digitalglobe

How do you Move a Vast Imagery Archive?

May 15, 2017

Satellite imaging company DigitalGlobe recently wrote an interesting blog post about their vast archive of imagery and how they are moving it to the cloud.

Some interesting statistics from the post:
One image from WorldView-3 can be 30 gigabytes.
The DigitalGlobe archive is around 100 petabytes.
Their archive increases by 10 petabytes per year.

Comparing this to what we learned about Planet’s archive earlier this year, Planet has an archive of 7 petabytes. Before Planet launched the latest batch of 88 satellites and purchased Terra Bella from Google they were adding 7 terabytes daily or 2.5 petabytes per year to their archive. We estimate that they are now adding about two and a half times that daily or over 6 petabytes per year.

So, although Planet has the world’s largest imaging satellite fleet (actually, it is the world’s largest commercially-operated fleet of any type of satellite), their volume of imagery being gathered in terms of data storage required appears to be lower than DigitalGlobe’s. This is probably because DigitalGlobe’s satellites are higher resolution and thus produce much more data per area covered. It is also probable that DigitalGlobe’s satellites gather more optical wavelength bands.

DigitalGlobe have, until now, kept most of their archive in-house in a vast tape-based storage system. However, they are now moving to the cloud (Amazon Web Services). Rather than attempt to upload 100 petabytes to the internet, Amazon sent them a whole shipping container, a solution known as ‘Snowmobile’, complete with built in data storage, which DigitalGlobe then copied their data to and sent it back to Amazon. Interestingly, according to Amazon, 100 petabytes is the upper limit of their ‘Snowmobile’ solution (per truck).

We don’t know the exact size of the Google Earth database, but we estimate it to be several petabytes at least. So if you ever needed an offline version, then you would probably need a large suitcase-sized container at least.

The DigitalGlobe blog post features one of their images showing ‘Snowmobile’ outside DigitalGlobe’s headquarters but that image has not made it into Google Earth. However, we did manage to spot it in Street View:

Ground based imagery looking at an archive of satellite imagery.

DigitalGlobe is helping to combat slavery at sea

February 24, 2017

DigitalGlobe recently released this article which highlights the use, by the Associated Press, of high resolution imagery from DigitalGlobe’s WorldView-3 satellite to combat slavery on fishing ships. You can read various Associated Press stories on slavery here and a story that specifically involved DigitalGlobe imagery here. The imagery is used to not only spot suspicious behaviour, but also identify ships from their masts and other distinctive features.

DigitalGlobe correctly claims that it is currently the world leader in resolution for commercial satellite imagery. It is likely that some countries have spy satellites with better resolution. Although the article mentions WorldView-3, DigitalGlobe also has WorldView-4, which was launched late last year and has similar resolution. WorldView-4 only went into commercial use this month.

We have previously looked at Global Fishing Watch, which monitors fishing ship locations and uses that information to identify illegal fishing. Maybe they too could help identify behaviour indicative of slavery on fishing ships. Apparently fishing ships carrying slaves spend most of their time at sea and offload their catch to other ships rather than returning to port.

Using artificial intelligence on satellite imagery

January 17, 2017

We recently came across this interesting article which talks about how DigitalGlobe is applying AI to satellite imagery recognition.

Humans brains are still extremely useful, so DigitalGlobe also has a crowd-sourcing platform that can be used either on its own, or to train an artificial intelligence algorithm.

If you are interested in AI and satellite imagery then read the DigitalGlobe developer blog, which has a number of interesting articles, such as:

Detecting and measuring coastal change
Finding pools
Monitoring changes along pipeline routes
Detecting population centers in Nigeria

Detecting population centers in Nigeria. [Image: DigitalGlobe]

You can also sign up for a free evaluation account on DigitalGlobe’s AI platform GBDX.

Incidentally, we also found this comment in the article mentioned in the first paragraph above to be interesting:

DigitalGlobe also doesn’t release images of active U.S. combat areas.”

