Spy satellites for astronomy.
Robert Clark
"Featuring the latest in charged-coupled device (CCD) imaging
technology, the original KH-11's had a primary mirror 7 feet 8 inches
(2.33 meter) wide, although more recent models feature larger mirrors.
The secondary mirror, more than one foot in diameter, sharply focuses
the light coming off the primary mirror to produce images with a
resolution of 10 centimeters (3.93 inches).
"The latest KEYHOLE's reportedly feature a resolution of just 3
centimeters. Putting it simply, if you were to place a matchbox in the
middle of a suburban street a KH-11 would be able to photograph it."
http://www.space.com/SpaceReportersNetworkAstronomyDiscoveries/Carter_Spysats_022202.html
Now that's what I'm talking about!
Bob Clark
Martin Hardcastle
Robert Clark <rgrego...@yahoo.com> wrote:
>"The latest KEYHOLE's reportedly feature a resolution of just 3
>centimeters. Putting it simply, if you were to place a matchbox in the
>middle of a suburban street a KH-11 would be able to photograph it."
>http://www.space.com/SpaceReportersNetworkAstronomyDiscoveries/Carter_Spysats_022202.html
Clever trick, given that telescopes operating in the other direction
are limited by the atmosphere to a resolution of ~ 0.5 arcsec. If
these things are in low earth orbit they can't do much better than 1
meter resolution on the ground, surely?
Martin
--
Martin Hardcastle Department of Physics, University of Bristol
A little learning is a dangerous thing; / Drink deep, or taste not the
Pierian spring; / There shallow draughts intoxicate the brain ...
Please replace the xxx.xxx.xxx in the header with bristol.ac.uk to mail me
William C. Keel
> In article <832ea96d.02030...@posting.google.com>,
> Robert Clark <rgrego...@yahoo.com> wrote:
>>"The latest KEYHOLE's reportedly feature a resolution of just 3
>>centimeters. Putting it simply, if you were to place a matchbox in the
>>middle of a suburban street a KH-11 would be able to photograph it."
>>http://www.space.com/SpaceReportersNetworkAstronomyDiscoveries/Carter_Spysats_022202.html
> Clever trick, given that telescopes operating in the other direction
> are limited by the atmosphere to a resolution of ~ 0.5 arcsec. If
> these things are in low earth orbit they can't do much better than 1
> meter resolution on the ground, surely?
Though to be fair, these were the folks who could afford to put
serious money into adaptive optics first, and can also afford really
fast computers for post-processing down stream. When doing some
tests of algorithms for deconvolving HST images in the Bad Old
Days, I got a phone call from an Army contractor in Huntsville.
He wanted to find out everything I knew and spill nothing that he
knew... I didi end up convinced that they cxan play a bit faster and
looserthan astronomers can in superresolution games, since their
applications usually don't need precise radiometry or few-cm
location precision.
Statements such as this come out a lot, and nobody who knows is
likely to confirm or deny.
Bill Keel
The Small Kahuna
I don't think you thought this through, Martin. Astronomical telescopes
deal with an exceedingly low light flux and so use long exposure times
to form a decent image. During this long exposure, the atmosphere
changes considerably and in unknown and unpredictable ways. The density
variations change the index of refraction and this varying index of
refraction bends the image. The result is a smearing of the image and
as you point out, it is this smearing that limits the angular
resolution, not the optics, per se.
Now consider a KeyHole satellite. It is taking a picture in broad
daylight with a telescope with a rather large effective 'f' stop
(considering the magnification). This means that the exposure times can
be "short". A short exposure time means that the atmosphere does not
move much during the actual exposure, even if it *does* actually distort
the image. However, to the military intelligence people, this
distortion is unimportant as it does not smear the image, only offset
objects relative to each other (as well as vary the effective
magnification).
So the increased clarity is not because the effective resolution is so
much better than ground based telescopes as it is because the subject
has a very high light flux and so the exposure times are short so the
resolution is truly a function of the optics and not the atmosphere.
I'm surprised you make the assertion you do because the performance of
the KeyHole satellites are not a very well protected secret. Since they
really do work that well, there must be a rational explanation...
Dan McKenna
"William C. Keel" wrote:
> > are limited by the atmosphere to a resolution of ~ 0.5 arcsec. If
> > these things are in low earth orbit they can't do much better than 1
> > meter resolution on the ground, surely?
>
The down looking seeing is near the object and causes less distortion
than the ground based (up looking ) case.
>
> Though to be fair, these were the folks who could afford to put
> serious money into adaptive optics first, and can also afford really
> fast computers for post-processing down stream. When doing some
> tests of algorithms for deconvolving HST images in the Bad Old
> Days, I got a phone call from an Army contractor in Huntsville.
> He wanted to find out everything I knew and spill nothing that he
> knew... I didi end up convinced that they cxan play a bit faster and
> looserthan astronomers can in superresolution games, since their
> applications usually don't need precise radiometry or few-cm
> location precision.
>
> Statements such as this come out a lot, and nobody who knows is
> likely to confirm or deny.
>
I was at the Maui site and they removed the AO system and
replaced it with a speckle camera.
I was told that it looked cleaner (the setup on the telescope)
and that the big SGI box did a good job of real time reconstruction.
