Artist’s rendition of satellites orbiting the Earth – rottenman/123RF Stock Photo
This week we’re looking at the latest Earth Observation (EO) launches both in terms of satellites and machine learning datasets, together with the discovery of a lost satellite!
WorldView-Legion EO Satellites Finally Launch!
We were a little premature when we talked about the impending launch of Maxar’s WorldView Legion-1, and -2 EO satellites a couple of weeks ago. We were even more premature when we sent out a planned Tweet (or whatever they are called now) saying it had launched, when in fact the launched had been pulled – oops, the problem of automating Tweets a few days ahead! We can rely on our knowledgeable audience to make sure we don’t get away with such errors, and we didn’t! We apologise for our error.
We can now confirm that SpaceX’s Falcon 9 rocket successfully launched the pair of satellites on May 2nd from Vandenberg Space Force Base in California. Details of the satellites can be found in a previous blog. SpaceX is also expected to launch the remaining four satellites in this six strong constellation later this year.
New Training Datasets Supporting Machine Learning
At the start of the month, Satellogic released a large open dataset of high-resolution imagery curated from their catalogue to support the training of machine learning and artificial intelligence (AI) models in EO. It included around 6 million images from around the world at a sub-meter spatial resolution. The dataset has been released under a Creative Commons CC-BY 4.0 license, allowing for commercial use of the data with attribution, and access to the dataset can be found here.
Machine Learning training datasets is something we are passionate about – we’ve covered another dataset release previously – as it is so critical for the development of AI models. Over the last couple of years we’ve been working with the Open Geospatial Consortium on the topic, and last year we co-wrote the Testbed-18: Machine Learning Training Data Engineering Report to develop a foundation for future standardization of Training Datasets (TDS) for Earth Observation (EO) application. We were also part of the team who wrote the OGC Training Data Markup Language for Artificial Intelligence (TrainingDML-AI) Part 1: Conceptual Model Standard that was published eight months ago.
It is exciting to see this area of EO develop, and we’re looking forward to using Satellogic’s dataset in future work, and continuing to support this ongoing activity in a variety of ways.
Lost Satellite Rediscovered!
It’s not very often a lost satellite is rediscovered, but that is exactly what happened last month. After hiding in space for a quarter of a century, it’s reported that the Infra-Red Calibration Balloon (S73-7) has been found again using tracking data from the U.S. Space Force.
It was part of the US Air Force Space Test Program, and was deployed on the 10th April 1974 from the KH-9 Hexagon satellite into a 500 mile circular orbit. After launch, it was expected to inflate and operate as a calibration target for remote sensing equipment, however the inflation failed during deployment. The satellite has been a bit of a challenge for ground tracking stations over the years, disappearing in the 1970s and again in the 1990s, and has been unseen for the last twenty-five years before it was rediscovered last month. Part of the problem with tracking, is that as this is a balloon and so it is more difficult to track as, unlike metal, it does not show up as well on the radar.
Summary
As more and more satellites are launched, its important that we keep track of them to ensure that space remains space to operate in – although the journey of the Infra-Red Calibration Balloon shows how difficult this can be sometimes! It’s great that it has been rediscovered, and will be interesting to see whether it will do its disappearing act again in the future!
