Google Maps is so comprehensive that you can use it to plan a New York City subway trip down to the minute. But in the parts of West Africa affected by the Ebola epidemic, Google barely has the roads mapped out. More often than not, the names of villages are missing–and sometimes the village altogether.
“Google’s business model is selling advertising, so it’s simply the business case,” says Andrew Buck, a volunteer coordinator with the Humanitarian OpenStreetMap Team. “Starbucks isn’t paying for Google to advertise over there so there’s very little incentive for Google to improve its maps.”
Organizations working in the three hardest-hit countries, Sierra Leone, Guinea, and Liberia, need doctors and medical equipment. But they also need something more basic: maps to help aid workers get around the country and do the difficult job of checking village by village for victims of the disease.
That’s why organizations like the UN, Red Cross, and Doctors Without Borders have turned to OpenStreetMap (OSM) for their map data. Commonly referred to as the “Wikipedia for maps,” OSM is a crowdsourced mapping project that brings together mappers on the ground using GPS devices with map editors working remotely.
A subset of the OSM community, the Humanitarian OpenStreetMap Team (HOT) has taken on creating more robust maps of the affected countries in West Africa. HOT is a specialized team that responds to international humanitarian crises by corralling OSM volunteers to gear their efforts to impacted areas.
Using free and open source software designed by them specifically for this purpose, HOT has been able to produce detailed, freely available maps.
Gueckedou, Guinea is a large border city where it is believed the current Ebola epidemic originated. OSM’s map includes a much higher level of detail than Google’s, including names of neighborhoods, landmarks, and terrain detail. Google Maps does not even include the two rivers that meet in Gueckedou.
OSM’s maps of the region were not always this good. In fact, prior to the epidemic, the OSM map of Gueckedou looked very similar to Google’s–if not worse.
The detailed mapping of areas impacted by Ebola are thanks to the efforts of Humanitarian OpenStreetMap Team, which traces its roots to the Haitian earthquake in 2010.
OpenStreetMap had first been used during a response to a typhoon in the Philippines the year before, but In Haiti HOT coordinated directly with disaster response teams, learning what was most mission-critical for them and in turn training them on how to get the most out of OSM.
“During the earthquake in Haiti there was an overwhelming response from OpenStreetMap contributors,” says Kate Chapman, executive director of HOT.
With the knowledge that HOT was able to meet the challenges of real-world disasters, relief organizations started approaching them directly. Since then Buck and Pierre Béland have coordinated several HOT efforts, called “activations.” During the 2012 coup in Mali, they worked with the United Nations to map the country. And during Typhoon Haiyan, which hit the Philippines last year, which was the first time that the group worked closely with the Red Cross.
When Doctors Without Borders began working in West Africa on the Ebola epidemic in March, its aid workers were sorely lacking maps of the area. So they requested that HOT map three major towns in Guinea. Buck and Béland were quick to respond.
HOT works by getting satellite imagery of the areas they want to map. Volunteers sitting behind their monitors around the world then do the tedious work of translating those images into map data. Some satellite imagery is publicly available, some is purchased by organizations supporting OSM’s work, while other imagery is provided through special agreements with companies like Bing.
According to an internal Doctors Without Borders Ebola response case study, HOT mapped the three priority cities (Guéckédou, Macenta, and Kissidougou) in less than three days. “Within five days, 244 volunteers had mapped more than 90,000 buildings,” according to the report.
“If we get a high-priority request we can map that area pretty quickly and thoroughly–often in under 24 hours,” says Buck.
Since March the team working on West Africa has mapped a staggering 8,000,000 objects, which in this context means individual buildings, roads, rivers, fields, landmarks, and more.
Since Ebola is spread by touch, the main method of containing the outbreak is contact tracing: Following the steps of an infected person to see with whom else they might have come into contact. Without adequate maps, that job is near impossible.
According to Doctors Without Borders, “one the biggest challenges the team faced was that the names of many villages were unknown.” There are also many villages with similar or identical names. For example, the village name Bendou exists 14 times within the prefecture.
Many villages also have multiple names. The village of Rosos in Sierra Leone is also known as Sos, for example. If a sick patient tells a doctor they come from “Sos” and a map only knows that village as Rosos, it can lead to delays, and unnecessary trips to the wrong locations can expose staff to additional risks in a country where foreign doctors are not always welcome.
HOT overcomes this problem by integrating their maps with publicly available databases of place names like the National Geospatial-Intelligence Agency’s Names Server. As you can see in the screenshot below, Rosos has a custom alternate name value of “Sos” and it’s flagged that the source of the data is the GNS database. Now if a user searches for “Sos, Sierra Leone” in OSM Rosos will pop up as a search result.
Road mapping is another key area in which HOT helps aid workers. “You’re more likely to find sick cases in a town connected by a road,” says Buck. “So Doctors Without Borders will know where to concentrate efforts for follow-up visits and education.”
Having roads on their maps of course also serves a more basic purpose for relief workers: helping them get around. “They use our maps just for the roads to route trucks and personnel,” says Buck. OSM also has special features that designate road quality–is it dirt or paved? potholed or muddy?–which might be factored into logistics considerations.
That level of detail can’t always be recognized from satellite scans. So volunteers on the ground will feed information back upstream to HOT volunteers and ask them to integrate it into the maps.
Aid workers on the ground also feed information back to HOT to name specific neighborhoods within Gueckedou. Those names help doctors pinpoint where exactly someone is from in a large city. But those informal community names aren’t on any official maps and can’t be learned from satellites.
“All the neighborhood names in Gueckedou come from Doctors Without Borders when they surveyed people on the ground in March,” says Buck.
All of this is possible because of HOT’s intense volunteer coordination combined with OSM’s software, which is built from the ground up to meet the organization’s needs.
Humanitarian organizations send mapping requests to HOT, which manages the volunteers responding to the request. This reduces the workload for the requesting organization.
HOT’s online task manager facilitates crowdsourcing the mapping work. Tasks are tagged by priority and show whether they were requested by a specific organization or are for general crisis response. According to Buck, tasks that are requested specifically by organizations like Doctors Without Borders, the World Health Organization, and the World Food Program, tend to be completed the most quickly.
“People feel like they are having a direct impact when they see that a group like Doctors Without Borders needs their work,” he says.
The most basic “task” that needs to be filled is tracing over satellite imagery to create basemaps. That work is accomplished mainly with the Java OpenStreetMap editor (JOSM), an open source tool created by and for OSM volunteers.
JOSM allows mappers to do the digital equivalent of putting trace paper over an atlas to create a map. With an interface that feels like Paintbrush or MS Paint, volunteers can draw maps with specialized tools for buildings, roads, and water features. Tasks get divided up into “tiles” on the map grid that volunteers click on to begin working. A completed tile is reviewed by veteran OSM editors to verify that everything has been mapped and tagged accurately.
Buck gave me a personal tutorial of JOSM by sharing his screen over Skype. With less than a half hour’s training, I felt confident tracing buildings and rivers for OSM. So it really is something anyone can do.