Download imaging science in astronomy - RIT CIS

682
IMAGING SCIENCE IN ASTRONOMY
65 (a) E. V. Sayre and H. N. Lechtman, Stud. Conserv. 13,
161–185 (1968); (b) M. W. Ainsworth et al., Art and Autoradiography: Insights into the Genesis of Paintings by Rembrandt, Van Dyck, and Vermeer, The Metropolitan Museum
of Art, NY, 1982; (c) C. O. Fischer et al., Nucl. Instrum. Methods A 424, 258–262 (1999); (d) Images and supporting text
provided by Ward Laboratory, Cornell University, Ithaca,
NY 14853-7701.
66. D. L. Glackin and E. P. Korsmo, Jet Propulsion Laboratory,
Final Report 83-75, JPL Publications, Pasadena, 1983.
67. J. R. Druzik, D. Glackin, D. Lynn, and R. Quiros, 10th Annu.
Meet. Am. Inst. Conserv., 1982, pp. 71–72.
68. E. J. Wood, Textile Res. J. 60, 212–220 (1990).
69. F. Heitz, H. Maitre, and C. DeCouessin, IEEE Trans. Acous.,
Speech Signal Process. 38, 695–704 (1990).
70. J. Sobus, B. Pourdeyhimi, B. Xu, and Y. Ulcay, Textile Res. J.
62, 26–39 (1992).
71. K. Knox, R. Johnston, and R. L. Easton Jr., Opt. Photonics
News 8, 30–34 (1997).
72. L. Likforman-Sulem, H. Maitre, and C. Sirat, Pattern Recognition 24, 121–137 (1991).
73. E. Lang, and D. Watkinson, Conserv. News 47, 37–39 (1992).
74. F. Su, OE Reports SPIE 99, 1,8–(1992).
75. J. L. Kirsch and R. A. Kirsch, Leonardo 21, 437–444 (1988).
76. J. F. Asmus, Opt. Eng. 28, 800–804 (1989).
77. R. Sablatnig, P. Kammerer, and E. Zolda, Proc. 14th Int. Conf
Pattern Recognition, 1998, pp. 172–174.
78. L. R. Doyle, J. J. Lorre, and E. B. Doyle, Stud. Conserv. 31,
1–6 (1986).
79. J. Asmus, Byte Magazine March (1987).
80. P. Clogg, M. Diaz-Andreu, and B. Larkman, J. Archaeological
Sci. 27, 837 (2000).
81. P. Clogg and C. Caple, Imaging the Past, British Museum
Occasional Paper, London, 1996, p. 114.
IMAGING SCIENCE IN ASTRONOMY
JOEL H. KASTNER
Rochester Institute of Technology
Rochester, NY
INTRODUCTION
The vast majority of information about the universe is
collected via electromagnetic radiation. This radiation is
emitted by matter distributed across tremendous ranges
in temperature, density, and chemical composition. Thus,
more than any other science, astronomy depends on
innovative methods to extend image taking to new,
unexplored regions of the electromagnetic spectrum. To
bring sufficient breadth and depth to their studies,
astronomers also require imaging capability across a vast
range in spatial resolution and sensitivity, with emphasis
on achieving the highest possible resolution and signal
gain in a given wavelength regime. This simultaneous
quest for better wavelength coverage and ever higher
spatial resolution and sensitivity represents the driving
force for innovation and discovery in astronomical
imaging.
Classical astronomy — for example, the search for new
solar system objects and the classification of stars — is
still largely conducted in the optical wavelength regime
(400–700 nm). This has been the case, of course, since
humans first imagined the constellations, noted the
appearance of ‘‘wandering stars’’ (planets), and recorded
the appearance of transient phenomena such as comets
and novae. During the latter half of the twentieth century,
however, a revolution in astronomical imaging took
place (1). This relatively brief period in recorded history
saw the development and rapid refinement of techniques
for collecting and detecting electromagnetic radiation
across a far broader wavelength range, from the radio
through γ rays. Just as these techniques have reached
maturation, astronomers have also developed the means
to surmount apparently fundamental physical barriers
placed on image quality, such as the distorting effects
of refraction by Earth’s atmosphere and diffraction by a
single telescope of finite aperture. The accelerating pace of
these innovations has resulted in deeper understanding of,
and heightened appreciation for, both the rich diversity of
astrophysical phenomena and the fundamental, unsolved
mysteries of the cosmos.
OPENING THE WINDOWS: MULTIWAVELENGTH
IMAGING
For most of us, our eyes provide our first, fundamental
contact with the universe. It is interesting to ponder
how humans would conceive of the universe if we had
nothing more in the way of imaging apparatus at our
disposal, as was the case for astronomers before Galileo.
In contrast to the complex cosmologies currently pondered
in modern physics, most of which involve an expanding
universe shadowed by the afterglow of the Big Bang, the
‘‘first contact’’ provided by our eyes produces a model of
the universe that is entirely limited to the Sun, Moon,
and planets, the nearby stars, and the faint glow of
the collective background of stars in our own Milky
Way galaxy and a handful of other, nearby galaxies.
From this simple thought experiment, it is clear that the
bulk of the visible radiation arriving at Earth is emitted
by stars.
