International Archives of Photogrammetry and Remote Sensing. Vol. XXXII Part 7C2, UNISPACE III, Vienna, 1999
141
VMS PACE III- 1SPRS/NASA Seminar on
“Environment and Remote Sensing for Sustainable Development”
9:00 am -12:00 pm, 23 July 1999, VIC Room A
Vienna, Austria
In addition to the commercial ventures which are investing
hundreds of millions of dollars governments around the
world are spending collectively billions a year developing
and operating national and civil satellites. In aggregate, the
global remote sensing market, is in the range of multi-billions
annually.
It is important to note, however, that only a few commercial
sensor platforms are expected over the next decade while the
pyramid expands rapidly due to the multiple markets that can
use the same base information products. Information has the
unique attribute of being expandable and non-linear (i.e.
single source feeds multiple uses). In a digital form it is easy,
quick, and increasingly affordable to transport information in
electronic or media form.
If the COPUOS is to produce meaningful direction for the
remote sensing industry, it must do so recognizing the
information industry trends in relation to the remote sensing
industry. The next sections take a quick look at some of the
most significant trends as they affect the development of the
commercial remote sensing business.
At what rate is Earth information “content” being
created?
Today, a single satellite like IKONOS (Figure 1) is capable
of collecting Earth information at a staggering rate of over
100 million (mega) pixels per second. At this rate IKONOS
is capable of filling a 10-15 terabyte digital archive in less
than three montlis. For reference: the U.S. Library of
Congress, with its 10 million volumes of books, if digitized,
would be storable in that archive, a space today occupying
less than 10 cubic meters. For IKONOS, imagery at 1 meter,
the habitable world is about 100 tera pixels. This imagery,
compressed with a nearly lossless algorithm, can be stored in
two such mass storage units referenced above, i.e. about 25-
30 terabytes. Equally dramatic and significant is the
realization that since the 1960s the rate of remote sensing
data collection and transmission has increased by about a
factor of 10 every decade.
The ITU at WARC ’97 helped ensure tins growth by taking a
positive step in authorizing the 25.5 - 27.0 Ghz spectrum for
future use by next generation remote sensing systems. The
ITU is to be applauded for its forward planning action at
WARC ’97. By 2010 it is reasonable to assume data rate
increases of 5 to 10 times (through a combination of
bandwidth efficient modulation techniques and greater
bandwidth allocations). Such rates will result in reducing the
time to fill the 10-15 terabyte archive from about 10-12
weeks to once a week for a single satellite. When all satellite
sources are considered, it is clear that a deluge of
data/information is occurring.
The truly demanding challenge becomes one of applying
information technology to service the demand side, i.e.. the
base of the information “pyramid” versus aerospace
technology which promotes the supply-side, vertex, of the
pyramid. Limited by information “exploitation” systems and
technology, the remote sensing industry advancement, both
nationally and commercially, will be limited if such
processing, exploitation and information management
systems are not rapidly developed. The promotion of user
application tools, techniques, and technologies is where the
UN can take a leadership position in advancing the concept
of Program Information Applications (PIA) in developing
conn tries.
What cost to store, process, and disseminate information?
A key economic factor in the commercialization of remote
sensing is the continuing and rapid decline in the cost of
storage and the computer processors. Today, it is possible to
store data at a cost approaching a $1 per gigabyte. Processors
running at 400-500 Mhz are in the range of $1-2,000: even
high end pixel processing intensive computers are available
as workstations, desktop “appliances”, for under $5,000.
Soon, Gliz speed PCs will be available and will compete
with these “high-end", $5,000 workstations. This computer
industry trend has been characterized by Moore’s “Law”
vvliich forecasts a redoubling of performance to price every
18-24 montlis. This trend has been consistent since the days
of UNISPACE II.
Similarly, data transmission costs have been dropping rapidly
over the decades as both electronic dissemination costs and
media (tape, disks) costs have dropped from $l,000’s per
gigabyte in the 1970s Lo today about $1 - 10 per gigabyte for
media recordings and surface “transmission” costs. When
wide-band, bandwidth on demand, telecommunication
systems arrive and transmission costs approach $1-10 per
gigabyte, more and more remote sensing data will move
electronically via terrestrial (fiber) and space communication
systems.
The entire world will continue to benefit from such
advancement if proper planning, promotion, and expenditures
are encouraged and funded. Here is where data fonnats and
standards are most useful - at the back-end of an open
information system. The UN should promote such back-end
standardization, and application specific hardware, software,
and analytic information tools. Information manufacturing
will be a specialty trade leveraging aerospace and high end
computer processing teclinology while information
engineering will leverage the new domain of information
sciences and technologies to address the end users’ needs.
Cost of goods/price of goods are critically
dependent on the resolution of the information.
Some new thinking is required here.
Now we come to the interesting matter of cost. If the “pixel”
is envisioned as the information atom, then as the atom size
gets smaller, the cost of producing the smaller atom
increases. This is actually a very basic physics concept that
drives the system economics and for which a “cold fusion”
solution has not been discovered nor likely to be soon. To
first order then, the cost/price increases as the square area of
the pixel decreases; not quite as a 1/square pixel area, but not
too far off this relation. As better technology is available, the
cost per unit area at increasing resolution will go down just
like the cost of goods of communication bandwidth. For most
remote sensing (space and aerial), the critical metric is the
cost per megapixel; in communication systems it is cost per
megahertz (bits). For 1 meter imagery the cost is in the range
of $25-75 per square kilometer or, since a square kilometer
equals a megapixel, (at 1 meter pixel size), $25-75 per
megapixel. Since compression of pixel to bits relates
