How much satellite data does it take to map the skies of our big blue planet? It depends on who you ask. A few years ago, a Russian weather satellite took 121-megapixel (MP) images of Earth1. At the time, these photos were the highest-resolution, single pictures of our world. To put that into perspective, a popular, high-end smartphone takes 12 MP images, which are one-tenth of the size2.
If you ask the U.S. federal research agency responsible for aerospace research, they will have a different answer. Instead of taking individual photos, the organization maps Earth’s atmosphere by stitching together images from multiple flybys. This was the same way that astrophysicists generated the first image of a supermassive black hole – which took nearly five petabytes of data.
Satellite data is big data, and dealing with it requires smarter ways of collection, analysis, and archival.
A Constellation of Satellites
So, how exactly do satellites help store the data that maps Earth’s atmosphere? It takes a network of these devices, operated by either governments or private businesses, working together in harmony.
The first type are known as geostationary satellites. These satellites are placed directly above the equator and match the rotational speed and direction of the earth. Polar-orbiting satellites, on the other hand, revolve around the North and South Poles. Their speeds and directions can vary. Together, these satellites collect images of Earth from all angles, which are passed to ground stations for further use.
But, collecting satellite imagery isn’t as easy as snapping a photo on a smartphone. There are many characteristics of the data to keep in mind. First among them is resolution – how clear an image from space will appear:
- Spatial resolution – the pixel size correlated to the image subject’s surface area
- Spectral resolution – the wavelength(s) that images are taken at (visible, infrared, etc.)
- Temporal resolution – the length of time between images collected at a specific location
- Radiometric resolution – the levels of brightness and contrast that an image records
By adjusting these individual levers, satellites can fine-tune the images that they collect to suit their clients’ specific needs. Turning this data into insights, however, takes another big step of data analysis and processing.
From Satellite Images to Insights
With the sheer size of the earth and high-resolution of its images, the repositories that store satellite data are quite large. But, turning those files into useful visualizations that can be licensed to government and other businesses for a profit takes analysis. Today, that processing is often being carried out by techniques such as artificial intelligence (AI) and machine learning (ML).
With ML, powerful models can be built to better understand satellite imagery. For instance, machine learning can auto-generate geographical tags for an image based on satellite and airborne data. In addition, ML can help clean up a satellite image by using a method called “image destriping” to remove unwanted stripes and streaks in the raw data.
Whereas ML is used to understand satellite data, AI is used to apply satellite data to real-world problems. For climate scientists, intelligent analysis of satellite imagery can help monitor crop health, vegetation, and water sanitation. This includes deforestation, which a leading research university held a global predictive analytics challenge to address3. For companies in the business of weather data, AI applied to satellite data helps create extended forecasts, seasonal patterns, and severe weather warnings that are invaluable to the logistics and shipping of their clients.
Through AI and ML, satellite data analysis helps inform smarter business decisions and environmentally conscious public policies. For data that goes unused, archiving becomes the next step to inform future applications.
Archiving Satellite Data for Future Use
Depending on the satellite operator, access to archived satellite images can be restricted. Typically, commercial satellite companies avoid placing their imagery into the public domain. Instead, their data is archived in large databases – either cloud-based or on-premises – and licensed to third-parties for a fee.
For government satellite operators, though, satellite data is stored and made publically accessible through shared processing programs. A perfect example is one of the world’s largest active archives of environmental data. The archive brings together three national data centers to provide open access to over 25 petabytes of data about the weather, climate, coasts, oceans, and land masses4.
The actual process to archive the data is clearly defined. Before being archived, two options are available depending on the frequency and volume of data being produced. After the archive request is assessed, the satellite data is properly assessed, documented, transferred, and sent to long-term storage. Each month, the research organization archives nearly 26 terabytes of environmental data across 130 platforms4. Their data is open to the public through an extensive online catalog.
Learn Beyond Satellite Data to Space Data
Technology has made possible higher-resolution images from our friendly home planet to the depths of our universe. Explore more stories from our series on data from space.
- Solving the Data Transmission Challenge in Deep Space
- Scientific Computing Helps Researchers Explore the Universe
- Signals from Orbit Provide New Earth Insights
- Russian satellite’s 121-megapixel image of Earth is most detailed yet. https://www.theverge.com/2012/5/12/3016254/russian-satellite-earth-from-space-121-megapixels
- iPhone Photo Sizes: 2007-2018. https://lifeinlofi.com/more/iphone-photo-sizes-2007-2013/
- WiDS 2019 Datathon. https://www.widsconference.org/datathon.html
- National Centers for Environmental Information (NCEI). https://www.ncei.noaa.gov/