But the apparent predominance of visible light from
the Sun and nearby stars is in fact merely an accident
of our particular position in the universe, combined
with the evolutionary adaptation that gave our eyes
maximal sensitivity at wavelengths of electromagnetic
radiation that are near the maximum of the Sun’s energy
output. The Sun provides by far the majority of the
visible radiation arriving at Earth strictly by virtue
of its proximity. The brightest star in the night sky,
Sirius (in the constellation Canis Major), actually has
an intrinsic luminosity about 50 times larger than that of
the Sun, but is about 8.6 light years distant (a light year
is the distance traveled by light in one year, 9 × 1012 km;
the Sun is about 8 light minutes from Earth). In turn,
Sirius is only about one-ten-thousandth as luminous as
the star Rigel (in the neighboring constellation Orion),
IMAGING SCIENCE IN ASTRONOMY
683
but Sirius appears several times brighter than Rigel
because it is about 50 times closer to us. Like the Sun,
which has a surface temperature of about 6,000 K, most
of the brightest stars have surfaces within the range
of temperatures across which hot objects radiate very
efficiently (if not predominantly) in the visible region.
Representative stellar surface temperatures are 3,000 K
for reddish Betelgeuse, a red supergiant in Orion; 10,000 K
for Sirius; and 15,000 K for the blue supergiant Rigel
(Fig. 1).
Thermal Continuum Emission
The tendency of objects at the temperatures of the Sun
and stars to emit in the visible can be understood to first
order via Planck’s Law, which describes the wavelength
dependence of radiation emitted by a perfect blackbody.
The peak of the Planck function lies within the visible
regime for an object at a temperature of 6,000 K. This same
fundamental physical principle tells us that objects much
hotter or cooler than the Sun should radiate predominantly
at wavelengths much shorter or longer than visible,
respectively. Indeed, for a perfect blackbody, the peak
wavelength of radiation is given by Wien’s displacement
law (2),
0.51
.
(1)
λ(cm) ∼
T(K)
This relationship between the temperatures of objects and
the wavelengths of their emergent radiation allows us to
understand why Betelgeuse appears reddish and Rigel
appears blue (Fig. 2).
106
105
Rigel
Deneb
Betetgeuse
M
Spica
Superglants
ts
lan
dg
e
Capella R Aldebaran
Vega
Arcturus
Sirius A
Pollux
Altair
Procyon A
qu
se
103
n
ai
102
ce
en
Intrinsic brightness (Sun = 1)
104
10
1
Sun
10−1
W
hit
e
10−2
dw
ar
ts
Sirius B
Procyon B
10−3
40,000
20,000
10,000 6,000 4,000 3,000 2,000
Figure 2. Wide-field photograph of Orion, illustrating the
difference in color between the relatively cool star Betelgeuse
(upper left) and the hot star Rigel (lower right). The large, red
object at the lower center of the image, just below Orion’s belt,
is the Orion Nebula (see Fig. 7). (Photo credit: Till Credner,
AlltheSky.com) See color insert.
The same, simple relationship also provides powerful
insight into astrophysical processes that occur across a
very wide range of energy regimes (Fig. 3). The lowest
energies and hence longest (radio) wavelengths reveal
‘‘cold’’ phenomena, such as emission from dust and gas in
optically opaque clouds distributed throughout interstellar
space in our galaxy. At the highest energies and hence
shortest wavelengths (characteristic of X rays and γ rays),
astronomers probe the ‘‘hottest’’ objects, such as the
explosions of supermassive stars or the last vestiges
of superheated material that is about to spiral into a
black hole.
Stars' surface temperature (K)
Figure 1. The Hertzsprung–Russell diagram. The diagram
shows the main sequence (Sun-like stars that are fusing hydrogen
to helium in the cores), red giants, supergiants, and white
dwarfs. In addition, the positions of the Sun, the twelve brightest
stars visible from the Northern Hemisphere, and the white
dwarf companions of Sirius and Procyon are indicated [Source:
NASA (http://observe.ivv.nasa.gov/nasa/core.shtml.html)]. See
color insert.
Nonthermal Continuum Emission
Certain radiative phenomena in astrophysics do not
strongly depend on the precise temperature of the material
and are instead sensitive probes of material density and/or
chemical composition (3,4). For example, the emission
from ‘‘jets’’ ejected from supermassive black holes at
the centers of certain galaxies (Fig. 4) is said to be
684
IMAGING SCIENCE IN ASTRONOMY
Figure 3. Schematic diagram showing various regimes of the electromagnetic spectrum in terms
of temperatures corresponding to emission in that regime. The diagram also illustrates the
wavelength ‘‘niches’’ of NASA’s four orbiting ‘‘Great Observatories.’’ [Source: NASA/Chandra
X-Ray Center (http://chandra.harvard.edu)]. See color insert.
‘‘nonthermal’’ because its source is high-velocity electrons
that orbit around magnetic field lines. Other, similar
examples are the emission from filaments of ionized gas
located near the center of our own galaxy and from the
chaotic remnant of the explosion of a massive star in
1054 A.D. (the ‘‘Crab Nebula’’). Such so-called ‘‘synchrotron
radiation’’ often dominates radiation emitted in the radio
wavelength regime (Fig. 5). Indeed, if human eyes were
sensitive to radio rather than to visible wavelengths, the
early mariners probably would have navigated by the
Galactic Center and the Crab because they appear from
Earth as the brightest stationary radio continuum sources
in the northern sky. The synchrotron emission from the
Crab is particularly noteworthy; it can be detected across
a very broad wavelength range from radio through X ray
(Fig. 6).