Most AO high order systems especially D.O.D. systems are not good for astronomy because they
don't have a stable PSF.
Dan
>
> Bill Keel
Henry Spencer
Martin Hardcastle <m.hard...@xxx.xxx.xxx> wrote:
>>"The latest KEYHOLE's reportedly feature a resolution of just 3
>
>Clever trick, given that telescopes operating in the other direction
>are limited by the atmosphere to a resolution of ~ 0.5 arcsec. If
>these things are in low earth orbit they can't do much better than 1
>meter resolution on the ground, surely?
From 300km, 0.5 arcsec is about 70cm. But bear in mind the "shower
curtain effect": much of the atmospheric turbulence is located quite
close to the ground, which means that it hurts ground-based observers
looking up much worse than it hurts orbital observers looking down.
--
Many things changed on Sept. 11, but the | Henry Spencer he...@spsystems.net
importance of freedom did not. -SpaceNews| (aka he...@zoo.toronto.edu)
Chris Franks
"Henry Spencer" <he...@spsystems.net> wrote
>
> From 300km, 0.5 arcsec is about 70cm. But bear in mind the "shower
> curtain effect": much of the atmospheric turbulence is located quite
> close to the ground, which means that it hurts ground-based observers
> looking up much worse than it hurts orbital observers looking down.
This is often called the "vellum" effect, which we noticed in
engineering
school when making mechanical drawings.
David M. Palmer
<m.hard...@xxx.xxx.xxx> wrote:
> In article <832ea96d.02030...@posting.google.com>,
> Robert Clark <rgrego...@yahoo.com> wrote:
> >"The latest KEYHOLE's reportedly feature a resolution of just 3
> >centimeters.
> Clever trick, given that telescopes operating in the other direction
> are limited by the atmosphere to a resolution of ~ 0.5 arcsec. If
> these things are in low earth orbit they can't do much better than 1
> meter resolution on the ground, surely?
If the scattering is 0.5 arcsec = 2.4 microradians, and the scattering
occurs at a characteristic altitude of 10 km, then that causes 2.4 cm
(ground measure) of scattering.
--
David M. Palmer dmpa...@email.com (formerly @clark.net, @ematic.com)
Tamas Feher
It was claimed that the limit allowed by laws of physics is 5 centimeters.
Sincerely: Tamas Feher.
Martin Hardcastle
I didn't, but I'm not sure I agree with your reasons why not...
[snip]
>Now consider a KeyHole satellite. It is taking a picture in broad
>daylight with a telescope with a rather large effective 'f' stop
>(considering the magnification). This means that the exposure times can
>be "short". A short exposure time means that the atmosphere does not
>move much during the actual exposure, even if it *does* actually distort
>the image.
The timescales involved for effective AO (again, from the ground) are
of the order of milliseconds -- these are the timescales required to
track variations in the atmosphere, or equivalently the timescales on
which you can ignore those variations. So, back of the envelope
suggests that you're in a fairly photon-limited regime at 500 km
altitude on these timescales for objects illuminated by sunlight. You
need, at the minimum, to make a very large number of images and do
some clever post-processing to reliably detect and image objects of
centimeter sizes. As Bill Keel pointed out, the military are exactly
the guys who have the resources to throw at this sort of problem, so I
don't deny it can be done, though I'd be surprised if it were done
routinely. (It's a harder job than normal AO because you aren't likely
to have a convenient point source to hand.)
The real bogus thing about the number I quoted was my failure to think
about the altitude of the atmospheric stuff, as several people pointed
out. That makes it wrong by a large factor, though of course the
figure of 0.5 arcsec I used in the first place is unrealistically good
for normal conditions, getting me a small factor back.
>I'm surprised you make the assertion you do because the performance of
>the KeyHole satellites are not a very well protected secret. Since they
>really do work that well, there must be a rational explanation...
I can think of a group of people in whose interest it is that the
world at large believe these numbers. That naturally makes me
curious about their accuracy, in the absence of actual data.
Allen Thomson
[snippage]
> The timescales involved for effective AO (again, from the ground) are
> of the order of milliseconds -- these are the timescales required to
> track variations in the atmosphere, or equivalently the timescales on
> which you can ignore those variations.
There's also the spatial scale of the variations to consider --
those (call them "cells") tend to be in the range of tens of
centimeters, give or take a factor of ten or so. Looking up,
an astronomical telescope with a typically narrow field of view
only needs to cope with a relative few of these cells every few
milliseconds. Looking down, you would have to measure and compensate
for all the cells over the entire field of view every few
milliseconds. Say you're looking at a scene 1 km on a side and
the cell size is 30 cm; that's 1.1e7 cells. Somehow measure those
once every ten milliseconds with, say six-bit accuracy, and that's
6.7 gigabits per second that you have to process against the actual
image stream.
And I'm not at all sure how you'd measure the index of refraction
fluctuations in the first place. Maybe some sort of laser probing
technique?
So I don't think there's much chance that adaptive optics or
post-processing is being used to compensate for the effects of
"seeing" on spysat pictures. That having been said, adaptive optics
to maintain the figure of a thin or segmented mirror seem perfectly
feasible, but that's a completely different application.
(I also think that there's nothing keeping the current crop of
satellites with mirrors in the reported 2 to 3 meter range from
getting essentially diffraction limited images of the ground.)