Документ взят из кэша поисковой машины. Адрес оригинального документа : http://www.mso.anu.edu.au/~kcf/pubs_top20/103.pdf Дата изменения: Sun Oct 16 06:14:10 2011 Дата индексирования: Tue Oct 2 01:39:26 2012 Кодировка: Поисковые слова: m 5

The Astrophysical Journal, 493:L53­L57, 1998 Februar y 1
1998. The American Astronomical Society. All rights reser ved. Printed in U.S.A.

CONSTRAINTS ON COSMOLOGICAL MODELS FROM HUBBLE SPACE TELESCOPE OBSERVATIONS OF HIGH-z SUPERNOVAE P. M. Garnavich,1 R. P. Kirshner,1 P. Challis,1 J. Tonry,2 R. L. Gilliland,3 R. C. Smith,4 A. Clocchiatti,5 A. Diercks,6 A. V Filippenko,7 M. Hamuy,8 C. J. Hogan,6 B. Leibundgut,9 M. M. Phillips,5 D. Reiss,6 A. G. Riess,7 B. P. Schmidt,10 R. A. Schommer,5 J. Spyromilio,9 C. Stubbs,6 N. B. Suntzeff,5 and L. Wells8
Received 1997 October 14; accepted 1997 December 9; published 1998 January 14

ABSTRACT We have coordinated Hubble Space Telescope (HST) photometr y with ground-based discover y for three supernovae: Type Ia supernovae near z 0.5 (SN 1997ce, SN 1997cj) and a third event at z 0.97 (SN 1997ck). The superb spatial resolution of HST separates each supernova from its host galaxy and leads to good precision in the light cur ves. We use these light cur ves and relations between luminosity, light-cur ve shape, and color calibrated from low-z samples to derive relative luminosity distances that are accurate to 10% at z 0.5 and 20% at z 1. When the HST sample is combined with the distance to SN 1995K (z 0.48), analyzed by the same precepts, we find that matter alone is insufficient to produce a flat universe. Specifically, for Q m QL 1, Q m is less than 1 with more than 95% confidence, and our best estimate of Qm is 0.1 0.5 if QL 0. Although this result is based on a ver y small sample whose systematics remain to be explored, it demonstrates the power of HST measurements for high-redshift supernovae. Subject headings: cosmology: obser vations -- galaxies: distances and redshifts -- supernovae: general -- supernovae: individual (SN 1995K, SN 1997ce, SN 1997cj, SN 1997ck)
1. INTRODUCTION

Direct measurement of global cur vature and deceleration of the universe has challenged the best efforts of obser vers for many decades (Humason, Mayall, & Sandage 1956; Baum 1957; Sandage 1988). Progress has been stymied by a lack of reliable standard candles and yardsticks and the difficulty of making precise measurements on faint objects at high redshift. Recent advances now offer the hope of solving this classical problem: the empirical calibration of Type Ia supernovae (SNe Ia) as precise distance indicators, and new technology that allows the measurement of supernova (SN) properties at large distances. In addition, search strategies pioneered by Hamuy et al. (1993) and Perlmutter et al. (1997a) regularly yield a number of SNe that can be efficiently studied using prescheduled telescopes. This Letter reports results from coordinated ground-based and Hubble Space Telescope (HST) obser vations of distant SNe up to z 1, which extend the luminosity-distance relation to redshifts where cosmological effects can be measured (NЬrgaard-Nielsen et al. 1989; Perlmutter et al. 1997b; Schmidt 1997). Type Ia SNe proved to be excellent distance indicators. They are not perfect standard candles, but their luminosities correlate with light-cur ve shape so that differences in intrinsic brightness can be assessed by obser ving SN light cur ves (Phillips 1993;
1 Har vard-Smithsonian Center for Astrophysics, 60 Garden Street, Cambridge, MA 02138. 2 Institute for Astronomy, University of Hawaii, Manoa, HI 96822. 3 Space Telescope Science Institute, 3700 San Martin Drive, Baltimore, MD 21218. 4 University of Michigan, Department of Astronomy, 834 Dennison, Ann Arbor, MI 48109. 5 Cerro Tololo Inter-American Obser vator y, Casilla 603, La Serena, Chile. 6 Department of Astronomy, University of Washington, Seattle, WA 98195. 7 Department of Astronomy, University of California, Berkeley, CA 94720. 8 Steward Obser vator y, University of Arizona, Tucson, AZ 85721. 9 European Southern Obser vator y, Karl-Schwarzschild-Strasse 2, Garching, Germany. 10 Mount Stromlo and Siding Spring Obser vator y, Private Bag, Weston Creek P.O., Australia.

