|
In an earlier
issue of this magazine ( Astronomical
Ballooning) I described some of the 'behind-the-sceens'
activities that take place during the launch of a balloon-borne,
infrared telescope. In this issue, we are going to have a
look at how the data we get from such a flight is analyzed.
There are basically
two questions that have to be answered with the data; "Where
is the far-infrared emission coming from?" and "How
bright was it in the direction we were looking?". As
I mentioned in the last article, a typical flight produces
as much information as a full set of the Encyclopedia Britanica,
but the information is stored in long strings of 1s and 0s;
the basic alphabet of the computer language. To translate
the computer 'words' into recognizable maps of an infrared
source, a carefully thought-out series of steps have to be
followed, culminating in maps of each object in the survey.
These maps form the backbone of whatever theory we wish to
propose to explain what we are seeing. For example, by knowing
that most of the infrared emission comes from regions near
the edge of a dense molecular cloud, we might be able to say
something about where luminous stars are born in these interstellar
clouds; "Do they favor the surfaces of these enormous,
interstellar clouds or their deep interiors where the density
of gas is highest?"
WHERE
DID WE LOOK?
The 30 feet
of flight film taken by a 35mm camera bolted to the side of
the telescope, tells us where our detectors, our infrared
'eyes', were looking during the entire, 10-hour flight. As
the telescope is made to sweep across a particular part of
the sky, bright stars travel across the field of the camera
lens and leave behind trailed images on the film. A rotating
shutter segments these trails by exposing the film in a precise
sequence throughout the entire 32 second exposure. Since the
start of the exposure is known very accuratly, we can measure
15 points along each of the trails and determine where the
telescope was pointing at specific times during each scan.
From this information, we produce a list of star positions
at each of the 15 measurement times. The star positions are
expressed in terms of the internal coordinates used by the
measuring instrument. To translate these positions into the
familiar RA and DEC that astronomers prefer to use, we have
to find the RAs and Decs of each of the measured stars, using
a star catalogue. These reference star positions, together
with the measurements of the star trails, are then processed
by a VAX 11/780 computer to find the actual RA and Dec of
the center of the camera field at 7 instants during a scan.
Generally speaking,
the optical axis of the camera is not the same as either the
telescope or the detector optical axes. To find out where
each detector was pointing, we usually select a bright planet
such as Mars or Jupiter whose trail we can measure on the
film and which is also a strong far-infrared source. We then
find out when each detector showed the highest, far-infrared
signal. What this means is that the planet crossed through
the center of the detector's field-of-view. Once the 'peak
signal' time is known, we can use this along with the planet's
trailed image on the film, to figure out where each dete ctor
was looking relative to the center of the cameras field-of-view.
The last step
is to take this pointing information and find a mathematical
equation to represent this 'aspect' data in a more compact
way. This also lets us interpolate to find positions for times
other than the 7 instants for which the aspect results were
derived originally. An example of a typical scan reduced in
this way looks like this
RA(1950.0) =
6.268 + .00197 t -.00000084 t
Dec(1950.0)=
173.83 - .11600 t +.0000186 t
where t is the
time, in seconds, since the start of the hour. All of the
32 scans that make up a single, 32-line raster are analyzed
in this way. Once we have found-out where our detectors were
pointing in the sky at a given time, we can then go on to
the next stage in the analysis.
WHAT
DID WE SEE?
Even at an altitude
of 18 miles, the atmosphere of the earth is still not completely
transparent to far-infrared radiation. To far-infrared eyes,
the rarefied atmosphere at 95,000' produces a haze, against
which it is possible to see only a handful of the brightest
sources in the sky. The telescope mirror itself is also a
warm surface with a temperature of about 77 K which contributes
its own undesireable 'background' emission to that of the
weak emissions from extraterrestrial sources. To get around
these problems, astronomers have come up with an ingeneous
solution; What we do instead of letting the telescope stare
at one part of the sky, straining to detect weak signals,
is to first measure the infrared brightness at the 'source'
position and then tilt the secondary mirror on the telescope
slightly and make a measurement of an adjacent part of the
sky. By subtracting the 'sky' measurement from the 'source'
measurement the remainder will be the weak signal produced
by the source itself without the sky background. By 'chopping'
in this way, 16 times a second, we can actually detect sources
that are thousands of times fainter than the sky brightness.
Unfortunatly, a small but annoying price has to be paid for
this advantage. Because we are subtracting the far-infrared
emission from two nearby points in the sky, it is virtually
impossible to study weak sources that are much larger than
the separation between the source and background positions
we measured. Since the surface brightness of such extended
sources can be very uniform, they contribute almost equally
to the radiation measured in the source and reference positions.
As a result, extended sources often get 'subtracted' out of
the data. Astronomers must always be mindful of this possibility
when analyzing their data since it could give us a false estimate
of a sources size and total luminosity.