This at least partially explains the censorship (lack of imagery updates) of certain countries in Google Earth. It is not clear whether this is a decision made unilaterally by DigitalGlobe.

WorldView-4 and SkySat launches

September 15, 2016

The next couple of days will see two significant launches for satellite imaging. If all goes well, the first launch will be four SkySat satellites owned by Google’s Terra Bella. They are expected to launch with a Peruvian reconnaissance satellite aboard a European Vega rocket. See the count down clock and more launch details here. The launch takes place from ZLV, Kourou, French Guiana. Next will be WorldView-4. It is a DigitalGlobe satellite with similar specifications to WorldView-3, the current leader in high resolution commercial satellite imagery. According to spaceflightnow.com the launch is scheduled to take place from SLC-3E, Vandenberg Air Force Base, California.

[ Update: We didn’t realise at the time of writing that PerúSAT-1 which is being launched together with the SkySat satellites is also an Earth observation satellite with a resolution of 70 cm per pixel panchromatic and 2 m per pixel for colour. Learn more about it here]

We already looked at WorldView-4 last month, so today we are focusing on the SkySat satellites. Those being launched today are SkySats 4 through 7. SkySat 1 was launched in November 2013, SkySat-2 in July 2014 and SkySat-3 in June 2016. We have seen imagery from them a number of times, including imagery of the damage caused by Italy’s recent earthquake, a poppy display at the Tower of London and the Burning Man festival. We also once came across a SkySat image in the Sahara, which has since disappeared from Google Earth.

The SkySat satellites have an imagery resolution of about 90 cm per pixel. This is not as good as WorldView-4’s 30 cm per pixel, but is better than Planet Lab’s Dove satellites, which have a resolution of 3-5 m per pixel. It is also better resolution than the best imagery Google Earth currently has for some locations, so we hope Google considers using Terra Bella imagery to fill in the gaps in Google Earth.

Unlike SkySats 1 and 2, SkySat-3 has propulsion, which gives it greater flexibility in capturing images. Presumably 4 thorough 7 also have propulsion. Read more about the differences between SkySat’s 1 and 2 and SkySat-3 here.

The SkySat satellites being prepared for launch as tweeted by CNES. Image credit ESA-CNES-ARIANESPACE.

Can we see Antarctic seals in Google Earth?

August 4, 2016

We recently came across this interesting article on the DigitalGlobe blog. It is about using imagery crowd-sourcing site Tomnod to count seals in DigitalGlobe satellite imagery of Antarctica. The very best commercial satellite imagery available today has a resolution of about 30 cm per pixel, with most imagery that we call ‘high resolution’ satellite imagery closer to 50 cm per pixel. This usually means that animals are only a few pixels in an image and difficult or impossible to spot. In the cases where we have seen animals in Google Earth it has typically been in aerial imagery, such as the National Geographic Africa Megaflyover images or areas like the US which have a lot of aerial imagery. And there is, of course, Street View.

Antarctica has the advantage that we are looking for dark spots on a fairly featureless white background, which makes spotting seals possible. Much of Antarctica doesn’t have high resolution satellite imagery but there are some areas that have DigitalGlobe strips of imagery. So, given the tips in the DigitalGlobe blog post, we wondered whether we could find any seals in the imagery.

We aren’t certain if these dark spots are seals, but they do match the description given.

This is rather low resolution imagery but it does look like there might be groups of seals gathered around this crack in the ice.

We are fairly sure this is a positive find and that those are seals.

For the locations above, download this KML file.

Despite looking rather white and boring from a distance, it is well worth zooming in on Antarctica and looking around, as there is a remarkable variety of landscapes formed by glaciers and shifting sea ice. For best results switch to ‘historical imagery’ as it is much easier to find the high resolution patches that way.

We also tried looking for seals with the Tomnod website, but were unsuccessful. Google Earth is a much better interface for this sort of thing as it is much easier to find major features, such as cracks in the ice, then zoom in to them. The actual survey the scientists want, however, has to be done on Tomnod, because Google Earth does not have access to all the imagery that they want searched.