Dear Readers,
Welcome to issue 7 (November 30, 1998) of the MIPS/IRAC GTO newsletter. This issue features two new white paper submissions by William Reach of the SIRTF Science Center and a letter from Marcia Rieke.
The deadline for submissions to the next issue of the newsletter is the 14th of December. It will be the last newsletter of the year, so if you want to enter the new year on a positive note, I encourage you to make a contribution to the next newsletter.
Doug Kelly, editor (dkelly@as.arizona.edu)
A cosmic background radiation due to the energy produced in the earliest stars, which led to the formation of the elements heavier than He as well as the old populations of elliptical galaxies and the bulges of spiral galaxies, is expected to produce an isotropic sky brightness spanning a wide range of wavelengths. Direct photospheric emission appears in the rest-frame visible range, redshifted to the near-infrared, while light absorbed by dust within the host galaxy or protogalaxy will appear as thermal emission from the mid- to far-infrared range. The measurement of this background radiation is of fundamental importance for understanding the origin and early evolution of galaxies. This goal motivated the Diffuse Infrared Background Experiment (DIRBE) on the Cosmic Background Explorer (COBE) mission, which measured the diffuse sky brightness in 10 bands from 1.25 to 240 microns. The results of that mission in the far-infrared, where a significant background was detected, are revolutionizing models of the early evolution of galaxies. The far-infrared background has been partially resolved using follow-up observations with the Infrared Space Observatory, and further follow-up observations with radio and optical telescopes will reveal the nature of these galaxies, which contribute most of the energy output of the epoch of galaxy formation.
Measurement of the cosmic infrared background (CIRB) in the near- to mid-infrared has been thwarted due to the bright foregrounds produced by starlight and zodiacal light. The COBE measurements resulted in upper limits at 1.25 to 100 microns. In the near-infrared, these upper limits are dominated by uncertainties in the subtraction of starlight from the broad-beam (42') observations. Using the Infrared Array Camera (IRAC) on SIRTF, we will completely eliminate this problem because essentially all stars in the Galaxy will be resolved into point sources at high to moderate galactic latitudes. In this proposal we describe an observing program to look `between the stars' and measure the brightness of the CIRB at 3.6 to 8 microns.
We believe the CIRB to be due to the combined emission of galaxies throughout history. The brightness of the CIRB depends on the angular resolution of the observation. An observation with a broad beam will detect the integrated light of all galaxies; this was the strategy of COBE, which was unfortunately limited by our inability to remove starlight accurately. There are two ways that we can improve upon the present situation:
Both of these approaches are possible with SIRTF/IRAC, though both are subject to some technical difficulties.
The diffuse sky brightness will be measured in essentially every IRAC observation. Deep observations will obviously reveal the most sensitive limit, as we will be able to look not only between the stars but between a fraction of the galaxies as well. The brightness of the CIRB will reduce as the sensitivity for detecting point sources increases, until we reach the confusion limit of the telescope, where individual galaxies are no longer resolved from each other even if they are bright enough to be detected. If all stars are detected and excluded, then the sky brightness is presumed to be composed only of zodiacal light and galaxies. The brightness of the zodiacal light is smooth on small angular scales, and it is comparable in magnitude to the total brightness of integrated starlight at 3.6 microns; it must be removed from our observation by subtraction of a model (as described below). If the galaxies brighter than F_0 can be detected and resolved, and if all stars are detected and excluded, then the brightness of the CIRB due to galaxies fainter than the detection limit, I_{CIRB}(F_0) is equal to the integral from 0 to F_0 of (dN/dF)FdF, where (dN/dF)dF is the number of galaxies in the flux range F to F+dF. The value of I_{CIRB}(F_0) for the IRAC detection limit, as well as the source counts above this limit, will be the direct IRAC measurement of the CIRB.
The second strategy is to use COBE/DIRBE as the measurement of absolute sky brightness and to use SIRTF/IRAC as a star detector; the difference between the two is the total CIRB, I_{CIRB}(infinity). This project is very simple in principle but complicated in practice because it requires mapping large fields with IRAC and it requires careful matching of the calibration of the two instruments so that two nearly equal numbers can be subtracted. In this first proposal draft, we will not explore this option in great detail. The strategy would be to make a few large-field (~1 sq.deg), relatively shallow rasters and to extract all of the point sources. The primary waveband would be 3.6 microns, which is similar to a COBE/DIRBE waveband; a reasonable measurement is also possible at 4.5 microns, which is a narrower waveband and shifted from the COBE/DIRBE 4.9 micron waveband. This experiment is probably not worth attempting at longer wavelengths because the COBE/DIRBE 12 micron waveband is much wider and significantly different from the IRAC 8 micron waveband.