Monochromatic (‘‘Line’’) Emission and Absorption
Deducing Chemical Compositions. Astronomers use
electronic transitions of atoms (as well as electronic,
vibrational, and rotational transitions of molecules) as
Rosetta stones to understand the chemical makeup of
gas in a wide variety of astrophysical environments.
Because each element or molecule radiates (and absorbs
radiation) at a discrete and generally well-determined set
of wavelengths — specified by that element’s particular
subatomic structure — detection of an excess (or deficit) of
emission at one of these specific wavelengths1 is both
necessary and sufficient to determine the presence of
that element or molecule. Hence, our knowledge of the
origin and evolution of the elements that make up the
universe is derived from astronomical spectroscopy (which
might also be considered multiband, one-dimensional
imaging).
Spectra obtained by disparate means across a very
broad range of wavelengths can be used to ascertain
both chemical compositions and physical conditions (i.e.,
temperatures and densities) of astronomical sources
because the emissive characteristics of a given element
depend on the physical conditions of the gas or dust
in which it resides. For example, cold (100 K), largely
neutral hydrogen gas emits strongly in the radio at 21 cm,
whereas hot (10,000 K), largely ionized hydrogen gas
emits at a series of optical wavelengths (known as the
Balmer series). The former conditions are typical of the
gas that permeates interstellar space in our own galaxy
and in external galaxies, and the latter conditions are
typical of gas in the proximity of very hot stars, which
are sources of ionizing ultraviolet light. Such ionized
gas also tends to glow brightly in the emission lines of
heavier elements such as oxygen, nitrogen, sulfur, and
iron (Fig. 7).
1 Such spectral features are called ‘‘lines,’’ because they appeared
as dark lines in early spectra of the Sun.
IMAGING SCIENCE IN ASTRONOMY
Figure 4. At a distance of 11 million light years, Centaurus A
is the nearest example of a so-called ‘‘active galaxy.’’ This radio
image shows opposing ‘‘jets’’ of high energy particles blasting out
from its center [Source: National Radio Astronomy Observatory
(NRAO)]. See color insert.
Figure 5. The Crab Nebula is the remnant of a supernova
explosion that was seen from the earth in 1054 A.D. It is 6,000
light years from Earth. This radio image shows the complex
arrangement of gas filaments left in the wake of the explosion
(Source: NRAO). See color insert.
Deducing Radial Velocities from Spectral Lines. Atomic
and molecular emission lines also serve as probes of bulk
motion. If a given source has a component of velocity
along our line of sight, then its emission lines will be
685
Figure 6. X-ray image of the innermost region of the Crab
Nebula. This image covers a field of view about one-quarter
that of the radio image in the previous figure. The image shows
tilted rings or waves of high-energy particles that appear to have
been flung outward across a distance of a light year from the
central star (Source: Chandra X-Ray Center). See color insert.
Figure 7. Color mosaic of the central part of the Great Nebula
in Orion, obtained by the Hubble Space Telescope. Light emitted
by ionized oxygen is shown as blue, ionized hydrogen emission
is shown as green, and ionized nitrogen emission as red. The
sources of ionization of the nebula are the hot, blue-white stars
of the young Trapezium cluster, which is embedded in nebulosity
just left of center in the image (Source: NASA and C.R. O’Dell
and S.K. Wong). See color insert.
Doppler shifted away from the rest wavelength. The
absorption or emission lines of sources that approach
686
IMAGING SCIENCE IN ASTRONOMY
Velocity (km/s)
30,000
20,000
10,000
0
0
100
200
300
Distance (Mpc)
400
500
Figure 8. Plot of recession velocity vs. distance [in megaparsecs
(Mpc); 1 Mpc ≈ 3 × 1019 km] for a sample of galaxies. This figure
illustrates that, to very high accuracy, the recession velocity
of a distant galaxy, as measured from its redshift, is directly
proportional to its distance. This correlation was first established
in 1929 by Edwin Hubble and underpins the Big Bang model for
the origin of the Universe (Figure courtesy Edward L. Wright, 
1996).
us are shifted to shorter wavelengths and are said to
be ‘‘blueshifted,’’ whereas the lines of sources moving
away from us are shifted to longer wavelengths and are
said to be ‘‘redshifted.’’ The observation by Hubble in
1929 that emission lines of distant galaxies are uniformly
redshifted and that these redshifts increase monotonically
as the distances of the galaxies increase, underpins
modern theories of the expansion of the universe2 (Fig. 8).
Images obtained at multiple wavelengths that span the
rest wavelength of a bright spectral line can allow
astronomers to deduce the spatial variation of lineof-sight velocity for a source whose velocity gradients
are large. Such velocity mapping, which is presently
feasible at wavelengths from the radio through the optical,
helps elucidate the three-dimensional structure of sources
(Fig. 9).
Multiwavelength Astronomical Imaging: An Example
Planetary nebulae represent the last stages of dying, Sunlike stars. These highly photogenic nebulae are formed
after the nuclear fuel at the core of a Sun-like star has
been spent, that is, the bulk of the core hydrogen has been
converted to helium. The exhaustion of core hydrogen
and the subsequent nuclear fusion, in concentric shells, of
hydrogen into helium and helium into carbon around the
2
In practice, all astrophysical sources that emit line radiation — even those within our solar system — will appear Doppler
shifted, due for example, to the Earth’s motion around the Sun.
