Satellites Helping Countries shape their Economies
For almost six decades since Sputnik first circled the Earth, satellites have been the exclusive domain of the richest governments and companies, costing billions and weighing tons. Now with the advancements in space flight and digital cameras, cheap orbiting cameras are becoming ubiquitous.
When you visit the Gallery section on NASA’s EPIC Website you see some of the most beautiful images of Earth taken from shoe-box shaped satellites sometimes hundreds of miles high. But if you ask a data scientist, who really has an eye for data, he would tell you that these images reveal more than just how beautiful the earth is. If proper resources are implemented, these images can predict a wide range of economic and societal trends happening down below.
For example, In Myanmar, night lights indicate a slower growth rate than what the World Bank estimates suggest. In Africa, photos of homes with metal roofs can show transition of poverty. In China, trucks in factory parking lot can indicate industrial output.
Images from these and other satellites, combined with big-data software can help us observe things that are too large to be seen by the human eye.
I had a chance to read about one of the companies working in this domain (thus the urge to write this). They use advanced image processing software and machine learning algorithms on satellite images of 60 US retail and restaurant chains to estimate their product sales. While one other product of theirs generates a global poverty map and predicts illegal deforestation by watching road construction and other signs of logging.
These technologies are breathing new life into economics, where current policy is driven by information gathering techniques dating back to the messenger pigeon era. Implementing machine learning and advanced image processing algorithms, images can be processed into insightful statistics within days, not months or years, and can also help re-evaluate old data which might reveal new answers.
But, despite these advances, satellite images may never be able to provide data for some sophisticated economic indicators like inflation or employment, much less for more-abstract gauges like market perception or lending rates. Even the ability to count every car in a Wal-Mart parking lot several times a day can’t show how much customers spent.
Ground-based surveys will still be needed to validate the satellite data, but what really is positive here is the fact that satellite analysis combining surveys are vastly more powerful than either one alone.
“One should believe that this is only the beginning of a remote sensing data and software revolution, and that as these tools develop they will be of increasing importance to not just economists but to the entire world as a whole.”
Read more at medium.com/@jitesh.bits