Space

Subject: Re: New article on CORONA history
From: aufsj@imap2.asu.edu
Newsgroups: sci.space.history,sci.space.policy,alt.politics.org.cia
Date: 6 Feb 1997 09:09:26 GMT
Organization: Arizona State University
Lines: 101
Message-ID: <5dc746$lgp@news.asu.edu>
References: 
[deletions]
:       "We approached these questions by preparing simulated 
:    satellite photographs from high-resolution U-2 coverage and 
:    giving them to the photointerpreters to see how recognition 
:    varied with resolution.  The experiments confirmed my impression 
:    that a substantial improvement in resolution was needed.  The 
:    Drell group judged it unlikely that we could push the Corona 
:    system to that new level by further improvements. Corona's basic
:    design had inherent limits, and we had reached them."
:     In any event, it looks as if the basic Corona design should 
: have been improvable down to about a meter ground resolved 
: distance, or NIIRS 5 in today's parlance.  This is a peculiar 
: result in light of the first question put to the Drell panel:  
: "what resolution do the photointerpreters need to find and 
: identify strategic installations in broad area coverage" and the 
: answer, "The experiments confirmed my impression that a 
: substantial improvement in resolution was needed." 
:     If you look at almost any task-vs-resolution table, it's 
: clear that one or even two meter resolution is more than adequate 
: for "finding and identifying strategic installations."  What he 
: should have been asking was either "how can we increase area 
: coverage while keeping enough resolution to discover strategic 
: installations?" or "how can we increase resolution to do 
: technical intelligence and other fine-scale photointerpretation 
: tasks?"  These are different and somewhat antithetic questions. 
	It seems to me that he may have been talking about a somewhat 
different but very key issue: The tradeoffs between area coverage and 
analytical ability with reference to limited operator assets. Back in 
those days there was not much in the way of computerized image software, 
indeed at first most of the data was not even digitized and stored on 
computers or such media, it was just film. This, of course, meant that 
people had to sit down and physically look at the film to get data, and 
with thh satellite taking pictures of thousands of square miles, one 
could quickly run into the problem of more photos than people to look at 
them.
	A complicating factor is that resolutions of tens of meters often 
*aren't* of good enough resolution to distinguish facilities of 
"strategic" importance. Major space launch facilities and many of the 
fixed air defense sites had prominent geographic patterns to be certain, 
but many (I would say most) places of interest did not.  For example, you 
get imagery that shows a building with cars parked outside. It could be 
about anything. But get higher resolution and you can see if they are zil 
limos, military issue, or just plain "cars".  Many factories look pretty 
much the same at 10 meter resolution, but with 1 meter you can see what 
the gear on the loading docks is--military or just general issue?  Yet 
another very common problem from the 60s (and even sometimes today), just 
*exactly* what kind of antenna is on that building? And what are its 
pointing angles? 
	In other words, the more detail the more likely you are to be 
able to accurately analyze the scene, but the more detail (i.e. the 
higher resolution) the more film you generate, often to the point that 
you can't even look at all of it.  The thrust was to find some adequate 
"middle ground", enough detail to discern targets of interest from just 
plain 'stuff', and yet low enough resolution so that you could hope to be 
able to look at it all in some systematic manner.
	I recall reading a note on the "open skies" proposal along the 
same lines. Some general was pointing out that the proposal was made with 
the assumption that we could use the U-2 at 90,000+ feet and take swaths 
XX miles wide. If the Soviets 'agreed in principle', and then turned 
around and said that the US aircraft could fly no higher than 5,000 feet, 
the effective swaths could only be YY miles wide and so much film would 
be generated that we couldn't analyze it all properly. "Open Skies" never 
went anywhere (back then) but it is an interesing example of one of the 
conundrums faced in intel, as well as an early example of why arms 
control and other treaty teams need not just political scientists but 
real ( :-) ) ones as well.
	The migration to "close look" and "area look" satellites didn't 
entirely solve this problem. If the area data looked interesting then one 
could schedule a look with high-res. But if the original problem is that 
the area data *isn't* indicative of high interest activity, then what?  
The answer was catalogue work and systematic work. People often focus on 
taking pictures of crisis and superfacilities, but most of the work was 
drudgery (though to some very interesting drudgery).  Areas would be 
systematically imaged and people would spend countless hours looking for 
anything of interest. The problem was that this was not very fast at all, 
if it took literally years to look over even a portion of the USSR (let 
alone the world) then important stuff might slip by. This is where 
'cueing' from other intelligence sources becomes vital. For example, the 
big Soviet ABM radar site out in the middle of nowhere was photoed in the 
late sixties only after the Navy located its emissions.
	I think that the real revolution in *modern* imagery is not the 
resolution of the photos or even the speed with which pictures can be put 
on a persons computer screen. Rather, the most important advances are the 
ability of computers to do searches and analysis, even if only of a 
limited capability, against huge volumes of data. Just as the first 
computerized translation and speech recognition computers revolutionized 
the communication side of SIGINT starting back in the 60s, modern 
capabilites are starting to revolutionize imagery.
regards,
-----------------------------------------------------------------------
Steven J Forsberg  at  aufsj@imap2.asu.edu           Wizard 87-01