Hence it is necessary to account properly for ‘‘local’’ sources of
Doppler shifts when deducing the line-of-sight velocity component
of interest.
spent core causes the atmosphere of the star to expand,
forming a red giant. Although the extended atmospheres
of red giants are ‘‘cool’’ enough (∼3,000 K) for dust grains
to condense out of the stellar gas, red giant luminosities
can be huge (more than 10,000 times that of the Sun).
This radiant energy pushes dust away from the outer
atmosphere of the star at speeds of 10–20 km s−1 . The
outflowing dust then collides with and accelerates the
gas away from the star, as well. Eventually enough of
the atmosphere is removed so that the hot, inert stellar
core is revealed. This hot core is destined to become
a fossil remnant of the original star: a white dwarf.
But before the ejected atmosphere departs the scene
entirely, it is ionized by the intense ultraviolet light from
the emerging white dwarf, which has cooled from core
nuclear fusion temperatures (107 to 108 K) to a ‘‘mere’’
105 K or so. The ionizing radiation from the white dwarf
causes the ejected gas to fluoresce, thereby producing a
planetary nebula.
Because the varied conditions that characterize the
evolution of planetary nebulae result in a wide variety of
phenomena in any given nebula, such objects demand a
multiwavelength approach to imaging. A case in point is
the young planetary nebula BD +30° 3639 (Fig. 10). This
planetary nebula emits strongly at wavelengths ranging
from radio through X ray. The Chandra X-ray image shows
a region of X-ray emission that seems to fit perfectly inside
the shell of ionized and molecular gas seen in Hubble Space
Telescope images and in other high-resolution images
obtained from the ground. The optical and X-ray emitting
regions of BD +30° 3639, which lies about 5,000 light
years away, are roughly 1 million times the volume of our
solar system. The X-ray emission apparently originates in
thin gas that is heated by collisions between the ‘‘new’’
wind blown by the white dwarf, which is seen at the
center of the optical and infrared images, and the ‘‘old,’’
photoionized red giant wind, which appears as a shell
of ∼10,000 K gas surrounding the ‘‘hot bubble’’ of X-ray
emission.
REQUIREMENTS AND LIMITATIONS
To understand the requirements placed on spatial
resolution and sensitivity in astronomical imaging, we
must consider the angular sizes and energy fluxes of
astronomical objects and phenomena of interest. In turn,
there are three fundamental sources of limitation on the
resolution and limiting sensitivity (and hence quality) of
astronomical images: the atmosphere, the telescope, and
the detector.
Spatial Resolution
Requirements: Angular Size Scales of Astronomical
Sources. Figure 11 shows schematically typical scales of
physical size and distance from Earth for representative
objects and phenomena studied by astronomers. Most of
the objects of intrinsically small size, like the Sun, Moon,
and the planets in our solar system, lie at small distances;
we can study these small objects in detail only because
they are relatively close, such that their angular sizes are
substantial.
IMAGING SCIENCE IN ASTRONOMY
687
38′′41′45′′
38′′41′40′′
38′′41′36′′
Dec. (2000.0)
38′′41′30′′
38′′41′45′′
38′′41′40′′
38′′41′35′′
38′′41′30′′
21h02m19.0
21h02m18.5
R.A. (2000.0)
21h02m18.0
Figure 9. Radio maps of the Egg Nebula, a dying star in the constellation Cygnus, showing
emission from the carbon monoxide molecule. At the lower left is shown blueshifted CO emission,
and at the lower right redshifted emission; the upper right panel shows the total intensity of CO
emission from the source. One interpretation for the localized appearance of the blueshifted and
redshifted CO emission is that the Egg Nebula is the source of a complex system of ‘‘molecular
jets,’’ shown schematically in the top left panel. Such jets may be quite common during the dying
stages of Sun-like stars [Source: Lucas et al. 2000 (5)]. See color insert.
Within our own Milky Way galaxy, we observe objects
that span a great range of angular size scales. The
angular size of a Sun-like star at even a modest
distance makes such stars a challenge to resolve spatially,
even with the best available techniques. On the other
hand, many structures of interest in our own Milky
Way galaxy, such as star-forming molecular clouds and
the expelled remnants of dying or expired stars, are
sufficiently large that their angular sizes are quite large.3
Certain giant molecular clouds, planetary nebulae, and
supernova remnants subtend solid angles similar to that
of the Moon.
Just as for stars, the angular sizes of external galaxies
span a very wide range. The Magellanic Clouds, which are
the nearest members of the Local Group of galaxies (of
which the Milky Way is the most massive and luminous
3
The ejected envelopes of certain dying, sun-like stars were long
ago dubbed ‘‘planetary nebulae’’ because their angular sizes and
round shapes resembled the planets Jupiter and Saturn.
member), are detectable and resolvable by the naked eye,
whereas the Andromeda galaxy (a Local Group member
that is a near-twin to the Milky Way) is detectable
and resolvable with the aid of binoculars. The angular
sizes of intrinsically similar galaxies in more distant
galaxy clusters span a range similar to that of the
planets in our solar system. The luminous cores of certain
distant galaxies (‘‘quasars’’) — which can outshine their
host galaxies — likely have sizes only on the order of
that of our solar system; yet these are some of the most
distant objects known, and hence quasars are exceedingly
small in angular size. Galaxy clusters themselves are
of relatively large angular size, simply by virtue of
their enormous size scales; indeed, such clusters (and
larger scale structures that consist of clusters of such
clusters) probably represent the largest gravitationally
bound structures in the universe. At still larger size
scales lies the cosmic background radiation, the radiative
remnant of the Big Bang itself. This radiation encompasses
4π steradians and has only very subtle variations in
intensity with position across the sky.