The absolute sky brightness down to the confusion limit will be measured as part of the IRAC Deep Survey Guaranteed Time Observation, in which a total of 32 hr of integration will be acquired per sky position. This measurement will be as deep as would ever be needed for the present experiment. The astrophysical limitation in measuring the CIRB will be the accurate subtraction of the zodiacal light. We propose to use a model for the interplanetary dust cloud, carefully calibrated to match the brightness scale of IRAC, to subtract the zodiacal light. To do this, we require accurate sky brightness measurements at a range of ecliptic latitudes or solar elongations. The difference in sky brightness between these various observations will be used to determine the scale factor by which the zodiacal light model must be multiplied to get it onto the same scale as the IRAC measurements.
The primary observing strategy for this project is to use the IRAC Deep Survey together with about two other fields observed through different paths through the interplanetary dust cloud. The new fields, not already included in the Deep Survey, do not need to cover wide areas. A simple strategy would be to observe with the same depth and dither but for a single pointing center (rather than a mosaic). With this strategy, the present project would take approximately 1/32 as long as the Deep Survey. The advantage of observing as deep as the Deep Survey is that both stars and galaxies can be removed to the same depth.
The most important technical limitation in measuring the CIRB is the amount of straylight contamination in the absolute sky brightness. The total observed sky brightness will be:
I_{obs} = (1 + f_{stray}) * ( I_{ZL} + I_{IS} + I_{CIRB}),
where I_{ZL} is the zodiacal light brightness, I_{IS} is the
integrated starlight brightness (for all stars, even bright ones),
and f_{stray} is the straylight fraction. At a design specification
of f_{stray} < 0.2, the SIRTF telescope could limit this project
more severely than the astrophysical uncertainties. Therefore,
we need an accurate measurement of the stray light. Such a
measurement could be obtained for example by observing at various
distances from extremely bright sources (e.g., Earth or Jupiter)
to derive the amplitude of the distant sidelobes of the telescope.
If such observations are not included in the IOC or routine-phase
calibration such that f_{stray} is not known to better than about
+/- 0.03, then the present project will have to include such a
calibration observation.
To Top
This program is designed to measure the total brightness of empty patches of sky, spanning a wide range of wavelengths, ecliptic latitude, and solar elongation. SIRTF is not designed for high sensitivity to very-large-scale phenomena such as the zodiacal light, but such observations will be important for the following practical applications:
The absolute zodiacal light observations are also of scientific interest if they are performed with enough frequency to monitor changes in the sky brightness as SIRTF drifts behind the Earth.
The time for the project described below is about 3000 s per observing epoch, with roughly equal time on MIPS and IRAC and roughly half the time spent moving from position to position. I did not add overhead for each AOT, only integration (ON and OFF) and 3 minutes per slew.
5 * (100 + 180) = 1400s
5 * (10*15 + 180) = 1650s
The most important requirement is that the sky brightness can be compared to `true' darkness. As the IRS does not have a shutter, we have disqualified it from this project. Both IRAC and MIPS contain internal shutters that appear suitable for these observations. For a `typical' zodiacal light brightness, Table 1 shows the estimated SNR in 500 s of integration. In the table, F_{bg} is the flux in a single pixel due to the background, I_{bg} is the surface brightness of the background, and SNR is the signal to noise in 500 s of integration after averaging all pixels in the array (based on the `SIRTF Instrumentation Summary' table on the time estimator web page). Sensitivity is not a critical issue except at 160 microns. The sky brightness at 160 microns is dominated by the interstellar medium rather than zodiacal light, and single-point absolute brightness measurements are of little value for studying cirrus.
Table 1: Signal-to-noise for zodiacal light
Instrument Wavelength F_{bg} I_{bg} SNR
(microns) (mJy) (MJy/sr)
------------------------------------------------------
IRAC 3.5 0.0058 0.17 590
IRAC 4.5 0.010 0.31 890
IRAC 6.3 0.097 2.9 1200
IRAC 8.0 0.31 9.2 2700
MIPS 24 5.1 45 4000
MIPS 70 26 13 1600
MIPS 160 3.7 1.8 3
------------------------------------------------------
For this program, the main emphasis is on the brightness of the zodiacal light, so we will observe a small number of positions spread out on the sky. The goal of observing more than one position is to have a few `tie points' between the sky brightness and a model. The targets are spread out, so we will make a separate AOT for each one (rather than a cluster). To monitor the time-variability of ths sky brightness, we will perform these observations at monthly intervals. It doesn't really matter if the spacing between observations is 1.0 months, so we will not constrain the time too tightly. The basic coordinate system for these observations is (differential ecliptic longitude, solar elongation) or (lambda - lambda_sun,beta). To spread a few points over the sky, we use the following: (90deg,0deg), (105deg,0deg), (-90deg,90deg), (90deg,90deg), (90deg,-90deg). This way we can measure the polar brightness, the north-south polar brightness difference, the in-ecliptic brightness tangent to the Earth's orbit, and the brightness difference leading and trailing the Earth. There are 5 positions and 2 instruments, so there would be 10 AOTs per epoch. The RA and Dec would be calculated for a specific date at the middle of the desired epoch. Each AOT would specify that it is a FIXED TIME observation, to be performed within a 1 week interval.
