Bringing satellite data down to Earth

Jun 12, 2017 | Citizen Science

By Linda See, IIASA Ecosystems Services and Management Program

Satellites have changed the way that we see the world. For more than 40 years, we have had regular images of the Earth’s surface, which have allowed us to monitor deforestation, visualize dramatic changes in urbanization, and comprehensively map the Earth’s surface. Without satellites, our understanding of the impacts that humans are having on the terrestrial ecosystem would be much diminished.

The Sentinel-2 satellite provides high-resolution land-cover data. © ESA/ATG medialab

Over the past decade, many more satellites have been launched, with improvements in how much detail we can see and the frequency at which locations are revisited. This means that we can monitor changes in the landscape more effectively, particularly in areas where optical imagery is used and cloud cover is frequent. Yet perhaps even more important than these technological innovations, one of the most pivotal changes in satellite remote sensing was when NASA opened up free access to Landsat imagery in 2008. As a result, there has been a rapid uptake in the use of the data, and researchers and organizations have produced many new global products based on these data, such as Matt Hansen’s forest cover maps, JRC’s water and global human settlement layers, and global land cover maps (FROM-GLC and GlobeLand30) produced by different groups in China.

Complementing Landsat, the European Space Agency’s (ESA) Sentinel-2 satellites provide even higher spatial and temporal resolution, and once fully operational, coverage of the Earth will be provided every five days. Like NASA, ESA has also made the data freely available. However, the volume of data is much higher, on the order of 1.6 terabytes per day. These data volumes, as well as the need to pre-process the imagery, can pose real problems to new users. Pre-processing can also lead to incredible duplication of effort if done independently by many different organizations around the world. For example, I attended a recent World Cover conference hosted by ESA, and there were many impressive presentations of new applications and products that use these openly available data streams. But most had one thing in common: they all downloaded and processed the imagery before it was used. For large map producers, control over the pre-processing of the imagery might be desirable, but this is a daunting task for novice users wanting to really exploit the data.

  • An example of a Sentinel-2 satellite image showing agricultural fields in Kansas, USA, displayed with natural colors in the Sentinel Playground. ©ESA | Sinergise

  • An example of a Sentinel-2 satellite image showing the status of vegetation for an area around IIASA, Austria, displayed in the Sentinel Playground ©ESA | Sinergise

In order to remove these barriers, we need new ways of providing access to the data that don’t involve downloading and pre-processing every new data point. In some respects this could be similar to the way in which Google and Bing provide access to very high-resolution satellite imagery in a seamless way. But it’s not just about visualization, or Google and Bing would be sufficient for most user needs. Instead it’s about being able to use the underlying spectral information to create derived products on the fly. The Google Earth Engine might provide some of these capabilities, but the learning curve is pretty steep and some programming knowledge is required.

Instead, what we need is an even simpler system like that produced by Sinergise in Slovenia. In collaboration with Amazon Web Services, the Sentinel Hub provides access to all Sentinel-2 data in one place, with many different ways to view the imagery, including derived products such as vegetation status or on-the-fly creation of user-defined indices. Such a system opens up new possibilities for environmental monitoring without the need to have either remote sensing expertise, programming ability, or in-house processing power. An exemplary web application using Sentinel Hub services, the Sentinel Playground, allows users to browse the full global multi-spectral Sentinel-2 archive in matter of seconds.

This is why we have chosen Sentinel Hub to provide data for our LandSense Citizen Observatory, an initiative to harness remote sensing data for land cover monitoring by citizens. We will access a range of services from vegetation monitoring through to land cover change detection and place the power of remote sensing within the grasp of the crowd.

Without these types of innovations, exploitation of the huge volumes of satellite data from Sentinel-2, and other newly emerging sources of satellite data, will remain within the domain of a small group of experts, creating a barrier that restricts many potential applications of the data. Instead we must encourage developments like Sentinel Hub to ensure that satellite remote sensing becomes truly usable by the masses in ways that benefits everyone.

This article gives the views of the author, and not the position of the Nexus blog, nor of the International Institute for Applied Systems Analysis.