Late March, just coming back from Burundi, I have been invited by Guido Lemoine from the Institute for the Protection and Security of the Citizen – IPSC, Geospatial EMergency Management Action – GEMMA, that is part of the Joint Research Center (the European Commission’s in-house science service). The aim was to organize a one day workshop about HOT and OSM data in crisis contexts, as well as to have specific discussions with some groups or team within this huge research center located in Ispra, Italy. So yes, this is not the freshest piece of news ever published here, but anyway JRC’s interest for OpenStreetMap and concrete activities are still ongoing; thus I think this is an interesting reality for the HOT community to know about, moreover I guess this is not well-known by most of us.
So basically the morning (10AM-12PM) was dedicated to an open presentation followed by questions to all the interested departments in the IPSC conference room. More than 30 people showed up from various departments and after my presentation questions were about attributes, quality, possibility to lock data considered as updated, what is quite common regarding OSM. The INSPIRE Are3na team was also interested to find the ways to transform OSM data into INSPIRE frame. INSPIRE is so far a thematic that has not be really discussed in the OSM community, orsometimes considered as not open. Might be an opportunity to start a discussion with the technical people in charge in European Commission. Another unusual question was about the possibility to find slum mapped areas. It is true we do not have a specific wikipage about this and, as far as I know, no specific tagging about this topic.
In the afternoon, I met different groups or departments that wanted to dig more on specific matters. Generally, these JRC departments are very interested in OSM or even use massively OSM data, and would like to go further (how to contribute, how to adapt themselves to OSM to fit their needs, etc.).
As stated in their webpage, “GDACS is a cooperation framework between the United Nations, the European Commission and disaster managers worldwide to improve alerts, information exchange and coordination in the first phase after major sudden-onset disasters." GDACS intervenes in various thematics:
- trainings (eg on the methodology they developed for tsunamis)
- crisis room if needed
- work with national civil protections, like in Haiti
Their Alert system for various natural disasters, including vulnerabilities, is used to produce crisis maps. Before satellite imagery are available, daily routine maps, situation reports are done. They always look for the best available data in their areas of interest and OSM data is one main source. They created ArcGIS rendering for Haiti with OSM.
They wanted to talk with me about various topics:
- how HOT is or can be triggered? We looked together at the digitalhumanitarians.org website, especially the activation graph. But obviously, if ever JRC needs OSM baseline data in an emergency context, they can directly contact HOT through the HOT list
- place names at different admin levels + disputed places (eg Kosovo or Kashmir). How this can be addressed in OSM, as using it as a background can bring dispute issues for GDACS? I made them read the Disputes wikipage, so that they figure out this issue is under discussion within the community, and specific rendering for every case might be created in the future to solve the issue
- buildings. Attributes that would be useful for GDACS like materials, etc. They are basically very interested in Preparedness and they would like to use/create consistent models with the OSM community. They had in mind what has been done "in Padang". Of course I redirected them to Kate
- quality of data. To point out what are the areas that can be considered as good. I reminded them about the fact nothing can be locked in OSM, but that it is possible to design a tool to have an overview of any kind of change on an Area of Interest, through a analysis of diffs
IPSC GEMMA pilots an outsourced mapping service, called Copernicus Emergency Management Service, producing on demand reference maps for European Commission, on both non rush (a few weeks delay pre- and post-disaster) and rush (within a single day, for disaster response) modes, as explained in this document. OSM is one of the main data source used to produce those maps, once imported into GIS vector formats like shapefile.
At IPSC – Globsec – Isferea ("Geo-Spatial Information Analysis for Security and Stability"), I met Mayeul Kauffmann, researcher, consultant at JRC and OSM fan who developed various projects directly or potentially related to OSM. Posted on SourceForge, OSM2postgresql is a SQL bash that solves any topological issue thanks to more tham 1300 lines of code and does not need to input a preset because it uses the tagshstore system to store absolutely all the attributes in only one field with no loss of attributes or columns threshold. The demo with QGIS for rendering and QGIS server to create tiles based was really impressive. Those who would like to know more and test it can visit Mayeul’s channel here. He also developed an algorithm to improve GPS traces accuracy and then imagery georeferencing by using a mean distance combined with analysis tools to check and eliminate too extreme values. From a starting 20 m accuracy, finally it ends up with a 1-2 m georeferencing accuracy. EC-JRC lab also developed an automatic tool to recognize and create built-up areas. Result with high res imagery like over Ispra are really impressive, very close to the areas drawn by the OSM contributors. Mayeul developed a kind of Crowdsource Image Recognition like we use for South Mopti, Mali, but to validate if the automatic built up area is good or not.
As data consumers, food of thoughts, task force or tools developers, these JRC institutes and labs are thus participating directly in the main OSM fields or would be keen on doing it. Considering their scopes sometimes relate even directly to emergency response or crisis preparedness, working closer with them would certainly be fruitful.