For example, we could choose November 17 as the center of a measurement epoch. The Sun is at lambda_sun=237deg, so the ecliptic coordinates of the five pointings are [lambda,beta] = [327deg,0deg], [342deg,0deg], [147deg,0deg], [237deg,90deg], [237deg,-90deg]. This observing date needs to be chosen in advance. Then the observer has to look at the IRAS sky maps to make sure there are no stars, galaxies, or interstellar clouds in the field of view; the positions are adjusted by up to a degree to avoid problems. These positions can then be converted to Ra, Dec. In the Universal AOT Front End, Time-Critical Information field, we would enter `Observation to be performed between 15--19 November 2003.'
An alternative approach is to create an ephemeris for each position. Then we could have `bogus' objects in the moving target database that move with the Sun. In this case, one would simply specify a start date and the ephemeris will position the telescope at the appropriate offsets from the Sun. The disadvantage is that one would not be able to fine-tune the pointing to avoid bright sources; therefore, we prefer the solution described in the previous paragraph.
The specially-designed MIPS AOT for total power measurements will allow chopping between the `shutter'--(actually implemented by the scan mirror)--and the sky. (In this AOT, only half of the pixels on the 70 micron array actually view the `shutter', so the SNR is sq.root(2) worse than listed above.) The AOT consists of a simple cycle, ON--SHUTTER--ON--SHUTTER... The only free parameters on the instrument part of the AOT are the integration time per cycle and the number of cycles. If we aim for 50 s integration time, with the standard 10 s per cycle, then that would mean 5 cycles, for a total time (ON+SHUTTER) of 100 s. The values to be chosen here depend on the detailed time-dependent response of the detectors to sudden brightness changes and on the rate of cosmic rays (which eliminate a cycle). We will need to supply information on the detector response to square wave input of various amplitude and period so that observers can decide these parameters. Otherwise, we will all just use standard values that have hopefully been well-chosen by the instrument team.
There is only 1 AOT for IRAC, and its detailed procedure is not presently spelled out in available documents. It is not possible at this time to decide whether the AOT parameters are optimal to perform the total-power measurement to be performed here. However, the observation is clearly possible with the present AOT. The important question is the method and precision of the dark current measurement. Comparable times need to be spent on the dark current and on the sky. If there is only one dark current measurement per AOT, then it might be best to perform a chained set of a few short AOTs, each one performing a dark then a sky measurement. For the Design Reference Mission, we will instead use a single AOT per position per epoch and request an accurate dark current measurement within 12 hours or so. The SNR is very high, and cosmic rays can be easily rejected by robust averaging of all the brightness values in the array. We will use an exposure time of 15 sec and take 10 images. The observing mode will be `single point observation', and `both' F.O.V. will be selected. We will get a total of 20 frames per array because the observing sequence will put first one then the other FOV onto our nominal coordinates. Only 10 of the frames will be at the same coordinates; these could be useful to identify point sources. No dithering or rastering is needed for this observation.
The only significant problems are:
Minor items:
The actions that arose from our discussion are:
Various talks from the previous sessions at the Science Working Group indicated that the following survey areas and strategies are under consideration:
Total time for MIPS portion of survey: ~300 hrs
"Ultra-deep" survey in 2 5'x5' fields to the confusion limit, at 24um and in super-resolution mode at 70um. This survey would complement the "deep" 160um survey by setting similar limits on the total flux produced by resolvable galaxies. Total time is TBD but of order 80-100hrs.
"Ultra-deep" survey of 5x5 arc minutes to detect L* galaxies to z~5.
Total IRAC time of ~100 hrs.
"Shallow" survey covering at least a square degree to account for all the background flux at 3.5um. This survey could be identical to the IRAC coverage needed for the MIPS "deep" survey.
Because of the focal plane layouts of both instruments, surveying in strips will be the preferred mode for efficiency. The MIPS areas will typically be 10'x30'-60'.
Before plunging into a series of talks to learn the latest about deep imaging and surveys, we discussed the issue of low redshift analogues to high redshift galaxies. Recent ISO results such as those reported in astro-ph/9811126 about the low metallicity galaxy SBS0335-052 suggest that we may need to broaden our thinking way beyond models that suppose that galaxy activity is a linear combination of Circinus and M82. Observing some high redshift galaxies with particular properties such as Lyman-break galaxies may also be necessary. We touched on extragalactic projects other than deep surveys only briefly.
Note that HDF-S done in parallel mode so that NICMOS, WFPC2, and STIS observed different fields. Data will be released on Nov. 23.