Hamuy et al. 1996, hereafter H96; Riess, Press, & Kirshner 1996, hereafter RPK). The techniques of RPK and H96 produce relative distance estimates to better than 10% despite a spread of than 1 mag in intrinsic brightness. Colors of SNe Ia provide a way to correct for extinction by dust in both our Galaxy and the host. Since cosmological models predict the shape of the relation between luminosity distance and redshift, relative distances constrain cur vature and deceleration independent of the Hubble constant. By using the same methods at high and low redshift, we minimize the systematic errors that could undermine an enterprise of this type. Even evolutionar y effects, which have bedeviled all previous attempts to measure q0, can be calibrated by studying contemporar y samples in populations of different ages and metallicity; Schmidt et al. (1998, hereafter SSP98) demonstrate that any residual population-dependent bias in luminosity is less than m M 0.06 mag. Our team has undertaken a program to discover and study SNe Ia at z 1 0.3 (Leibundgut et al. 1996; Riess et al. 1997). Results from our first discover y (SN 1995K at z 0.48), a detailed description of our techniques, and a discussion of possible systematic errors are given by SSP98. Here, we present preliminar y results that combine HST and ground-based photometr y. We analyze light cur ves of three SNe with redshifts between 0.4 and 1.0 using the multicolor light-cur ve shape (MLCS) method (RPK) and a template-fitting technique (Hamuy et al. 1995) to determine distances. We produce a consistent Hubble diagram in the range 0.01 ! z ! 1.0 and use the results to constrain cosmological models. We also compare our conclusions with the results from seven high-z SNe studied from the ground by Perlmutter et al. (1997b).
2. OBSERVATIONS

The search for distant SNe was conducted at the CanadaFrance-Hawaii Telescope (CFHT) on 1997 April 29 and 30 (UT) using the UH8K CCD mosaic. Four fields, selected and scheduled long in advance for HST visibility, were imaged in broadband V and I filters, for a total search area of about 1 L53


L54

GARNAVICH ET AL.

Vol. 493

deg2. We compared these images with template frames of the same regions taken April 4 and 9 (UT) to detect variable objects. Seeing on the search frames was 0 . 6 and 0 . 7 (FWHM) on the template frames. A minimum of three 1200 s exposures were taken of each field and combined with a median filter. The magnitude limits were typically I ! 2 4.5 . In software, the two sets of images were aligned, convolved to match pointspread functions, scaled, and subtracted. The subtracted images were then searched in software and inspected by eye for residual point sources. Twelve SN candidates were identified from the search. Two objects had blue V I colors unlike any high-redshift SN Ia. Spectra of eight of the remaining candidates were obtained with the Multiple-Mirror Telescope on May 1 and 2 and with the Keck 10 m telescope on May 4. Four of the objects were identified as SNe Ia (Tonr y et al. 1997). We selected SNe 1997ca, 1997ce, 1997cj, and 1997ck for HST obser vations based on the HST scheduling requirement to have exactly one target in each of the four fields. Subsequent photometr y of SN 1997ca indicated that it was in fact a SN II at low redshift. Our first WFPC2 images were obtained on 1997 May 12, just a week after the Keck spectra. We obser ved each SN in the WF3 chip on six visits spanning approximately 3 weeks in the SN rest frame. Each epoch was allotted one HST orbit. For the z 0.5 targets, the orbit was divided into an 800 s exposure in the F675W filter and a 1100 s exposure with the F814W filter. These filters approximate standard BV bandpasses at z 0.5, and we have computed K-corrections for both the HST and ground-based obser vations according to SSP98. At z 1, the F850LP is well matched to the rest-frame B band. Exposures of SN 1997ck in the F850LP filter fill the target visibility window with a minimum total exposure of 2200 s. All observations were divided into two exposures, and we combined the cosmic-ray split images using the default parameters in the

