Artificial intelligence and satellite data could change the way we map global poverty

Satellites staring down at Earth can see a lot from their posts in space.

Powerful eyes in the sky can pick out homes, natural formations, the pyramids and even small cars driving on roads. And now, scientists are using the wealth of data collected by these satellites to solve major problems on Earth.

A new study published in the journal Science this week uses machine learning -- a type of artificial intelligence that lets computer algorithms change when given new data -- coupled with satellite imagery to map poverty in Nigeria, Uganda, Tanzania, Rwanda and Malawi.

This new technique could help revolutionize the way groups find impoverished areas and eventually get relief to people living in those specific parts of the world.

Today, aid workers conduct surveys on the ground in poor areas, but that kind of mapping is incredibly time-consuming and even expensive. Developing a faster, more automated system could save time and money.

Plus, it could help get aid to those who need it more quickly.

"I think the goal is to understand the world much better, in particular the world where a lot of poor people live. And that includes understanding their livelihoods in terms of their sources of income, how their agriculture is performing, how different sectors of the economy perform, and, more specifically, what actually is effective at improving conditions," co-author of the study David Lobell told Mashable in an interview.

"All of that is really difficult if you don't have good measurements."

Original image has been replaced. Credit: Mashable

How it works

But creating the tools used for the study wasn't exactly simple.

The authors of the new study needed to figure out a way to predict poor communities without clearly knowing if they are poor or not.

"There are few places in the world where we can tell the computer with certainty whether the people living there are rich or poor," study co-author Neal Jean, said in a statement. "This makes it hard to extract useful information from the huge amount of daytime satellite imagery that's available."

To solve this issue, the scientists used day and nighttime satellite imagery to see which parts of the countries are brighter than others.

Usually, areas with less artificial light are less developed. Therefore, the scientists mapped those dimmer parts of the map at night with daytime photos of the same areas, allowing the computer to pick out patterns -- like road conditions or metal roofs versus thatched roofs -- that indicate a less-developed and possibly poorer region.

According to the study, this method is 81 percent more effective at predicting poverty in an area under the poverty line than a method using nighttime imagery alone. The authors also think that the algorithm may be effective when used in multiple countries and regions.

Artificial intelligence and satellite data

Satellite data is ubiquitous in our modern world.

Corporations use the high-resolution imagery to track specific data points like the number of cars in a parking lot at certain times of day or year.

Pairing machine learning with satellite images is a new but growing field, however.

Companies like Orbital Insight are planning to use satellite data and artificial intelligence to track deforestation and communities at-risk for it.

Because satellite companies like Planet and DigitalGlobe are continuing to launch commercial satellites that can beam images down to Earth, the pipeline filled with high-quality photos of Earth from space won't be slowing down anytime soon.

"There's a lot of competition in the area of high-resolution imagery right now, and then there's also a lot of interest from these companies to have a positive impact on the world as long as it doesn't undercut their business," Lobell said.

This kind of project could meet that goal, Lobell added.

Where to go from here

While the new study's methods may save time and money, they aren't necessarily a replacement for on-the-ground surveys, according to Marc Levy, deputy director of the Center for International Earth Science Information Network at Columbia University.

"The study demonstrates the power of combining multiple data streams to measure things that matter.  Satellites plus surveys are vastly more powerful than either one alone," Levy, who was not involved in the new study, told Mashable via email. "This insight has been borne out in repeated studies across many issue areas, and this paper shows how to leverage it to get timely, accurate, spatially precise estimates of poverty."

Levy also thinks there's still room to grow for this kind of work.

"If you consider the core insight of the paper that combining data streams adds inferential power, then a logical next step is to hypothesize the most-valuable addition to the satellite, survey mix," Levy said.

"Based on what has been learned in other fields, I think a combination of selective, dedicated field campaigns and broad-based crowd-source campaigns would have a very high chance of dramatically improving the accuracy of the results presented here."

Although the results are encouraging, Levy doesn't necessarily think these kinds of methods will be implemented right away on a large scale, and there might be challenges to broadening these models to the rest of the world.

Levy said that the five countries examined for this study are "much more similar to each other than they are as a group to other world regions," so applying this same model to other countries might come with unique challenges.

But Levy is hopeful none-the-less, saying that he expects those challenges can be overcome.