Remote Sensing - The future of astronomical interferometry
Written by Sten Odenwald

Copyright (C) 1995

This web page is located at the Astronomy Cafe

http://www.astronomycafe.net

Astronomy Cafe


Pioneer...Voyager...Magellan. As the roll-call of 'firsts' in imaging our planetary neighbors draws to a close, the search for new solar systems to explore will loom as an ever more tantalizing prospect. Believe it or not, much of the ground work for this next, great adventure has already been lain. Long before NASA celebrates its 100th anniversary, using instruments now on the drawing boards, astronomers may well have cataloged all of the Jupiter-sized planets orbiting stars within 50 light years of the earth. Even the bright stars we admire in the night sky will have had their sunspots and prominences scrutinized. All this will come to pass, not because humans will travel to these distant worlds in starships, but because we will have taken remote sensing to its ultimate, natural limit.

To understand how these discoveries might come about is to rediscover the basic physical principles that allow us to see our universe as clearly as we do. Every amateur astronomer is familiar with the simple relationship A = 20 W/D describing how well a lens with a diameter of D centimeters will let you resolve features subtending A arcseconds, working at a wavelength of W microns. For the human eye, we can discern features about 60 arcseconds across; good enough to see large craters and mare on the moon. The Hubble Telescope with its 4-meter mirror can do much better than this, and has an acuity of 0.07 arc seconds. With it, we can obtain photos of Jupiter and Saturn nearly as good as those taken by the Voyager spacecraft.

But to distingish even a big planet like Jupiter orbiting the nearest star requires a resolving power of 0.0002 arc seconds, and at optical wavelengths implies a mirror 500 meters in diameter. At the much longer wavelengths used by radio astronomers, often measured in centimeters, most radio telescopes have dishes between 10 to 100 meters across providing a resolution at the shorter wavelengths that seldom exceed a few tens of arc seconds. For example, the 140-foot dish at Green Bank has a resolution at wavelengths of 1 centimeter of only one arcminute. At 1 millimeter wavelengths, the 12-meter radio telescope at Kitt Peak has a resolution of only 30 arcseconds. Even the 1000-foot Arecibo telescope operating at 21-centimeter has a reslution of only about 2 arcminutes: twice as poor as the human eye!

To acheive the highest resolutions, clearly the vastly shorter wavelengths of optical technology provide major advantages over radio wavelengths, yet it is impossible to build a single, optically perfect mirror 500 meters across to resolve distant extra-solar planets. Fortunately, astronomers have found a clever way around this problem: Two smaller mirrors separated by 500 meters can acheive the same resolving power as a single mirror of the same diameter.

INTERFEROMETRY

Interferometry, unlike natural vision, is a new way of seeing unanticipated by nature. No known organism sees its world in quite this way. If human eyes used this physical principle, our vision would be limited, not by the diameter of the lenses in our eyes, but by their 3 inch separation. Our acuity would be 25 times better than it is, at least in the horizontal plane.

Astronomical interferometers form an image in much the same way that a holograph is used to record the 3-dimensional image of an object on a piece of photographic film. The light reaching each pair of telescopes is combined to produce an interference pattern of the image of a celectial object. This interference pattern, or fringe, is a measure of just how much structure the wavefront from the source has along the orienation angle of the telescopes, and at the resolution scale set by their separations. When enough of these fringes have been collected from a variety of telescope orientations and separations, a complete interferogram is created containing all the information in the wavefront from the object. Just as for a hologram, this interferogram can be processed to reconstitute the image of the source itself.

All of this sounds very complicated... and it is. When you look into the eyepiece of your telescope, the optical surfaces of the mirrors and lenses bring the light from the object to a common focus to form an image which is itself a collection of bright and dark regions. Each pair of diametrically opposite spots on your mirror act as mineature telescopes in an interferometer and transmit their interference pattern to a common focal point. Uncounted trillions of these pairs work together to build-up a complex pattern at the focal point. In-phase components appear as bright spots, and out-of- phase pairs cancil to form blackness or shadows. This combination of light and dark is what forms what our brain interprets as an image. Radio astronomers were the first to use this technique in the 1950's. Since then, optical and infrared astronomers have begun to extend this technique to even shorter wavelengths.

If it were possible to build interferometers of any size, there would be much for them to see. The world of high resolution astronomy is a mathematical world bounded by distance, size and angular scale (Figure 1). The human eye and most single aperture telescopes crowd the lower reaches of this mathematical universe. Radio and optical interferometers now in use have begun to explore the vast middle ground extending to 0.0001 arcseconds. The high frontier, however, is vacant and portends a staggering view of the universe we can only dimly comprehend today. Consider just a few of the possibilites.

MEGAVISION : One Micro Arc Second

The disks of planets like Jupiter would be resolvable as far away as 3000 light years. Earth-sized planets orbiting nearby stars would be discerned as clearly as we see Jupiter with a 6" telescope. The motions of distant galaxies would be detectable in a single human lifetime as would the motions of individual stars in nearby galaxies.

