to the intense requirements of disaster response, and likewise, for forward thinking responders to understand the value of OSM. All “good news”, but of course the floods in Manila are just one of several sudden onset disasters the world face right now, and the unfortunate fact is that they will happen again. This positive step, to my mind, means that we’re just a little bit more prepared.

I really admire how MapAction works (and wish I had taken the chance to train with them when I lived in the UK). They quickly respond to disasters in the first crucial moments, collect data and create maps. They’re agile, but still understand well institutional needs. Mostly, their tools are ESRI oriented, but they have a desire to learn. I expect they used the Shapefile exports from Cloudmade.

This kind of use of OSM, roads for background contextual data in a PDF, is just a first step. OSM is figuring out how to make paper products, and integrate many web and mobile toolkits into deployable, off the grid, interactive applications. We’re developing techniques for authorities to approve revisions of OSM. We’re working closer with responders, in forums they’re familiar with, and soon enough the PDFs on ReliefWeb will contain the message “if you want to download or edit the data in this map, just open this url”. Disasters inevitably strike hardest on the poor, especially the improvised urban living of slums, and we’re working to map these invisible places. Check into Humanitarian OSM Team for more.

Humanitarian OpenStreetMap Team, Next. And also wine.

following on from

Camp Roberts takes place quarterly. The first Crisis Mapping Conference takes place in October .. I was fortunate to find a too brief moment to catch up with co-organizer Patrick Meier in Paso Robles. The Hackistan code sprint looks to be taking place in October. We’re working on holding events in Africa around the AfricaGIS conference, and do some follow up activities with the SOTM Scholarship recipients, especially incorporating Walking Papers in slum mapping.

The Humanitarian OpenStreetMap Team is working hard to take stock of what resources, materials, and organization we’ll need to fully engage with the emergency response community. Great ally Nicolas Chavent, formerly of the UNJLC, had joined to help with his tremendous experience.

Tangent, of course Burning Man is coming up, and Burning Man Earth will be on the playa in full. Here’s an early version of our map. Using very much the same toolkit and approach as Camp Roberts .. agile programming with great frameworks like GeoDjango, offline apps, open data, lightweight APIs .. we’re building out the digital nervous system of Black Rock City. BRC has always been a place for experimentation, a living one week urban laboratory in the harshest of environments. It’s the birth place of the Headmap manifesto. The fun we have out there is going to have beneficial reflections on the “default world”.

A little specific tech for the next Camp Roberts

We’re not done of course. The grid overlay on Walking Papers need tweaking, and probably migration to a real coordinate system. There are standard emergency iconography and schemas, and we need to get proper mapping in place, as we have for the UNSDI-t. My feeling is that OSM tags need a little bit of structured around them, but not explicitly in them, to provide better flow between the Map Features descriptions, tagging process, rendered icons, configuration options in editors, and mapping to other systems. OAuth and OpenID need fleshing out so that things like GeoChat can fully post updates on behalf of OSM mappers. In the long term, the tools for synchronizing OSM from multiple sources are needed, something approaching multimaster-sync, but based in better UI tools for sorting through geographic change and conflicts.

And we definitely need an icon for Kitfoxes!

With all the imagery processing happening, OpenAerialMap was mourned again. The NGA imagery is free to use, but it’s only usable form right now is a drive on the way to Afghanistan. There are lots of ideas on how to relaunch and scale up OAM, and there may be some efforts in the space in the near future.

Finally I have to give thanks (and gloat about) our accommodation for the week. Josh’s buddy Steve Martell runs Kaleidos Wine, and he generously offered space on his vineyard for us to camp, as well as quite a few glasses of wine. It was absolutely killer hanging out there, playing with the dogs and enjoying the incredible view. We also got a tour of the Firestone Brewery from their friend Matthew Brynildson, and that was just incredible. Yea, those 08:00 briefings were pretty blurry for us.

Paso Robles is lovely. If you get down there, check out the Kaliedos tasting room .. fun spot, great unpretentious wine.

Camp Roberts, Open Works

following on from..

OpenStreetMap was just one node in the flow. Through GeoWeb standards .. tiles, GeoRSS, KML .. Sahana was also receiving SMS updates, and DevelopmentSeed built an impressive election monitoring app for Afghanistan.

One piece we didn’t have time to implement was feedback from OSM into Google. Turns out, Sean Wholtman on the enterprise team has developed a process to bring OSM data into Google Fusion! Very eager to see how that develops for Google and OSM.


Preparing for my presentation at the Camp Roberts mapping party, I looked up the above back-of-the-napkin image from my presentation at SOTM two years ago, OpenStreetMap, a disaster waiting to happen. The reflection on the present data integration, data flow, and policy, was startling. Pre-deploy vector and imagery data is gathered. During response, remote sensors in the form of satellite imagery, UAVs (and kites!) are pulled in. Reports are received via SMS and radio. Data is shared among local components, and with the greater web, via lightweight standards. And boggling to me, paper is a two-way medium!

Camp Roberts fulfilled the rough blurry vision of this sketch. Amazing.

But this was just an exercise, right? Turns out all our work is getting immediate use in Afghanistan. Todd Huffman brought a MacMini and portable drive for us to install our work on. The drive got all the imagery. Andrew Turner had already set up the Mini as a GeoCommons appliance, and loaded up loads of data sets for Afghanistan, and locally produced OSM tiles. Walking Papers is there now too.

