- Glenn
Eye in the Sky

Monitoring people and the planet from space Imagine a world where a satellite image instead of a non-existent income statement, is used to determine whether a small scale farmer in Mozambique is granted credit or not. As we go about our work on the African continent, we’re finding that the answers to some of the world’s most intractable developmental problems may lie in the sky above us. Humanity’s fascination with outer space has led us to gaze longingly at the universe surrounding us for centuries, wondering what magic awaits to be discovered out there. This obsession inspired massive investment of capital and human energy in the 20th century to explore our solar system and the universe. But only in 1960, when the first satellite image of the earth was captured, did we begin to realize the value of turning our gaze to look at our own planet from space. Since then almost 200 satellite missions with the capability to image the entire earth have been launched by over 30 countries, generating massive volumes of data every second.
The applications of earth observation data include weather forecasting, climate prediction, monitoring agricultural yield, change in land cover, and mapping economic indicators. Data that previously would have taken years, decades or even longer to collect are being assembled into collections that cover the entire globe and can be remeasured daily. A few years ago it would have seemed ludicrous to propose that we could count all the trees in the world (roughly 3 trillion in case you were wondering). Today, enabled by advances in satellite technology, optics and computing, this and other amazing achievements are within our grasp. It is easy to be misled into believing that earth observation data will eliminate the need for intensive ground-based data collection. Why go door to door surveying households about their economic well-being, or counting trees in a forest when we can simply measure it from space? The reality is that the raw data collected by all earth observing satellites is simply the reflectance of light in different wavelengths. This raw information is translated into meaningful measures of the planet and its people, using the ground-based data we are familiar with. These are used in combination with sophisticated machine learning algorithms that detect how these ground-based measurements are related to light reflectance measured by satellites. We still need the manually collected data, perhaps more than ever. The power of earth observation data comes from the ability it conveys to scale ground-based data from point locations in space and time to broad geographic and temporal extents. An interesting application where this power is evident is in the mapping of poverty. Artificial illumination of town and cities is a useful indicator of development and economic activity. By measuring the intensity of nighttime illumination and combining this with estimated Gross Domestic Product, it is possible to rapidly map economic activity and supplement data where existing information is sparse or unreliable. The methods and applications of earth observation data to mapping human wellbeing are increasing in sophistication. Recent work combining household surveys with daytime satellite imagery using neural networks, helped estimate household consumption, expenditure and assets in developing countries, allowing these important measures of development to be monitored far more intensely than conventional surveys would allow. Every year, new satellites are launched equipped with sensors able to monitor the earth at ever increasing detail, with much of this data freely available. With increases in computing power and more sophisticated machine learning methods being used to combine this data with painstakingly collected ground data, new opportunities for learning about the state of our planet and its people are arising. So the next time you gaze skywards at night and see the glimmer of a satellite speeding overhead, pause for a moment to appreciate the wealth of information being provided by that shooting star. Perhaps the Mozambican small scale farmer will be looking at it too.