688
IMAGING SCIENCE IN ASTRONOMY
Figure 10. Optical (left), infrared (center), and X-ray (right) images of the planetary nebula
BD +30° 3639 [Source: Kastner et al. 2000 (6)]. The optical image was obtained by the Wide
Field/Planetary Camera 2 aboard the Hubble Space Telescope in the light of doubly ionized
sulfur at a wavelength of 9,532 Å. The infrared image was obtained by the 8-meter Gemini North
telescope at a wavelength of 2.2 µm (also referred to as the infrared K band). The X-ray image was
obtained by the Advanced CCD Imaging Spectrometer aboard the Chandra X-Ray Observatory,
and covers the wavelength range from ∼7 Å to ∼30 Å. Images are presented at the same spatial
scale. See color insert.
Of course, even within our solar system, there are
sources of great interest (e.g., the primordial, comet-like
bodies of the Kuiper Belt) that are sufficiently small that
they are unresolvable by present imaging techniques.
Sources of large angular sizes (such as molecular clouds,
planetary nebulae, supernova remnants, and galaxy
clusters) typically show a great wealth of structural
detail when imaged at high spatial resolution. Thus,
our knowledge of objects at all size and distance scales
improves with any increase in spatial resolving power at
a given wavelength.
Limitations
Atmosphere. Time- and position-dependent refraction
by turbulent cells in the atmosphere causes astronomical
point sources, such as stars, to ‘‘scintillate’’; i.e., stars
twinkle. Scintillation occurs when previously planeparallel wave fronts from very distant sources encounter
atmospheric cells and become distorted. Astronomers use
the term ‘‘seeing’’ to characterize such atmospheric image
distortion; the ‘‘seeing disk’’ represents the diameter of
an unresolved (point) source that has been smeared by
atmospheric distortion. Seeing varies widely from site to
site, but optical seeing disks at visual wavelengths are
typically not smaller than (that is, the seeing is not better
than) ∼1 at most mountaintop observatories.
Telescope. The diameter of a telescope places a
fundamental limitation on the angular resolution at
a given wavelength. Specifically, the limiting angular
resolution (in radians) is given by
θ ≈ 1.2
λ
d
(2)
where θ is the angle subtended by a resolution element,
λ is the wavelength of interest, and d is the telescope
diameter. This relationship follows from consideration
of simple interference effects of wave fronts incident
on a circular aperture, in direct analogy to planeparallel waves of wavelength λ incident on a single
slit of size d. The resulting intensity distribution for a
point source (known as the ‘‘point-spread function’’) is
in fact a classical diffraction pattern, a central disk (the
‘‘Airy disk’’) surrounded by alternating bright and dark
annuli. In ground-based optical astronomy using large
telescopes, atmospheric scintillation usually dominates
over telescope diffraction (that is, the ‘‘seeing disk’’
is much larger than the ‘‘Airy disk’’), and such a
diffraction pattern is not observed. However, in spacebased optical astronomy or in ground-based infrared and
radio astronomy, diffraction represents the fundamental
limitation on spatial resolution.
Detector. Charge-coupled devices (CCDs) have been
actively used in optical astronomy for more than two
decades. During this period, CCD pixel sizes have steadily
decreased, and array formats have steadily grown. As
a result, CCDs have remained small and still maintain
good spatial coverage. Detector array development at
other wavelength regimes lags behind the optical, to
various degrees, in number and spacing of pixels. However,
almost all regimes, from X ray to radio, now employ some
form of detector array. Sizes range from the suite of ten
1, 024 × 1, 024 X-ray-sensitive CCDs aboard the orbiting
Chandra X-Ray Observatory to the 37- and 91-element
bolometer arrays used for submillimeter-wave imaging
by the James Clerk Maxwell Telescope on Mauna Kea.
These devices have a common goal of achieving a balance
between optimal (Nyquist) sampling of the point-spread
function and maximal image (field) size.
Sensitivity
Requirements: Energy Fluxes of Astronomical Sources.
Astronomical sources span an enormous range of intrinsic
luminosity. Figure 12 readily shows that the least
luminous objects known tend to be close to Earth (e.g.,
IMAGING SCIENCE IN ASTRONOMY
689
1015
Virgo_cluster
1010
M31
Cygnus_GMC
Radius (astron. units)
Cas_A
105
Ring_Nebula
CS_disk
3C273
Betelgeuse
100
Sun
Alpha_Cen
Jupiter
Moon
Pluto
10−5
Crab_pulsar
10−10
10−10
10−5
100
Distance (light years)
105
1010
Figure 11. Physical radii vs. distances (from Earth) for representative astronomical sources (7).
