[HOT] Map4Haiti One month later - A major but forgotten crisis

Pierre Béland pierzenh at yahoo.fr
Sun Nov 6 22:41:57 UTC 2016


October 4, the Category 4 Matthew Hurricane with winds over 230 km/h hit the south-west point of Haiti.  Many communities along the coast in southern Haiti as well as inland were very devastated by Hurricane Matthew. The regions most affected are Grand'Anse, Nippes, and Sud.


One month later, we read worrisome reports. FAO is reporting that agriculture has been virtually riped off in a region were people live from subsistence farming. Food security of 800,000 haitians is seriously threatened. More then 80% of houses are severly damaged or destroyed. Moreover, OCHA reports 774 schools damaged or destroyed. It is even reported that hundred of communities did not yet have any help over the last month. 
https://twitter.com/FAOemergencies/status/792689507719983104
https://www.humanitarianresponse.info/en/operations/haiti/document/haiti-hurricane-matthew-situation-report-no-19-2-nov-2016-en

The OSM-Haiti community and the OSM volunteers that activated with them, we concentrate in the Grand'Anse, Nippes, and Sud Departments on the western part of the Haiti south Peninsula. The West Africa OSM contributors organized yesterday a mapathon to help Haiti. 

After the first two weeks of october, less edits are reported. I invite the OSM contributors  to react and respond to this humanitarian crisis as we did for previous ones. We need to support the humanitarian organizations with Quality data. Tasks are still available for those that want to help.http://taches.francophonelibre.org/

While mapping, I see a lot of errors such as trees interpreted as huts (there are generally no huts in Haiti), houses where we see no houses on images, unsquarred buildings, areas not completed.  Plus beginners participated to some complex tasks with  images not aligned to Bing or fuzzy and difficult to interpret. The Special DigitalGlobe imagery that was available for the Haiti response was not aligned with the Bing Imagery. With a mix of buildings traced with either Bing or the Special DG Image, we need to get back and realign the data before completing adding features.
If we look retrospectively to mapping done in Haiti over 2016, we see that there were1. Mapathons organized in early 2016 with unsufficient validations
2. New Mapathons organized for the first two weeks of the Haiti Response with insufficient validations and access to complex tasks by New contributors.
For the month of october, 2,419 OSM contributors participated to the Haiti response. They edited 3.2 million objects. Among them, 934 contributors (39% of contributors) edited less then 200 objects (2% of edits).  This correspond to a maximum of 40 houses. Most of them had no experience and often did not interpret the imagery correctly. In Pascal Neis statistics, these contributors are classified Hit and Run contributors. For their first and often only contribution, they make a lot of mistakes and need a good monitoring and support. Their work is spread in all the area that we cover and it is difficult to later spot and correct their mistakes.
Quality problems
Since 2012, I lead most of the major OSM humanitarian responses. I remember how pleased we were for the Mali and Haiyan/Philippines Responses to gather on the IRC with developpers and humanitarian organizations, to develop tools, to innovate to assure we can respond quickly to deliver a Quality map to the first responders that will land in a country they dont know, to help them navigate with OSM tools, to spot major features such as schools, to extract valuable OSM data for the planning of their response.  These responses since the Haiti earthquake in January 2010 has given a lot of credibility to the OSM community. It really has placed the OSM map on the desk of the UN and humanitarian agencies, help the Word Food Program to distibute goods, help doctors to travel in the remote areas of Nepal.
In parallel, a lot of media have talked about these actions and this brought more and more Mapathons that we control less and less. We need to address this problem and realize that the Haiti 2016 disaster is a major disaster. It needs a better response from the OSM community. We need to innovate again, and find ways for the various communities, Mapathons organizers to innovate in the way to respond. 

I saw the gradual impact of Mapathons on the quality of data produced in the context of crisis such as Ebola, Nepal and Haiti.  Mapathons were also organized for Haiti early in 2016 to prepare for eventual disasters. For the first 2 weeks of november, HOT and Missing Maps organized also Mapathons. This bring in important number of beginners.

In the context of Crisis like this Haiti Matthew Hurricane, we have more general, systematic quality problems. From my experience leading nearly all the major OSM Responses since 2012, plus seeing the increase of Mapathons and quality problems, I say that we need to  address the Mapathons Quality problems. Other then developping tutorials, we need some Accountability from the organizers of these Mapathons. It is not enough to open a Tasking Manager Project and let anyone with no experience map incorrectly or partially. Organizers should assure that as a Team, they can deliver a Quality OSM Map.  
With the popularity of the Mapathons in Crisis context, we did hit for the Nepal 2015 Response records of 1 million edits per day for almost a week.  At the same time, we did experience a lot of quality problems. We had a lot of discussions on the HOT list about that last year. After I lead the Nepal Response, I reported at the Worshop in Bergen about quaity problems and the necessity to find solutions with thousand of Hit and Run contributors bringing in a lot of quality problems. I think that we should also avoid Hit and Run Mapathons that cover too large, that spread inaccurate data on the map and often discourage experienced contributors to participate.
http://www.slideshare.net/pierzen/how-openstreetmap-responds-to-disaster-crisis-digital-revolutions-workshop-bergen-norway-20151102
If you organize mapathons, it is very important that enough experienced contributors participate to the Mapathon and revise the work of beginners. Tasking projects could cover smaller areas to assure that Mapathons contributors do not spread mapping in large areas without at the end completing the mapping. We should also select easy tasks and monitor constantly their work. We surely need good monitoring tools for Mapathon organizers to assure the quality of data produce, and that the projects are completed.
regard
  
Pierre 




   

   
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstreetmap.org/pipermail/hot/attachments/20161106/66ba451c/attachment.html>


More information about the HOT mailing list