Swarms of Small Satellites Set To Deliver Close To Real-Time Imagery of Earth

ananyo writes "A swarm of small satellites set to deliver close to real-time imagery of swathes of the planet is launching today. San Francisco-based Planet Labs, founded in 2010 by three former NASA scientists, is scheduled to launch 28 of its 'Doves' on 9 January. Each toaster-sized device weighs about 5 kilograms and can take images at a resolution of 3–5 metres. Meanwhile Skybox Imaging plans to launch a swarm of 24 satellites, each weighing about 100 kilograms, which will take images of 1 meter resolution or better. Skybox launched its first satellite on 21 November (and captured the first HD video of the world from space) and plans to launch another this year, followed by the remainder between 2015 and 2017. In a first — at least for civilian satellites — Skybox's devices will also stream short segments of near-live high-resolution video footage of the planet. So, too, will UrtheCast, a start-up based in Vancouver, Canada, whose cameras will hitch a ride on the International Space Station. Because the swarms are still to be launched, scientists have yet to fully assess the quality of the imagery. But the satellites' spatial resolutions of 1–5 metres are much higher than those of most scientific satellites. Landsat, NASA's Earth-observation workhorse, for example, has a resolution of 15–100 metres depending on the spectral frequency, with 30 metres in the visible-light range."

Comments

HD

[Russ1642]

I'm sure film cameras have been in space before and they are much higher quality than HD.

Re:HD

[Charliemopps]

No, they'd drop the film via parachute and it got picked up by the navy. The highest resolution they got was 1ft.

http://en.wikipedia.org/wiki/Corona_(satellite)

Re:HD

[deroby]

Using toaster-size satellites, I very much doubt so.

I seem to remember that doing so is impossible using 'normal' optics due simple physics. IIRC there is a limit to the resolution (expressed in radians) you can get for a given frequency for a given lens . Given the distance above our head these things fly this means there is a 'hard' maximum resolution these have and given what I remember from the article it was quite a bit above the ability to read the screen on your phone. The same /article/ (for the love of god I can't remember whether it was in print or some blog or something) stipulated that those pictures you see where they show you the playing-cards a man is holding are actually aerial pictures taken from specialised planes who happen to fly a LOT lower to the ground and also can carry big (heavy) lenses more easily.

PS: I mention 'normal optics' because apparently (same long forgotten source) it should be possible to get much higher resolution by combining different satellites looking at the same target but flying some distance apart and combining their 'view' using some fancy mathematics.

Doing some googling I stumble upon this that seems to conform the above : http://cosmoquest.org/forum/archive/index.php/t-2500.html

Re:HD

[kyrsjo]

The problem isn't the resolution of the sensor chip (5 MPix), but the angular resolution of the optics, which is limited by diffraction (wave physics phenomenon) to some quantity which is dependent on the aperture of the lens. Bigger apperture (diameter of the light-opening) => higher angular resolution.

The other problem is that satelites are quite far up, so you get less spatial resolution (cm on the ground) per unit angular resolution than you would get by being closer (putting the camera on a plane).

The diffraction limit puts a hard limit on the achievable resolution no matter the quality of optics and sensor, and if you want to increase that limit, you need a bigger diameter telescope. This is why people are talking about "2 meter" and "5 meter" etc. mirrors on astronomical telescopes.

The father of this post mentions that it is possible to get around this problem by using more than one satelite (or the same satelite at two different points in its orbit), effectively creating a much larger aperture. This is called syntetic aperture, and for this to work you need to record both the amplitude and phase of the incoming wave. It works great for radar applications as the wave used here is slow enough to be followed by electronics, but visible light is just way to fast and we can only record incident power (amplitude^2) and thus cannot do an off-line computation of the interference patterns wanted.

Apples and Oranges

[Remus Shepherd]

I work on the Landsat program. The article pulls Landsat out as an example of mid-resolution satellites, but it's really an apples-to-oranges comparison. Landsat 8 has 11 spectral bands, including thermal IR, a Cirrus band, a coastal aerosol band, and so on. All of these are used for scientific purposes. The Dove and Skysat instruments have 3 or 4 bands, just enough to get an RGB picture and maybe some chlorophyll distinction for agriculture producers.

Landsat is used to study land cover change, find new resources, map fire scars, and other applications that require precision and data depth; the swarm satellites will be used to make maps and that's about it. Both are important, but comparing one to the other is like comparing a smart car to a grain combine. They're used for totally different purposes.

Real time.

[DarthVain]

Um No. As someone who has worked with remote sensing imagery the resolution is not all that good. Landsat is ancient and there are loads of other high resolution options. The resolutions these "swarms" have are not particularity good.

The interesting part is the in "real time" bit.

However in that I am even doubtful unless they are using very large values for "real" time.

Anyone that has worked with this kind of data will tell you A) it is usually HUGE, and B) marginally compresses. Data has to be sent from satellite to ground. That means transmission. At what speed? Unless they have discovered a way of sending data magically faster than the rest of the world, it is still constrained by that. In addition, most of the time these sorts of images need to be processed, and with the volume of data we are talking about, even machines with a lot of processing power can be pushed to the limit, doing small chunks for days.

That said, I would be really excited if it really worked. This would allow for all sorts of scientific observations, and resource management, and a host of other things. (if only in small limited areas which are predetermined by orbit)

I get a kick out of all the TV and movies like Enemy of the State depicting satellite tracking and zooming etc... Enhance! Sure it is possible that some secret agency somewhere has some magic technology that does this. Then again has your encounters with any other branch of government given you the idea that this might be a possibility?

Maybe the real story here is that that satellites are really cheap to build and launch, meaning that we may have access to half decent (1-5m resolutions) coverage over most of the globe soon most of the time. Still processing may be a problem. So rather than a very poor update measured in years, you might get something better is much less time. Maybe. I presume these are commercial, and will have to make a profit off these somehow. Selling the data to Google Maps perhaps?