SUPERVISION : One Nano Arc Second

The surfaces of planets orbiting Alpha Centauri are seen as clearly as Magellan now shows us Venus. We begin to resolve planets like Jupiter within the Andromeda galaxy, and the surface markings on individal red giant stars out to the Virgo Cluster. The motions of distant quasars would be discerned within a few years.

HYPERVISION : One Pico Arc Second

Features one foot across on the surfaces of planets orbiting nearby stars could be seen. Magellan-like images of all planets anywhere in the Local Group of galaxies would be possible. Jupiter-sized planets within distant quasars could be studied. Sunspots and prominences on stars anywhere in the visible universe could be studied in detail.

Beyond Hypervision, the mind reels at even more stupendous possibilities. Unfortunatly, the laws of physics present us with three formidible obstacles to overcome in making these possibilities a scientific reality: High resolution necessarily means having to build large costly instruments; Not all surface details are bright enough to be seen at every conceivable wavelength, distance and resolution; and finally, since everything in the universe is in motion, we have only a limited amount of time to form a sharp image before the objects move and cause image blurring.

A QUESTION OF MAGNITUDE

To acheive the highest resolutions with the smallest instruments, we must form our images using light of the shortest wavelengths. An optical interferometer 100 km across, not much larger than the Very Large Array in Socorro, New Mexico used by radio astronomers, will certainly let us acheive Megavision, but a Hypervision optical interferometer will have to be as big as the earth's orbit! At radio wavelengths measured in centimeters, even a modest Supervision interferometer would be as big as the solar system. These are not limitations to be transcended by some clever, future superscience. These limitations are set by the quantum nature of light itself. It is a curious parallel that to explore the sub-nuclear world, we must also build enormous machines such as the Superconducting Super Collider which are themselves nearly 100 kilometers in diameter.

For each 10-fold improvement in instrument size have come a host of new technological problems to be reconed with. Single telescopes require individual mirrors or lenses to be fabricated with surfaces figured to 1/8 wave tolerances or better to insure that the images they form are distortion-free. Arrays of telescopes acting as interferometers also require accurate surfaces, but there are in addition entirely new issues to be delt with. Not the least of these is the accurate timing of the incoming light wavefronts to insure proper fringe formation. Radio telescopes use highly accurate maser clocks, while optical and infrared interferometers under construction use complex combinations of lasers and optical modulation techniques to count and track the interference fringes. A precise knowledge of the exact geometry of the array is also needed. Compensation for the earth's ionosphere and atmosphere must also be performed to suppress 'twinkeling'; a pernicious form of wavefront distortion easily seen with the unaided eye.

STAR LIGHT...STAR BRIGHT

A part from the shear size of the instruments involved, there is an even ugglier problem waiting in the wings like a mugger on a dark night. Having built these gargantuan, solar system girding interferometers, and having developed new technologies to get them to work, there may not be anything for them to see!

Among the brightest and most numerous visual objects in the universe are the lowly stars. A patch of the sun 100 km on a side subtends 0.1 arc second from the earth. It emits 6 trillion trillion trillion photons into space every second in blinding, optical radiance. But, after traveling only 8 parsecs, these photons expand to become such a tenuous hemispherical shell of particles, no more than 100 pass through each square meter of your detector every second.

An optical interferometer consisting of two Hubble telescopes separated by 1400 kilometers would just be able to resolve this distant star patch with an resolution of 0.0000001 arcseconds. Their combined collecting area would gather 2500 photons each second, just enough light to form a single recognizable interference fringe. But just as you cannot throw away 99 percent of the information in a hologram and expect to see a high definition image afterwards, a single interferogram cannot recover all of the information in the image of the star at 100 km resolution. Thousands of additional interferograms from many orientations and separations of the telescope mirrors must be gathered and combined. This of course takes time. The situation is far worse for viewing the surfaces of planets beyond our solar system.

Although reflected sunlight from a planet is lost in the glare of the star's light, their contrast in the infrared can be hundreds of times greater because planets emit most of their light in the infrared. Even so, a 100 km patch of a planet like the earth is about 100,000 times fainter in the infrared than a similar sized piece of a star's surface. At 8 parsecs, only one photon from such a surface will pass through a one meter square collecting area every 20 minutes. It would take one month to accumulate a single interferogram; and a thousand times that long to complete a full image. For even more distant planets or stars, photons from even continent-sized areas of their surfaces will arrive so infrequently, markings at sub-planetary scales would be virtually invisible.

At other wavelengths, we are left with an even narrower selection of objects bright enough to study. Planets emit virtually no radio waves. Only violent storms on stellar surfaces, supernovae, pulsars, stellar nurseries, maser spots, or galactic cores produce radio waves in detectable quantities. X-rays and ultraviolet light are the preferred spectral regions for studying young supernovae, neutron stars and the mysterious inner regions of black hole accretion disks. The universe seen with Supervision begins to thin out rapidly, with fewer and fewer interesting things to see for each additional dollar invested. A disturbing parallel to modern high-energy physics, where billions will be spent to find a 'missing' quark and the Higgs boson. But at the scale of stars and galaxies, do there still exist new types of phenomena that we haven't already seen? .