The secondary
mirror on the cassegrain telescope is oscillated back and
forth by electromagnets to let the radiation from the reference
and source positions enter the detectors field-of-view. The
voltage that we measure from each detector after amplification,
by 90,000 times, is then a modulated, 16 cycle per second
'square-wave' representing alternatly the intensity of the
source and background positions. These voltages are digitized
and then telemetered from the balloon platform to the ground
station where they are recorded on the flight magnetic tapes
in terms of '10-bit' computer words. This allows the detector
voltages to be broken down into 1024 brightness levels. Back
at the Lab, a computer reads each data tape, subtracts the
source and background measurements, and averages them over
1/8 second intervals throughout the flight. Once the sections
of the data corresponding to our object of study have been
identified, this averaged data will be processed a third time.
At this time, the aspect information, giving the detectors
RA and Dec in time, will be combined with the averaged detector
voltage at each instant, to give us the intensity of the source
at each point that was covered during the mapping operation.
From this, contour maps can then be drawn, showing levels
of constant brightness.
"Where's
the Flux?" or "When does 735 = .0000000012 Watts/meter2?"
By combining
the detector output and aspect data, a progam instructs the
computer to plot the detector voltage at each point in the
sky covered by the survey. A set of contours are then drawn
by hand on this map, showing regions that have the same brightness
level. This brings us nearly to the end of the map making
process. What we will publish in the scientific paper on this
object will be the contour plot itself and not the individual
measurements made by each detector. This makes the maps more
readable and less expensive to publish. To complete the mapmaking
process we have to translate the brightness levels from the
internal language of the computers 1024 voltage level steps
to an actual number representing the amount of radiation that
fell on the telescope itself; a quantity called the flux,
usually expressed as a certain number of watts of power falling
on each square meter of the mirrors surface. This is the calibration
process.
To calibrate
our intensity measurements, a bright object is selected whose
spectrum is accuratly known. During a recent flight we use
the planet Mars. A mathematical model of how the planets surface
responds to being heated by the suns incoming radiation, tells
us that the surface temperature of Mars, at the time it was
observed, was 241 K. This value changes slightly as the planets
distance from the earth and the sun varies. To calculate how
bright the planet was, as viewed by the detector, we need
to know one last important detail; the response curve of the
detector. This curve tells how efficient the detector is at
sensing radiation at various wavelengths. As an example, the
optical detector called the human eye can only sense radiation
with wavelengths between 4000 and 7000 A. Its peak efficiency
occurs at 5700 A where most of the sun's radiation appears.
A similar response curve can be measured for our detector
under laboratory conditions. A computer is then used to calculate
how bright a 241 K blackbody will appear to our detectors
that only see the part of the spectrum defined by the limits
of the response curve. The calibration program also tells
us how much of the total radiation produced by a source, will
be missed by confining our attention to only a narrow part
of the spectrum. For Mars, our far-infrared detectors that
see only the radiation between wavelengths of 40 to 250 micrometers,
will only detect 1% of the radiation emitted by Mars. From
a careful study of the data, we discover that a voltage level
of 735 at the medium gain setting corresponded to a flux of
.0000000012 Watts/meter2. This number can then be used, with
slight modifications, to set the scale for all other sources
we observe. Incidently, at the highest gain setting, our instrument
together with the 1-meter telescope, can just detect a flux
of 2 x 10^-12 watts/meter2; this is about how much warmth
you would receive if you were standing 60 miles away from
a 100 Watt bulb.
HOW
LONG DOES IT TAKE?"
From start to
finish, the entire data analysis process takes several months
and occasionally can take the better part of a year if other
research has to be conducted at the same time. Astronomers
rarely have the luxury to work on a single research project
full time, but often divide their time among several other
projects in various stages of completion. Many astronomical
papers got their start while the astronomer was flying on
a commercial jet between observatories! A schedule might look
like this one week:
Monday.....
Work on the balloon data from the flight last April.
Tuesday.....
Work on the Astrophysical Journal article on IC-443 based
on observations I made at the VLA last January.
Wednesday.....
Work on the VLA maps of the Galactic Center far-infrared sources
that I observed last April.
Thursday.....
Work on the radio source survey to finish the first draft
of a paper.
Friday.....
Analyze the carbon monoxide observations of NGC-4088 and 3556
made last March.
and then change
the following week as you find yourself 'hot on the trail'
of some new discovery suggested to you by your data. In addition,
many weeks of time can be taken-up during the year, carrying-out
new observations at the VLA or Kitt Peak; or even to plan
next years flight of the far-infrared telescope. Astronomers
almost always have a backlog of 'old' data, waiting to be
analyzed and written up for publication. There is, however,
a tremendous sense of excitement in waking-up on Monday morning
to work on data having to do with star forming regions in
our galaxy, and on some other day, to wrestle with understanding
what is going on inside the nucleus of a galaxy millions of
light years away!
|