STSDAS/CRREJ algorithm, which is designed to avoid confusing stellar images with cosmic rays in undersampled data. As a check, we performed aperture photometr y on bright, unsaturated stars obser ved at each epoch and found a scatter of less than 0.01 mag, consistent with the predicted statistical error. Figure 1 (Plate L7) displays images of each SN made by adding all the obser vations. We calibrated a sequence of stars near each SN using both HST and ground-based data. Magnitudes of stellar objects in the HST images were estimated using the prescription of Holtzman et al. (1995). Data numbers within a 0 . 3 radius aperture were summed, and we subtracted the background level estimated from a large region around the image. The small aperture minimized background noise. An aperture correction based on the point-spread function (PSF) of stars in the field brought the measurement to the equivalent of a 0 . 5 aperture, and the result converted to a magnitude in the HST filter system. For the F850LP filter, only a synthetic zero point (ZP) was available, which is estimated by Whitmore (1995) to be good to 3%. Obser vations of standard stars and three of the HST fields were obtained in the R and I bands under photometric conditions on three nights with the Hawaii 88 inch telescope in 1997 May. The F675W and F814W magnitudes for 15 stars in common with the Hawaii data were converted to RI magnitudes using ZPs and second-order color terms provided by Holtzman et al. (1995). There is a significant color residual between the two calibrations, but for the seven stars with colors similar to those of the SNe (R I ! 1.2 ), the average difference between ground-based (GRD) and HST magnitude estimates is only R GR D RHST 0.02 0.05 mag and IGR D IHST 0.00 0.0 4 mag, verifying the Holtzmann ZPs. From the ground, light from the host galaxy is a major source of uncertainty in measuring SN light cur ves. A template image

TABLE 1 Supernova Parameters and Error Budget Parameter 1995Ka (z 0.48) 1997ce (z 0.44) 1997cj (z 0.50) 1997ck (z 0.97)

Error Budget CT correction (mag) . Zero point (mag) . . . . . Evolutionb (mag) . . . . . Selection biasa (mag) Weak lensingc (mag) . K-correctionsd (mag) . Light-cur ve fit (RPK)e Light-cur ve fit (H96)e j of SNe Iaf (mag) . . .
max B max V

....... ....... ....... ....... ....... ....... (mag) (mag) .......

. . . . . .

. . . . . . . . ..

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

0.00 0.03 0.06 0.02 0.02 0.06 0.21 0.13 0.12

0.02 0.05 0.06 0.02 0.02 0.06 0.08 0.11 0.12 Parameters

0.02 0.05 0.06 0.02 0.02 0.06 0.19 0.11 0.12

0.02 0.05 0.06 0.02 0.04 0.06 0.42 0.21 0.12

m (mag) (H96) . . m (mag) (H96) . . D (mag) (RPK) . . . . . Dm15 (mag) (H96) . . AV (mag) (RPK) . . . . Galactic AV (mag) . . m M (mag) (RPK) m M (mag) (H96)g
a b c d e f g

. . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

22.89 23.00 0.07 1.15(09) 0.00 0.00 42.40(26) 42.25(20)

22.75 22.79 0.41 1.19(06) 0.00 0.03 41.83(18) 42.06(19)

23.19 23.19 0.04 1.16(03) 0.00 0.03 42.59(25) 42.48(19)

24.78 ) 0.01 1.00(17) ) 0.03 44.15(45) 44.06(27)

From SSP98. Current obser vational upper limit from SSP98. From Wambsganss et al. 1997. Includes propagated effect on extinction, 3.1jE(B V). Includes uncertainty in light-cur ve fit, extinction, and background subtraction. Dispersion in local SN Ia distances after LCS correction (RPK). Average of B and V.