That MacMini will be installed in the only bar in Eastern Afghanistan. It’s a gathering place for everyone working in the region, from government, NGOs, military and locals. In a social space is a system for social data. This is part of Star-Tides mission, to provide technological solutions that make humanitarian response more effective. Todd is also intending to incorporate Walking Papers in his work with young Afghani geeks at the Fab Lab Afghanistan hacker space, and with construction projects.

Surprising to me, there was simply very little questioning of the crowd sourced approach to data. It was well understood to be effective, and that issues like vandalism and quality are being addressed. The simple proof is that OSM provides the best map of Jalalabad, mostly contributed by Todd. The model works.

final thoughts in the next post

There is a Kitfox. More on OpenStreetMap at Camp Roberts

following from…

The starting point was 50 cm imagery of Aghanistan, released by the National Geospatial-Intelligence Agency. Todd Huffman, known to me from the Beer for Data program, negotiated free use of the invaluable imagery. OpenStreetMap’s response to Gaza demonstrates just one application of the power of easily available imagery. The US government, and many other governments, purchase this imagery with our tax money, and it’s totally reasonable and to be expected that such imagery should be put to its widest possible use in times of crisis.

The Google Enterprise team brought a Google Fusion server, and processed that imagery into Spherical Mercator tiles. With just a little config, everyone was able to incorporate these tiles in their systems (Andrew, you could also highlight tiles as an enabling GeoWeb standard).


Walking Papers blew everyone’s minds. It’s fair to say that paper was the highest tech on display, and emergency responders and other observers coming through the exercises were inevitably intrigued by the possibilities. We got a great deal done in intense coding sessions to make WP more ready for emergency response, most crucially decoupling WP from web services that may not be available in an emergency, or may simply be too expensive to access over satellite data connections common in these situations.

The OSM rails app and tile rendering system were straightforward to take offline, and run locally. Walking Papers took some hacking. The print and scan storage was optionally disengaged from S3. Aaron’s ws-compose went through heavy modification to work with configurable, local tile sources, and allow for overlay of multiple layers. We took the Google processed imagery, overlayed semi-transparent tiles from the local OSM server, and published both in the WP pdf. Our hacking was captured on this github branch, and with some work will make it back into the main codebase.

Walking Papers are meant to be scanned and then digitized into OSM. Camp Roberts threw up another usage scenario: a remote team prints out WP before deployment, but only has access to low-bandwidth communications, like SMS, to report back changes to the map. We worked closely with the InSTEDD team to integrate with GeoChat. On each WP print, a 100×100 grid is overlayed. Along with the bounding box of the paper, this 4 character coordinate can be translated into a fairly accurate lat/long.

When a WP is printed, the local GeoChat server is pinged with the paper ID and the bbox. Their SMS gateway and server was installed at the forward operating base, which provided internet connectivity and solar power, completely off the grid. During the field exercise, SMS reports were sent to GeoChat, including the paper ID and grid coordinate, along with text description and tags. Certain tags trigger GeoChat to pass the message on to the local OSM server, in the form of a changeset with one new node.

Simply, via Walking Papers and GeoChat, OpenStreetMap can now be updated with just a text message!


From there, OpenStreetMap generates new tiles. These tiles are automatically incorporated in other tools. Sahana users automatically see the Kitfox report. Now imagine that this a report of a bridge out, a collapsed building, a shelter for refugees. We have extremely easy to update and share base mapping. This is a big deal.

The OSM server was also producing Shapefiles, KML, Garmin maps, all of which have unexplored potential in disaster response.

The roundup in the next post…

OpenStreetMap at the Camp Roberts Disaster Response Experiments

An extraordinary week at the Camp Roberts experiments. Simply .. the Open approach to disaster response is becoming a reality.

John Crowley of Star-Tides, who many of us met at CrisisCamp (and I still haven’t written up that amazing weekend, it’s been that kind of summer) assembled a stellar crew of developers and thinkers from Google, DevelopmentSeed, InSTEDD, Sahana, TerraPan, and OpenStreetMap. Bring together these projects and talented individuals, poke and prod them with interestingly mis-congruent to agile development, Army style 08:00 daily briefings and “hot washes”, and the amazing collaborations develop.

Though a completely different world, it all had a similar feel to the week long locative arts workshop in Nottingham, organized by Ben Russell, that originally delivered me to OpenStreetMap. That journey, from art to emergency, says everything about the rapid changes Open source, data, and standards are triggering in the most crucial parts of our world.

Eric Gunderson captured the experience with brilliant photos.


Josh Livni and Mike Migurski generously took days out of their intensely creative day jobs to hack, brainstorm, and have a great time down in Paso Robles.


There is a Kitfox

This image says it all. The Kit fox is a threatened species in California, and specifically on the Camp Roberts base the fine for accidental killing is over $100k! This does lead to a lot of grumbled joking, but rather effectively raises awareness of its plight. It became a kind of mascot for our field tests.

The journey of this little report is a result of integration and data sharing among all our systems, rapid development, micro-power and UAVs, and an open data sharing policy that just a couple years ago were just back of the napkin drawings.

The dusty details in the next post…