One astronomical unit (AU) is the Earth–Sun distance (1.5 × 108 km). A light year is the distance
traveled by light in one year (9 × 1012 km). Represented in the figure are objects within our own
solar system, the nearby Sun-like star α Cen, the red supergiant Betelgeuse, the pulsar at the
center of the Crab Nebula supernova remnant, a typical circumstellar debris disk (‘‘CS disk’’),
a typical planetary nebula (the Ring Nebula), the supernova remnant Cas A, the galactic giant
molecular cloud located in the direction of the constellation Cygnus (‘‘Cygnus GMC’’), the nearby
Andromeda galaxy (M31), the quasar 3C 273, and the Virgo cluster of galaxies. Diagonal lines
represent lines of constant angular size, and angular size decreases from upper left to lower right.
small asteroids in the inner solar system), and the
most luminous sources known (e.g., the central engines
of active galaxies or the primordial cosmic background
radiation) are also the most distant. This tendency to
detect intrinsically more luminous sources at greater
distances follows directly from the expression for energy
flux received at Earth,
F=
L
,
4π D2
(3)
where F is the flux, L is luminosity, and D is distance.
Thus an astronomical imaging system that has a limiting
sensitivity F ≥ Fl penetrates to a limiting distance,
L
,
(4)
Dl ≤
4π Fl
for sources of uniform luminosity L. Real samples (of, e.g.,
stars or galaxies), of course, may include a wide range
of intrinsic luminosities. As a result, there tends to be
strong selection bias in astronomy, such that the number
and/or significance of intrinsically faint objects tends to be
underestimated in any sample of sources selected on the
basis of minimum flux.
For this reason in particular, astronomers require
increasingly sensitive imaging systems. To calibrate
detected fluxes properly, such systems must still retain
good dynamic range, so that the intensities of faint sources
can be accurately referenced to the intensities of bright,
well-calibrated sources. In addition, because a given source
of extended emission may display a wide variation in
surface brightness, a combination of high sensitivity and
good dynamic range frequently is required to characterize
690
IMAGING SCIENCE IN ASTRONOMY
1015
3C273
M31
1010
Total luminosity (solar units)
105
Betelgeuse
Cas_A
Ring_Nebula
Alpha_Cen
Sun
100
CS_disk
Crab_pulsar
10−5
Jupiter
10−10
Moon
Pluto
10−15 −10
10
10−5
100
Distance (light years)
105
1010
Figure 12. Intrinsic luminosities vs. distances (from Earth) for representative astronomical
sources; symbols are the same as in the previous figure. Luminosities are expressed in solar units,
where the solar luminosity is 4 × 1033 erg s−1 . Diagonal lines represent lines of constant apparent
brightness, and apparent brightness decreases from upper left to lower right.
source morphology adequately and, hence, deduce intrinsic
source structure.
Limitations
Atmosphere. The Earth’s atmosphere attenuates the
signals of most astronomical sources. Signal attenuation
is a function of both the path length through the
atmosphere between the source and telescope and the
atmosphere’s intrinsic opacity at the wavelength of
interest. Atmospheric attenuation tends to be smallest
at optical and longer radio wavelengths, at which the
atmosphere is essentially transparent. Attenuation is
largest at very short (γ ray, X ray and UV) wavelengths,
where the atmosphere is essentially opaque; attenuation
is also large in the infrared. In the infrared regime
especially, atmospheric transparency depends strongly
on wavelength because the main source of opacity is
absorption by molecules (in particular, water vapor).
The atmosphere also is a source of ‘‘background’’
radiation at most wavelengths, particularly in the thermal
infrared and far-infrared (2 µm ≤ λ ≤ 1 mm), at which
most of the blackbody radiation of the atmosphere
emerges. This background radiation tends to limit the
signal-to-noise ratio of infrared observations for which
other noise sources (such as detector noise) are minimal.
Elimination of thermal radiation from the atmosphere
provides a primary motivation for the forthcoming Space
Infrared Telescope Facility (SIRTF), the last in NASA’s
line of Great Observatories.
Telescope. Sensitivity (or image signal-to-noise ratio) is
directly proportional to the collecting area and efficiency
of the telescope optical surfaces (‘‘efficiency’’ here refers to
the fraction of photons incident on the telescope optical
surface that are transmitted to the camera or detector).4
Reflecting telescopes supplanted refracting telescopes at
the beginning of the twentieth century because large
primary mirrors could be supported more easily than large
4 The product of telescope collecting area and efficiency is referred
to as the effective area of the telescope.
IMAGING SCIENCE IN ASTRONOMY
objective lenses and the aluminized surface of a mirror
provides nearly 100% efficiency at optical wavelengths.
Furthermore, unlike lenses, paraboloid mirrors provide
images that are free of spherical or chromatic aberrations.
These same mirrors provide excellent efficiency and image
quality in the near-infrared, as well. Parabolic reflectors
are also used as the primary radiation collecting surfaces
in the radio regime, where the requirements of mirror
figure are less stringent (due to the relatively large
wavelengths of interest).
Detector. The photon counting efficiency of a detector
and sources of noise within the detector also dictate the
image signal-to-noise ratio. Photon counting efficiency is
usually referred to as detector quantum efficiency (QE).
Detector QEs at or higher than 80% are now feasible
in many wavelength regimes; however, such high QE
often comes at the price of the introduction of noise.
Typical image noise sources are read noise, the inherent
uncertainty in the signal readout of the detector, and dark
signal, the signal registered by the detector in the absence
of exposure to photons from an external source.