No. 2, 1998

CONSTRAINTS ON COSMOLOGICAL MODELS

L55

Fig. 2.--Rest frame B and V light cur ves of SNe 1997ce (top left), 1997cj (top right), and a B light cur ve for SN 1997ck (bottom). Filled points represent HST obser vations, while open points are from ground-based telescopes. The solid lines show the best-fit light cur ve from the MLCS method.

is often taken 1 yr after the SN maximum to subtract the host light. HST allows the SN and the galaxy light to be more easily separated, but the host background must still be removed. We create a template by subtracting the SN image from the HST data in software and assume no additional point sources exist at the position of the SN. At these large redshifts the hosts appear smooth so that this assumption is valid. For SNe 1997ce and 1997cj, an empirical PSF derived from nearby stars in the field was scaled to the peak brightness of the SN and subtracted from each obser vation. The four pixels at the SN position were smoothed and all the epochs combined to produce one high signal-to-noise ratio image. This template was subtracted from each HST epoch, leaving only the SN and other stars in the image to be measured with aperture photometr y. The PSF subtraction is not perfect, and we have estimated, through simulations, the uncertainty in the background (always less than 2%) and include this error with the uncertainty in each photometric measurement. The host galaxy for SN 1997ck was insignificant in the F850LP filter, so no correction was necessar y. Reduction of the ground-based data followed the procedures outlined by SSP98. The HST image templates were used to subtract the host galaxy background, and the relative brightness between the SN and other stars in the field was measured for each frame using PSF fitting routines in DoPHOT (Schechter, Mateo, & Saha 1993). Artificial stars were added and measured to estimate the uncertainty of each measurement, with these errors added in quadrature to those resulting from the imperfect template. In both cases, the ZP (and color term for the ground-based data) were determined using the F675W and F814W magnitudes of the field stars on the HST frame. K-corrections were applied to both the HST and ground-based data to bring the F675W and F814W magnitude estimates to rest-frame B and V, respectively. We list sources of statistical and possible systematic errors in Table 1. The total photometric uncertainty (including esti-

Fig. 3.--Hubble diagram for SNe Ia. The top panel shows the MLCS distance modulus vs. z for a large sample of low-redshift events and the four high-z SNe. The lower panel plots the magnitude difference between the obser ved SNe and the magnitude expected from a universe with Qm QL 0. Open circles show the SNe distances estimated using the techniques and scale from H96. The redshift for SNe where both techniques have been applied has been slightly shifted for clarity.

mated error in the extinction correction) is given as an error in the light-cur ve fit. Selection bias (includes Malmquist bias) results from having brightness and temporal discover y limits, but the error is small owing to the tight dispersion in the corrected intrinsic luminosities of SNe Ia (SSP98). The lumpiness of the universe causes weak gravitational lensing of distant SNe, and an estimate of the systematic effect has been made by Wambsganss et al. (1997). The direct uncertainty from Kcorrections is small (2%) but is magnified by a factor of 3 because of the propagation of color errors interpreted as extinction (SSP98). Even after correcting SNe luminosities using light-cur ve shape methods, there remains a dispersion of 0.12 mag (RPK) in the derived distances that is included in the error budget. The Holtzman calibration is expected to have a zeropoint uncertainty of 3%, but recent comparisons of Holtzmann magnitudes with ground-based data suggest that ZP offsets and uncertainties of 0.05 mag are possible, and errors this large cannot be excluded with our current ground-based data. Inefficient charge transfer (CT) in WF3 causes the apparent brightness of objects to var y with pixel position. Whitmore & Heyer (1997) show that the loss of charge along columns can be as large as 7% and depends on object brightness and background level as well as pixel position. We applied their corrections to all the magnitude estimates after interpolating to a 0 . 3 aperture. This made the estimated SN magnitudes brighter by between 3% and 5%, with the corrections being good to about 2%.


L56

GARNAVICH ET AL.