A MOVING EXPERIENCE

If we had all the time in the world, even miniscule rates like one photon per minute, with a sufficiently long exposure time, could be amplified to give useable numbers of photons from which to form an image. But virtually everything in the universe is in motion relative to the earth. At each resolution we wish to acheive, image motion sets a maximum limit to how long we can take to accumulate a complete image. A Megavision telescope for studying the surface of nearby stars 8 parsecs distant which move at 10's of kilometers per second, will have no more than a minute or so to complete the image. Supervision interferometers which have a thousand times the resolution will require correspondingly shorter exposure times to avoid image blurring. Even extragalactic studies of galaxies that typically move 100's of km/sec will demand progressively shorter integration times, just as the need for longer exposures to offset the faintness of these sources becomes critical.

One solution to this problem is to greatly increase the collecting area of each telescope in the interferometer array. Every doubling of telescope aperture quadruples the number of photons detected per unit time. In addition, the more telescopes you have in the array, the more interferograms you can form simultaneously. For instance, 10 telescopes provide information for (10x9/2 = ) 45 interferograms while 30 telescopes give (30x31/2=) 435 interferograms during each integration period. These telescopes can also be strategically placed to optimize the array's sensitivity for particular features. A planet like the earth, for example, may only have interesting structure between 100 km and its full diameter, so that at 8 parsecs, telescope separations between 14 and 1400 km are all that are needed to form an acceptable optical image. At the longer infrared wavelengths where planets are brightest, these separations will, of course, have to be 10 to 20 times longer to acheive the same resolution.

BACKGROUND NOISE

Potentially, the most serious limitation to any of these "photon-counting" interferometers is the large number of photons entering the telescopes caused by naturally occurring sources of light unrelated to the object you are trying to study. In the visual and infrared, the earth's atmosphere is the worst of these offenders, producing millions of background photons for every photon of interest from the object. Even from a space-based platform, interplanetary dust and the "zodiacal" light they scatter towards earth from the sun, though a million times less that the atmospheric background, still pose severe problems for studying faint sources. At radio wavelengths, interplanetary and interstellar charged particles form a lumpy haze causing distant radio sources to twinkle, and become indistinct at sufficiently high resolution and long wavelength. The telescope itself can also be a major source of background light for infrared imaging unless the telescope is cryogenically cooled to reduce its heat radiation. Then of course there is the background due to the Big Bang fireball which interferes with observations in the millimeter wavelength band.

So long as your measurements are not limited by the brightness of these diffuse backgrounds, every doubling of your observing time, doubles the number of photons you gather from the distant surface you are studying. But if this is not the case, and you are trying to detect light levels against the background sky limit, a far worse state of affairs takes control. For example, if the background level contributes 10000 photons during each integration period, and your source contributes only 1 photon during the same time, the random fluctuations in the detecyted number of background photons will be equal to the square root of the background level or plus or minus 100 photons. These will completely drown out the 1 photon signal you are trying to detect. The solution from purely statistical considerations is that you will have to accumulate at least 10,000 images for each fringe measurement so that the background noise averages out to plus or minus 1 photon. Eventually, even space-based observations run up against the limit of the interplanetary, interstellar and cosmic background light, necessitating integration times vastly greater than the limit set by the motions inherant in the objects being studied.

THROUGH A LOOKING GLASS...BRIGHTLY

You might think, after all of this, that the situation is pretty bleak. Compliments of Nature's quantum limitations on instrument size, background noise, and surface brightness, most of the regions in our mathematical universe of resolution appear as unreachable as the inside of a black hole. Even so, there is still much that could be done. To a distance of 3000 light years there would be over a billion stars to investigate, and potentially as many solar systems. Continents and weather systems on the planets orbiting the thousand or so nearby stars could under favorable circumstances be mapped and monitored for changes. More significantly, if there is enough light to form such an image, it could also be used to feed a spectroscope from which the planet's atmospheric chemistry could be deduced. The presence of free oxygen would immediatly indicate living organisms.

Long before the first space probes are launched out of our solar system to discover earth-like planets around other stars, astronomers will already have identified all such worlds out to hundreds of light years using Megavision technology, and at orders of magnitude lower costs. Finally, accurate proper motion studies of galaxies will complete a detailed picture of the motion of galaxies in the local universe. The dynamics of individual galaxies will be revealed for the first time, as will the subtelities of the expansion of the universe.

What the final outcome of reaching the Megavision or Supervision frontiers will be is unimaginable, however, it seems unlikely that astronomy as we know it will survive as an open-ended subject. Having more than likely settled the question of the destiny of the universe, and having received weather reports from planets orbiting nearby stars, it is hard to imagine that urgent new questions will remain. Fortunately for the continued vitality of astronomical research, the Megavision frontier is technologically within reach at most wavelengths, and what we discover there will keep us occupied for centuries.