Space data – or big data from space – is a term used to describe the camera and sensor information gathered by space-borne monitoring equipment (satellites) and the process of extrapolating patterns from it using analytical software. Think of it as ‘really big data’.
The area is currently experiencing a boom in investment from the private sector. Traditionally, only governments have had the financial muscle to send monitoring equipment into orbit. But technology advances and the rise of private space firms like SpaceX have dramatically reduced costs, improved reliability and increased launch frequencies.
In short, it’s now much cheaper to deploy satellites. A recent report by Sparks and Honey claimed that between 2012 and 2016, the number of satellites launched increased by 53%, and 6,200 small satellites are expected to be launched over the next ten years. “The growing footprint of small satellites and cheaper satellite costs are opening up a vast world of data and vantage points that have the potential to change industries and the way we live,” said the report.
If we can make sense of the information these satellites can gather, space data has the potential to revolutionise how we understand a wide array of industries and environmental phenomena. It’s the ultimate ‘helicopter’ view.
Space data is already used to monitor conflicts and track refugee movement. It is most obviously used for satellite navigation, and is integral to the development of driverless cars, and the monitoring and development of low carbon technologies.
But this is only the beginning. In agriculture, farmers can make use of satellite data to better understand what factors influence the growth of crops – weather patterns, exposure to sunlight, air quality, or pest activity. Space data can also be used to monitor flooding or sinkholes and, with access to historical data, help to predict future areas that may be at risk.
Antycip Simulation provides space/sensor/satellite planning, modelling and analysis software used by the defence, intelligence and commercial satellite communities. Their programmes provide the analytical framework to plan data collection, transmission, transport, and data analysis.
Imagine a vast world of data and vantage points that have the potential to change the way we live.
“One could argue that space data is a subset of big data,” suggests Frank Reynolds, Antycip European Marketing Manager, “with a current focus in advancing intelligent sensors and data science.
“There are multiple processes of data collection, such as Infrared: Electromagnetic radiation with longer wavelengths than visible light. On Earth (and in space), many objects emit infrared waves that cannot be seen by the human eye since they are below the visible spectrum of light. Satellites can detect infrared light from space using special instruments and sensors. There are also Ground Based Radars, which use phased array radar technology with the latest monolithic microwave integrated circuit technology radars can operate over long range and provide greater discrimination, so it can monitor very small objects in space or identify objects.”
How small? The UK-built Carbonite-2 satellite operated by Earth-i can deliver high-definition, full-colour videos from orbit that have a resolution of one metre. That’s good enough to track the movement of cars, boats and planes, even show whether the blades on a wind turbine are turning (and how fast).
“I think we will inevitably become more reliant on shared and fused data,” says Reynolds, “and as we have access to more and more of it there will need to be a change in how we process and make use of it; from expert-based strategy and direction to a more dynamic and learning oriented approach.”
The usefulness of the increasing amount of information generated by space or low orbit sensing devices is dependent on data analysis software. One firm in the field, Orbital Insight, develops software to supply actionable insights for the private and public sector by applying machine learning and computer vision technologies.
The firm uses deep learning algorithms to identify cars from satellite images at 55,000+ parking lots located in major retail chains across the US, using this to provide breakdowns and pattern analysis of consumer behaviour. The information can be used by businesses to optimise logistics and supply chain routes, achieve better visibility for risk management, monitor assets, and enhance understanding of site location and traffic density.
Orbital Insight’s software also monitors all the top oil storage regions in the world including OPEC, China, USA, and EMEA – culminating in coverage of over 5 billion barrels of crude oil storage capacity worldwide. Using artificial intelligence, Orbital Insight claims to provide the most comprehensive and consistent crude stock estimates in the world. Big data meets big oil.
Elsewhere, its forecasts of US corn and soy production can provide traders with crop yield estimates weeks before World Agricultural Supply and Demand Estimates (WASDE) are published. Algorithms are used to project expected crop yields via satellite, weather, and historical data in real-time to produce reports.
Other firms have developed space data applications for environmental protection. In 2016, satellites monitored some unusual activity in the Tereneyskoe Forest farm in Primorsky Krai, Russia. Earth-monitoring company Astro Digital noticed evidence of deforestation and informed the World Wildlife Fund, who were able to launch legal action. Once the firm’s algorithms detected the change, months of data from the European Space Agency's Sentinel-2 satellite was pulled as evidence.
Last year, the firm also demonstrated how space data can be used in building detection during the State of the Map conference in Colorado. OpenStreetMap is a crowdsourced online map of buildings, parks and other populated areas around the world, built by a community of participants. Astro Data demonstrated how using high frequency, moderate resolution satellite images, and machine learning can allow building construction to be automatically detected – using a series of medium resolution images from satellites and a model trained for detecting patterns.
“By engaging mappers in the cycle of detecting and verifying change, the analytical workflow becomes an active learning process and model results improve over time,” says Astro Digital. “The algorithm flags a new building, the community of mappers validate the accuracy of the new building, and the model learns from the user input. In this way, algorithms learn to mimic the user and can complete different regions of the map with increasing efficiency. This two-way interaction improves model results significantly without forcing user to do a lot of manual editing and labelling.”
As private sector investment in monitoring satellites skyrockets, so too will the amount of space data that is collected. With the right analytical tools and enough processing power to crunch the numbers, that reservoir of information can be applied to an increasing amount of industries and public interests, saving money, improving efficiency, and making life that little bit better for all of us.