Vol. 493

The total uncertainty in the derived distances is found by adding the individual errors shown in the Table 1 error budget in quadrature. The distance errors for the MLCS and templatefitting methods are ver y consistent except for SN1997ck, where the MLCS suffers from having only a B light cur ve to fit.
3. RESULTS AND DISCUSSION

The HST image shows that SN 1997ce occurred 0 . 4 south of the brighter of a pair of elongated galaxies. Keck spectra of SN 1997ce display a blue continuum with broad absorption bands that match those of a SN Ia at z 0.44. The light cur ve shows that SN 1997ce was discovered about 8 days before maximum light in the obser ver's frame. The MLCS method was used to analyze the rest-frame light cur ves (Fig. 2) and found that SN 1997ce declined slightly faster than a normal SN Ia (MLCS parameter D 0.41 mag). Template fitting agreed but found a smaller correction (Dm 15 1.19 mag). Differences in the derived luminosity between the two methods are not unusual for the nearby sample of SNe and underscore the benefit of increasing the sample size. The MLCS fit found no reddening of this SN, consistent with the color at maximum of B V 0. The host galaxy of SN 1997cj is a spiral with the SN offset 0 . 7 to the west. The MMT spectrum of SN 1997cj showed features consistent with a SN Ia at a redshift of 0.5, and a narrow emission line of [O ii] l3727 provided a precise redshift of z 0.50. Discover y was about 12 obser ver days before maximum light. Although the first planned HST visit was delayed due to a lack of guide stars, the light cur ves (Fig. 2) are still well defined by combined ground-based and HST data and show that SN 1997cj had a normal decline rate. MLCS fits to the B and V light cur ves show that the object is not reddened. The Keck spectrum of SN 1997ck was too weak to show the broad features of the SN, but strong [O ii] l3727 emission from the host galaxy indicated a redshift of z 0.97. The V I color was consistent with a SN Ia before maximum. The host is not easily seen in the F850LP HST images, but in the R and I ground-based frames it appears as a ver y elongated, low surface brightness patch extending 2 to the northwest. The blue color of the host may indicate a population of young stars, and there is a corresponding possibility of dust extinction. Because of the high redshift, only a rest-frame B light cur ve could be constructed. The color derived from the difference between the I and the F850LP filters is consistent with no reddening. However, the wavelength baseline is small, and the extinction to SN 1997ck remains uncertain. The SN was discovered 11 days before maximum light in the obser ver's frame (Fig. 2). The MLCS method, working with only one bandpass, finds that SN 1997ck has a normal decline rate. Template fitting suggests that this is a slightly overluminous SN. Although the evidence that this is truly a SN Ia is less certain for SN 1997ck than for the other objects, the light cur ve looks exactly like a SN Ia light cur ve at z 0.97 and the premaximum color is also consistent. But, as with some of the early obser vations of Perlmutter et al. (1997b), we do not have the data to exclude

conclusively other possibilities. We treat SN 1997ck as an unreddened SN Ia when we include it in the discussion that follows. Table 1 summarizes the error and distance estimates for this sample combined with SN 1995K from SSP98. A Hubble diagram reaching to z 1 is shown in Figure 3. The data adhere closely to the expectations of relativistic cosmology as shown by the model cur ves. The lower panel in Figure 3 compares the data with flat and open cosmological models containing nonrelativistic matter, Qm, and a cosmological constant, QL (Carroll, Press, & Turner 1992). First, we consider only the three confirmed SNe Ia that have rest-frame B and V light cur ves. With the constraint of a flat universe (Q m QL 1), minimizing x2 for the combined set 03 of low-z and high-z SNe, we find Q m 0.4 0..3 for the MLCS 0.3 technique and Q m 0.3 0.3 using the template-fitting method. Alternatively, with QL set to zero, we derive Q m 0.1 0.5. From either method we find that Q m ! 1 with 95% confidence. Including SN 1997ck in the analysis tightens the constraints on Qm and QL but does not alter the best-fit values. Our data suggest that the matter density is low. Either the universe is open or flat; if flat, then a cosmological constant makes a considerable contribution (which may conflict with limits from gravitational lensing statistics; Kochanek 1996). These conclusions agree with those from dynamical estimates of the density of clustered matter (Lin et al. 1996; Carlberg et al. 1996) and from the comparison of the Hubble time with estimates of the nuclear burning ages of globular clusters (Reid 1997). Our results contrast with those of Perlmutter et al. (1997b), which preferred a high matter density even in flat models. Although there are cosmological models acceptable to both data sets at the 68% level, the probability that the two samples have the same parent distribution is less than 10%. More SNe at z 1 0.4 , colors and spectra for all new objects, and a detailed comparison of the two approaches should resolve this apparent disagreement. We have shown that SNe with redshifts as large as z 1 can be discovered and successfully studied with a combination of ground-based telescopes and HST. Refinements will be made to this data set once template images are acquired with HST, and additional ground-based photometric data are obtained. Our initial sample of four SNe is inconsistent with a high matter density Q m 1, although the strength of these conclusions should be tempered by the less than perfect data set for SN 1997ck and the small size of our present sample. Additional objects will allow us to increase the precision of our measurement and test for sources of systematic error. We are ver y grateful to STScI director, R. Williams, for granting the HST director's discretionar y time. We thank S. Jha, G. Luppino, J. Jensen, R. Lucas, A. Patterson, D. Harmer, L. Cowie, E. Hu, D. Rawson, and J. Mould for their assistance. The High-z Supernova Team is supported by NASA through a grant from the Space Telescope Science Institute, which is operated by the Association of Universities for Research in Astronomy, Inc., under NASA contract NAS5-26555.

