
Researchers from Stanford University created a poverty map from satellite imagery data and World Bank information.
International organizations used artificial intelligence to create a poverty map for the regions where information is expensive, hard to collect, or publicly unavailable.
While satellites are known for being used in television broadcasts or smartphone maps, they are also monitoring crops, poverty, and illegal deforestation.
At this moment, there are thousands of satellites up in the Earth’s low orbit, and many of them provide valuable data for policymakers, watchdog groups and humanitarian organizations. For example, the World Bank uses satellite surveys to collect data remotely.
Another powerful image is the one taken during nighttime by the astronauts from the International Space Station while flying over the Korean Peninsula, showing the territory of North Korea submerged in darkness.
A team of scientists from Stanford University created an algorithm to use the satellite images in order to identify signs of status and to draft a poverty map of the continents. As a result, the researchers managed to describe the economic situation of five countries from Africa.
The scientists explain that there is an enormous amount of imagery produced by the satellites, and the challenge was to pinpoint the clues that are important in establishing the economic development of a region.
The information on the African continent is particularly scarce. For example, only 39 out of the 59 countries had maximum two national surveys at the beginning of the century. Moreover, the data is not always publicly available. Fourteen of the countries do not have any kind of national polls.
The World Bank worked with a satellite data analysis company to decide where to allocate $100 billion worth loans. The software counts cars, measures agriculture activity, and the shapes of the building.
A data scientist from the University of California at Berkeley used mobile phone data to analyze the patterns of calls in African countries.
Another example is the nonprofit organization that offers money to poor people. Their algorithm included the house roofs, as thatched roofs, foster individuals who are more likely to be poor and metal roof houses are more likely to be inhabited by people with a better economic status.
However, the new algorithm developed at the Stanford University managed to create a program that can identify all by itself the features of poverty in publicly available data from Google, the National Oceanic Atmospheric Administration, and the World Bank.
Nighttime images can show the poverty rate by the quantity of light in a region. The assumption is that the regions that have more light are also richer. However, the night imagery cannot offer information on the density of the population. To overcome the issues, the researchers used data from the World Bank to complete the information.
The team of researchers will further try to improve the algorithm and use the resolution details to obtain more information from the satellite imagery. They will also compare pictures taken at different points in time to determine the local development pace.
Image Source: Flickr