Spatial Networks recently attended two very fascinating conferences. The first was the annual meeting of the Association of American Geographers, which was aptly called AAG 2014. The second was the annual United States Geospatial Intelligence Foundation’s Symposium called GEOINT 2013*.
Presentations at both conferences demonstrated the growing field of remote sensing. At AAG, Gary R Watmough from the Earth Institute demonstrated how algorithms for analyzing satellite imagery can be used to identify rich and poor areas in India. In a presentation at AAG and a booth at GEOINT, geniuses from the Oak Ridge National Laboratory showcased LandScan, a project that uses remote sensing algorithms to predict population levels around the world. Many companies at GEOINT are actively developing algorithms for mapping the physical world with LiDAR. Although not presenting at the conferences, the World Resources Institute is helping run Global Forest Watch, an innovative program to monitor all forest around the world using remote sensing. The Harvard Humanitarian Initiative is also using remote sensing to detect large-scale attacks and human rights abuses. It is clear that remote sensing, especially satellite imagery analysis, is a diverse field full of potential.
With all of that said, for the remote sensing industry to reach its greatest potential it will have to leverage the data and services of on-the-ground data collectors, such as Spatial Networks. Imagery analysts often check if their algorithms work by comparing their results with census data and other open datasets. Unfortunately, censuses conducted around the world, especially in many conflict-affected countries, can be highly inaccurate, poorly organized and politically biased. Testing algorithms using biased data can lead to biased algorithms. Ground-truthers, like Spatial Networks, have a solution to this problem. We gather data on-the-ground from all over the world in hard-to-reach areas. If you want to know if your algorithm correctly made a prediction about a location, send us over there to find out!
Remote sensing and mobile data collection are essentially two sides of the same coin. They both seek to ascertain the truth and can work together to mutual benefit. Just as information collected on the ground can test the accuracy of remote sensing algorithms, these same algorithms can identify interesting areas as candidates for mobile data collection.