184
E. Eros
The American Red Cross is actively developing digital tools and ways to gamify
remote mapping, but the humanitarian community could benefit from additional
insight and tools for engaging remote volunteers. Additionally, ensuring data quality
becomes especially important when adding large numbers of new mappers into the
OSM community. The American Red Cross has emphasized the importance of data
validation and held events for more experienced mappers in order to check and correct
OSM data. More work remains in this area.
Machine learning algorithms offer another area for future research and exploration. For larger mapping needs where crowdsourcing is less feasible or timely,
organizations recently began to experiment with machine learning methods and outputs. For example, Facebook, Columbia University, and the World Bank pioneered a
method to identify buildings from high-resolution,
11 commercially available satellite
imagery (Tiecke 2016). The partners trained the model to adapt it to local conditions
in 18 different countries, ground truthed the results against household survey data
and existing crowdsourced OSM building data, and applied most recent government census population data to outputs. The result is openly accessible, nationwide
datasets that contain estimates of human population down to the 30-m scale. Data
are available for 18 countries at the time of publication (Facebook Connectivity Lab
and CIESIN 2016) and agencies like the Red Cross have begun testing their applications for humanitarian purposes, and in combination with crowdsourced and locally
derived data. As these techniques become even more refined, more work is necessary
to investigate how computer algorithms, crowdsourcing, and local knowledge can
work together to create valuable data for humanitarian purposes while also respecting
privacy and sensitivities.
For field mapping, sustainability and long-term engagement are areas for additional growth—especially for organizations working internationally. There have been
highly successful initiatives undertaken by HOT in Indonesia and by the World
Bank/KLL in Kathmandu, and mapping engagement has continued following the
end of the American Red Cross’ mapping project in West Africa. Again, this area
could benefit from additional insight into ways to effectively engage with local communities, promote local interest and ownership over data collection, and support a
community of mappers, who are trained and motivated to continue mapping activities
independently. Rural areas may present a particular challenge; previous tech hubs
and mapping initiatives have been based out of urban centers like Nairobi, Kathmandu, and Jakarta, which benefit from more developed infrastructure, the presence
of university students, and a more technically skilled volunteer base.
Finally, as the mapping community grows, it is important to bridge the gap between
those who are experienced in working with data and technology, and the broader community of practitioners and decision makers, who may be less comfortable with digital tools and resources. Base data and analytics bring higher value when more people
understand how to interpret and apply them to situations. For this reason, the Missing
Maps partnership has emphasized efforts to increase data literacy and what we call,
11 0.5 m resolution.
E. Eros
The American Red Cross is actively developing digital tools and ways to gamify
remote mapping, but the humanitarian community could benefit from additional
insight and tools for engaging remote volunteers. Additionally, ensuring data quality
becomes especially important when adding large numbers of new mappers into the
OSM community. The American Red Cross has emphasized the importance of data
validation and held events for more experienced mappers in order to check and correct
OSM data. More work remains in this area.
Machine learning algorithms offer another area for future research and exploration. For larger mapping needs where crowdsourcing is less feasible or timely,
organizations recently began to experiment with machine learning methods and outputs. For example, Facebook, Columbia University, and the World Bank pioneered a
method to identify buildings from high-resolution,
11 commercially available satellite
imagery (Tiecke 2016). The partners trained the model to adapt it to local conditions
in 18 different countries, ground truthed the results against household survey data
and existing crowdsourced OSM building data, and applied most recent government census population data to outputs. The result is openly accessible, nationwide
datasets that contain estimates of human population down to the 30-m scale. Data
are available for 18 countries at the time of publication (Facebook Connectivity Lab
and CIESIN 2016) and agencies like the Red Cross have begun testing their applications for humanitarian purposes, and in combination with crowdsourced and locally
derived data. As these techniques become even more refined, more work is necessary
to investigate how computer algorithms, crowdsourcing, and local knowledge can
work together to create valuable data for humanitarian purposes while also respecting
privacy and sensitivities.
For field mapping, sustainability and long-term engagement are areas for additional growth—especially for organizations working internationally. There have been
highly successful initiatives undertaken by HOT in Indonesia and by the World
Bank/KLL in Kathmandu, and mapping engagement has continued following the
end of the American Red Cross’ mapping project in West Africa. Again, this area
could benefit from additional insight into ways to effectively engage with local communities, promote local interest and ownership over data collection, and support a
community of mappers, who are trained and motivated to continue mapping activities
independently. Rural areas may present a particular challenge; previous tech hubs
and mapping initiatives have been based out of urban centers like Nairobi, Kathmandu, and Jakarta, which benefit from more developed infrastructure, the presence
of university students, and a more technically skilled volunteer base.
Finally, as the mapping community grows, it is important to bridge the gap between
those who are experienced in working with data and technology, and the broader community of practitioners and decision makers, who may be less comfortable with digital tools and resources. Base data and analytics bring higher value when more people
understand how to interpret and apply them to situations. For this reason, the Missing
Maps partnership has emphasized efforts to increase data literacy and what we call,
11 0.5 m resolution.