GO NICMOS HDF data -- same basic picture as from GTO HDF (i.e., nearly identical to I and shorter wavelengths) with most galaxies looking unchanged. A few galaxies show blobby structure in the UV and become normal looking spirals at the longest wavelengths.
Peter pointed out that the number counts in the HDF must be viewed with some caution as the HDF seems to be deficient in infrared-bright galaxies: (comparison is with a survey being executed by R. Elston, Peter, and A. Standford)
To a limit of K~20: Color No. in HDF No. in Elston et al. survey I-K > 4 0.26 / sq. arc min 2.5 / sq. arc min J-K > 1.9 0.6 " 3.4 " " All colors 12 15
Peter also noted that one of the fields in the Elston et al. survey (the Lynx field) will have a 190ksec exposure using AXAF in a collaboration including Rosati. Spinrad, Stern, and Dey will be acquiring Keck spectroscopy.
(see ApJ, Nov 20, 1998 issue for a compendium of COBE articles)
Long-wavelength background
1.17 +/- 0.32 MJy/sr at 140 um
1.09+/-0.2 MJy/sr at 240 um
T= 18.5deg +/-1.2K (from FIRAS data)
Ned described a program in which he had a student (V. Gorjain) find the darkest spot in the DIRBE 2.2um data (e.g. Spot with fewest stars). Gorjian carried out a K-band survey to count all the stars in this 2 sq. degrees. This leaves a residual of 9.5+/-5 kJy/sr at 2.2um (includes all external galaxies). The total K-band luminosity from galaxies in deep surveys is ~5-6 kJy/sr. In IRAC range, extrapolation from K suggests that adding up all galaxies will provide virtually all background -- surveying sq. deg should be adequate.
See http://nova.ias.fr/firback/
Marano 1-4 each .25 sq. deg (Field 1 most analyzed; did not do sub-pixel stepping while other fields observed were substepped). Because of the redundancy in the survey, group is very confident of source detections, and there is no doubt that there is an excess of sources relative to no-evolution predictions.
The initial calibration discrepancies were resolved by understanding how much flux from a source lies in the PSF wings outside the field-of-view of the ISOPHOT C200 array.
Puget doesn't think that SIRTF at 160um will have 10x smaller beam -- these large wings on PSF accounting for 50% of the flux must be due to aberrations and scattering in ISOPHOT.
MIPS needs to be careful about doing a calibration of off-axis PSF
ELAIS N1 Field -- 3 sq. deg
Source counts consistent in all fields, go to 100 mJy,
100 sources per sq. deg
In one .25 -deg area, 8/24 have been identified with ISOCAM sources
Will be launched in Feb-March, 1999.
21-27 um and 9-15 um bands
FOV = 33'x33' 128x128 Si:As BIBs
15.5 arcsec pixels ==> spatial resolution = 24" at 25um, 20" at 12um
Pointing reconstruction -- a couple of arc sec, may be an issue for
source identifications and IRS follow-up.
Of particular interest was Nick's presentation of the WIRE survey strategy. A "moderate" evolution case is the baseline for the initial WIRE observations:
Survey Sky Coverage Flux limits (5-s) No. of galaxies
25um 12 um
Moderate 598 sq. deg 1.5mJy 0.53mJy >50,000 230 places
on the sky
Deep 48 sq. deg 0.56mJy 0.22mJy 16,000
Confusion 0.75 sq. deg 0.34mJy 0.16mJy 500 3 places
Measurement on the sky
Note that WIRE fields will tend to be square in shape.
If early returns indicate that an even stronger evolution model is correct, there may be some limited ability to re-program fields.
Showed a list of candidate fields which overlaps substantially with those being considered for SIRTF.
We talked about several topics such as what groundbased data should be gathered before launch, more on survey strategies, and on tracking new results. Everyone agreed that VLA data would be desirable, but more information is needed before designing an optimum observing strategy. An inventory of existing data on proposed fields is also needed.
After Ned's presentation, it was very clear that many benefits would accrue from have an IRAC survey of the MIPS 1.8 square degrees. The survey fields need to be chosen so as to avoid cirrus and other interfering sources. The shapes of the survey fields need to be considered carefully since the shapes could affect analysis of fluctuations; square regions are possibly preferred. This argument must be balanced against the need to use SIRTF efficiently. Soifer also pointed out that SIRTF visibility should be taken into consideration, with high ecliptic latitudes preferred. A paper by Kaiser (ApJ, 498, p26) addresses sampling strategies for deep surveys.
Bill reach raised the issue of whether or not we need to track the zodiacal background during the mission. This needs more consideration. He also raised the issue of whether such measurements would be accounted against a project's time or whether it could be considered part of calibration. This issue should be raised before a larger group such as the SWG.
The list of issues given at the beginning summarizes the outcome of our
discussion.
To Top