Surmounting the Obstacles
Beating the Limitations of the Atmosphere: Adaptive
Optics and Space-Based Imaging. Adaptive optics techniques have been developed to mitigate the effects of
atmospheric scintillation. In such systems, the image
of a fiducial point source — either a bright star or a
laser-generated artificial ‘‘star’’ — is continuously monitored, and these data are used to drive a quasi-real-time
image correction system (typically a deformable or steerable mirror). Naturally — as has been demonstrated by
the spectacular success of the refurbished Hubble Space
Telescope — placement of the telescope above the Earth’s
atmosphere provides the most robust remedy for the effects
of atmospheric image distortion.
Beating the Limitations of Aperture: Interferometry. The
diffraction limit of a single telescope can be surmounted
by using two or more telescopes in tandem. This
technique is referred to as ‘‘interferometry’’ because it
uses the interference patterns produced by combination
of light waves from multiple sources. Therefore, the
angular resolution of such a multiple telescope system,
at least in one dimension, is limited by the longest
separation between telescopes, rather than by the aperture
of a single telescope. However, it is generally not
possible to ‘‘fill in’’ the gaps between two telescopes at
large separation by using many telescopes at smaller
separation. As a result, interferometry is generally
limited to relatively bright sources, and interferometric
image reconstruction techniques necessarily sacrifice
information at low spatial frequencies (i.e., large-scale
structure) in favor of recovering information at high
spatial frequency (fine spatial structure). Interferometry
has long been employed at radio wavelengths because
recombination of signals from multiple apertures is
relatively easy at long wavelengths. Indeed, the angular
resolution achieved routinely at centimeter wavelengths
by NRAO’s Very Large Array in New Mexico rivals or
691
exceeds that of optical imaging by the Hubble Space
Telescope. Recently, however, several optical and infrared
interferometers have been developed and successfully
deployed; examples include the Navy Prototype Optical
Interferometer at Anderson Mesa and the Infrared Optical
Telescope Array on Mt. Hopkins, both in Arizona, and the
optical interferometer operated at Mt. Wilson, California,
by the Center for High Angular Resolution Astronomy.
Beating the Limitations of Materials: Mirror Fabrication. The sheer weight of monolithic, precision-ground
mirrors and the difficulty of maintaining the requisite
precise figures renders them impractical for constructing
telescope apertures larger than about 8 meters in diameter. Hence, during the late 1980s and early 1990s, two
competing large mirror fabrication technologies emerged:
spin-cast and segmented mirrors (Fig. 13). Both methods
have yielded large mirrors whose apertures are far lighter
and more flexible than previously feasible. The former
method has yielded the 8-meter-class mirrors for facilities
such as the twin Gemini telescopes, and the latter method
has yielded the largest mirrors thus far, for the twin
10-meter Keck telescopes on Mauna Kea. It is not clear,
however, that either technique can yield optical-quality
mirrors larger than about 15 meters in diameter.
An entirely different mirror fabrication approach is
required at high energies because, for example, X rays
are readily absorbed (rather than reflected) by aluminized
glass mirrors when such mirrors are used at near-normal
incidence. The collection and focusing of X-ray photons
instead requires grazing incidence geometry to optimize
efficiency and nested mirrors to optimize collecting
surface (Fig. 14). The challenge now faced by high-energy
astronomers is to continue to increase the effective
area of such optical systems while meeting the strict
weight requirements imposed by space-based observing
platforms. It is not clear that facilities larger than the
present Chandra and XMM-Newton observatories are
practical given present fabrication technologies; indeed,
Chandra was the heaviest payload ever launched aboard
a NASA Space Shuttle.
THE SHAPE OF THINGS TO COME
Projects in Progress
At this time, several major new astronomical facilities
are partially or fully funded and are either in design or
under construction. All are expected to accelerate further
the steady progress in our understanding of the universe.
A comprehensive list is beyond the scope of this article;
however, we mention a few facilities of note.
• The Space Infrared Telescope Facility (SIRTF):
SIRTF is a modest-aperture (0.8 m) telescope
equipped with instruments of extraordinary
sensitivity for observations in the 3 to 170 µm
wavelength regime. SIRTF features a powerful
combination of sensitive, wide-field imaging and
spectroscopy at low to moderate resolution over this
wavelength range. It is well equipped to study (among
many other things) primordial galaxies, newborn
692
IMAGING SCIENCE IN ASTRONOMY
Figure 13. Photo of the segmented primary mirror of the 10-meter Keck telescope (Photo credit:
Andrew Perala and W.M. Keck Observatory). See color insert.
Field-of-view
— 5°
Doubly
reflected
X rays
Four nested hyperboloids
Focal
surface
Doubly
reflected
X rays
10 meters
X rays
X rays
Four nested paraboloids
Mirror elements are 0.8 m long and from 0.6 m to 1.2 m in diameter
Figure 14. Geometry of the nested mirrors aboard the orbiting Chandra X-Ray Observatory
[Source: NASA/Chandra X-Ray Center (http://chandra.harvard.edu)]. See color insert.
stars and planets, and dying stars because all of these
phenomena emit strongly in the mid- to far-infrared.
SIRTF has a projected 5-year lifetime and is expected
to be deployed into its Earth-trailing orbit in 2002.
• The Stratospheric Observatory for Infrared Astronomy (SOFIA): SOFIA will consist of a 2.5-meter
telescope and associated cameras and spectrometers
installed aboard a Boeing 747 aircraft. SOFIA will
be the largest airborne telescope in the world. Due to
its ability to surmount most of Earth’s atmosphere,
SOFIA will make infrared observations that are
impossible for even the largest and highest groundbased telescopes. The observatory is being developed
and operated for NASA by a consortium led by
the Universities Space Research Association (USRA).
