Dutch innovation makes processing satellite data easier and more efficient
Working with Earth observation data is more accessible, faster, and more efficient starting this month thanks to Ellipsis Map Engine. The innovative processing software was developed by the Dutch company Ellipsis Drive, with support from the Netherlands Space Agency.
Computers were invented to solve complex calculations. For a few years now, they have also been able to handle language well. Large language models, such as ChatGPT, provide answers to all your questions. But what do you do with a question that cannot be answered based on numbers or language? A question for which you need to analyze spatial data, such as observations from Earth observation satellites?
‘We have taken this challenge upon ourselves,’ says Rosalie van der Maas, co-founder and director of Ellipsis Drive. ‘We have developed processing software specifically intended to analyze raster data, a data type that had been overlooked by big tech until now.’
‘Calculating’ with images
Earth observation satellites collect pixels containing information about an exact location on Earth. Information about, for example, altitude, temperature, or land cover. If you want to work with this ‘raster data,’ you must first convert it into information that processing software can handle, and once the analysis is done, back into raster data.
‘In Earth observation projects, eighty percent of the time is currently lost to data processing,’ says Van der Maas. ‘Ellipsis Map Engine is the first processing software that operates ‘map natively’: you put an image into the computer and an image comes out that you can further work with. Under the hood, everything is automated, leaving you as a user with much more time to actually add value with Earth observation data.’
Dividing images
One of the biggest challenges Ellipsis Drive had to overcome was distributed computing. Suppose you want to know where trees provide shade throughout the Netherlands. You would have to search through an enormous number of images for trees and shade—too many for a single computer. The solution would then be to have several computers work on this task simultaneously. But then a problem arises: where do you cut the images into parts without the computers losing track of what they are seeing? And how do you then piece everything back together?
The trick is to provide each computer with sufficient information about the context of the pixels they are working with. So, no hard cut, but rather an overlap area. That is one of the things Ellipsis Map Engine does fully automatically. The processing software makes working with raster data flexible, scalable, and possible for anyone who masters the Python programming language.
Supported by InCubed
Over the past five months, HERE Technologies, Rabobank, NEO, and S[&]T, among others, tested a beta version of Ellipsis Map Engine. The European Space Agency used the engine to make Earth observation data available to emergency services more quickly during natural disasters. Starting this month, Ellipsis Map Engine is available to anyone who wants to experiment with it, with support from Ellipsis Drive.
Ellipsis Drive developed its Map Engine with support from the Netherlands Space Agency under ESA’s InCubed programme. This programme supports companies that develop innovative and commercially promising products and services using Earth observation data. ‘We are a relatively small company with fifteen people, ten of whom are in the Netherlands. This grant enabled us to develop much faster than expected,’ says Van der Maas.
She is pleased that it worked out. ‘We are not a space company, but an infrastructure provider that supports space companies. It took some convincing to show that working with Earth observation data becomes much more economical with our engine. Now that Ellipsis Map Engine is on the market, everyone can experience that.’