REFERENCES Baum, W. A. 1957, AJ, 62, 6 Carlberg, R., et al. 1996, ApJ, 462, 32 Carroll, S. M., Press, W. H., & Turner, E. L. 1992, ARA&A, 30, 499 Hamuy, M., et al. 1993, AJ, 106, 2392 Hamuy, M., Phillips, M. M., Maza, J. Suntzeff, N. B., Schommer, R. A., & Aviles, R. 1995, AJ, 109, 1 ґ Hamuy, M., Phillips, M. M., Suntzeff, N. B., Schommer, R. A., Maza, J., & Aviles, R. 1996, AJ, 112, 2398 (H96) ґ Holtzman, J. A., Burrows, C. J., Casertano, S., Hester, J. J., Trauger, J. T., Watson, A. M., & Worthey, G. 1995, PASP, 107, 1065 Humason, M. L., Mayall, N. U., & Sandage, A. R. 1956, ApJ, 61, 97 Kochanek, C. 1996, ApJ, 466, 638


No. 2, 1998

CONSTRAINTS ON COSMOLOGICAL MODELS

L57

Leibundgut, B., et al. 1996, ApJ, 466, L21 Lin, H., Kirshner, R. P., Shectman, S. A., Landy, S. D., Oemler, A., Tucker, D. L., & Schechter, P. L. 1996, ApJ, 471, 617 NЬrgaard-Nielsen, H. U., et al. 1989, Nature, 339, 523 Perlmutter, S., et al. 1997a, in Thermonuclear Supernovae, ed. P. Ruiz-Lapuente, R. Canal, & J. Isern (Dordrecht: Kluwer), 749 ------. 1997b, ApJ, 483, 565 Phillips, M. M. 1993, ApJ, 413, L105 Reid, I. N. 1997, AJ, 114, 161 Riess, A. G., Press, W. H., & Kirshner, R. P. 1996, ApJ, 473, 88 (RPK)

Riess, A. G., et al. 1997, AJ, 114, 722 Sandage, A. R. 1988, ARA&A, 26, 561 Schechter, P. L., Mateo, M., & Saha, A. 1993, PASP, 105, 1342 Schmidt, B. P. 1997, in Thermonuclear Supernovae, ed. P. Ruiz-Lapuente, R. Canal, & J. Isern (Dordrecht: Kluwer), 765 Schmidt, B. P., et al. 1998, ApJ, submitted (SSP98) Tonr y, J., et al. 1997, IAU Circ. 6646 Wambsganss, J., Cen, R., Xu, G., & Ostriker, J. P. 1997, ApJ, 475, L81 Whitmore, B. 1995, in STSci Workshop, Calibrating HST: Post Ser vice Mission, unpublished Whitmore, B., & Heyer, I. 1997, HST Instrument Sci. Rep., WFPC2 97-08


Fig. 1.--HST images of SNe 1997cj (top), 1997ce (center), and 1997ck (bottom). The full WF3 field (1 . 25) is shown on the left, while the right panel shows a 3 times magnification centered on the SN. Garnavich et al. (see 493, L54)

PLATE L7