SOFIA will be based at NASA’s Ames Research Center at Moffett Federal Airfield near Mountain View,
California. It is expected to begin flying in the year
IMAGING SCIENCE IN BIOCHEMISTRY
2004 and will remain operational for two decades.
Like SIRTF, SOFIA is part of NASA’s Origins Program, and hence its science goals are similar and
complementary to those of SIRTF.
• The Atacama Large Millimeter Array (ALMA):
ALMA will be a large array of radio telescopes
optimized for observations in the millimeter wavelength regime and situated high in the Atacama
desert in the Chilean Andes. Using a collecting area
of up to 10,000 square meters, ALMA will feature
roughly 10 times the collecting area of today’s largest
millimeter-wave telescope arrays. Its telescope-totelescope baselines will extend to 10 km, providing
angular resolution equivalent to that of a diffractionlimited optical telescope whose diameter is 4 meters.
ALMA observations will focus on emission from
molecules and dust from very compact sources, such
as galaxies at very high redshift and solar systems in
formation.
Recommendations of the Year 2000 Decadal Review
The National Research Council, the principal operating arm of the National Academy of Sciences and
the National Academy of Engineering, has mapped
out priorities for investments in astronomical research
during the next decade (8). The NRC study should
not be used as the sole (or perhaps even primary) means to assess future directions in astronomy, but this study, which was funded by NASA,
the National Science Foundation, and the Keck Foundation does offer insight into some potential groundbreaking developments in multiwavelength astronomical
imaging.
Highest priority in the NRC study was given to the
Next Generation Space Telescope (NGST). This 8-meterclass, infrared-optimized telescope will represent a major
improvement on the Hubble Space Telescope in both
sensitivity and spatial resolution and will extend spacebased infrared imaging into the largely untapped 2–5 µm
wavelength regime. This regime is optimal for studying
the earliest stages of star and galaxy formation. NGST
presently is scheduled for launch in 2007.
Several other major initiatives were also deemed crucial
to progress in astronomy by the NRC report. Development
of the ground-based Giant Segmented Mirror Telescope
was given particularly high priority. This instrument has
as its primary scientific goal the study of the evolution
of galaxies and the intergalactic medium. Other projects
singled out by the NRC report include
• Constellation-X Observatory, a next-generation
X-ray telescope designed to study the origin and
properties of black holes;
• a major expansion of the Very Large Array radio
telescope in New Mexico, designed to improve on its
already unique contributions to the study of distant
galaxies and the disk-shaped regions around stars
where planets form;
• a large ground-based survey telescope, designed to
perform repeated imaging of wide fields to search
for both variable sources and faint solar-system
693
objects (including near-Earth asteroids and some of
the most distant, undiscovered objects in the solar
system); and
• the Terrestrial Planet Finder, a NASA mission
designed to discover and study Earth-like planets
around other stars.
BIBLIOGRAPHY
1. A. Sandage, Ann. Rev. Astron. Astrophys. 37, 445–486 (1999).
2. K. R. Lang, Astrophysical Formulae, 3rd ed., Springer-Verlag,
Berlin, 1999.
3. G. B. Rybicki and A. P. Lightman, Radiative Processes in
Astrophysics, John Wiley & Sons, Inc., NY, 1979.
4. D. Osterbrock, Astrophysics of Gaseous Nebulae and Active
Galactic Nuclei, University Science Books, Mill Valley, 1989.
5. R. Lucas, P. Cox, and P. J. Huggins, in J. H. Kastner, N. Soker,
and S. Rappaport, eds., Asymmetrical Planetary Nebulae II:
From Origins to Microstructures, vol. 199, Astron. Soc. Pac.
Conf. Ser., 2000, p. 285.
6. J. H. Kastner, N. Soker, S. Vrtilek, and R. Dgani, Astrophys.
J. (Lett.) 545, 57–59 (2000).
7. C. W. Allen and A. N. Cox, Astrophysical Quantities, 4th ed.,
Springer Verlag, Berlin, 2000.
8. C. McKee et al., Astronomy and Astrophysics in the New
Millennium, National Academy Press, Washington, 2001.
IMAGING SCIENCE IN BIOCHEMISTRY
NICOLAS GUEX
TORSTEN SCHWEDE
MANUEL C. PEITSCH
Glaxo Smith Klime Research &
Development SA
Geneva, Switzerland
INTRODUCTION
Research in biology, aimed at understanding the fundamental processes of life, is both an experimental and an
observational science. During the last century, all classes
of biomolecules relevant to life have been discovered and
defined. Consequently, biology progressed from cataloging
species and their life styles to analyzing their underlying
molecular mechanisms. Among the molecules required by
life, proteins represent certainly the most fascinating class
because they are the actual ‘‘working molecules’’ involved
in both the processes of life and the structure of living
beings. Proteins carry out diverse functions, including signaling and chemical communication (for example kinases
and hormones), structure (keratin and collagen), transport of metabolites (hemoglobin), and transformation of
metabolites (enzymes).
In contrast to modeling and simulation, observation and
analysis are the main approaches used in biology. Early
biology dealt only with the observation of macroscopic
phenomena, which could be seen by the naked eye. The
development of microscopes permitted observation of the
smaller members of the living kingdom and hence of the
cells and the organelles they contain (Fig. 1). Observing