FIRST RESULTS FROM THE La Silla-QUEST SUPERNOVA SURVEY AND THE CARNEGIE SUPERNOVA PROJECT

The La Silla/QUEST survey started the Low Redshift Supernova Search in 2011 December (Baltay et al. 2013). The survey uses the 1 m ESO Schmidt telescope at the La Silla Observatory in Chile with the 10 square-degree QUEST camera (Baltay et al. 2007) located at the prime focus. The camera consists of 112 CCD detectors with $600\times 2400$ pixels each. The pixels are 13 μm square and correspond to 0.87 arcsec per pixel. With typically 60 s exposures in a broad g + r filter (4000–7000 Å), the limiting magnitude is around 21.5. About 1000 square degrees are scanned on a clear night and each field is imaged at least twice during the night at two-hour intervals to reduce contamination from Solar System objects that move across the image. The nightly pattern is repeated with a two day cadence. The search is blind in that it is sensitive to supernovae regardless of their proximity to or the type of host galaxies. Reference images, consisting of co-added images taken at least two weeks before the discovery images, are appropriately normalized and subtracted from the discovery images in order to isolate candidate transients. These initial candidates are filtered by imposing cuts based on the signal-to-noise ratio, point-spread function (PSF), and other shape parameters in order to remove noise artifacts and stellar transients. Images of each of the remaining candidates are visually inspected, along with plots of their historical light curves measured from previous LSQ survey images, to select candidates for spectroscopic classification. Of the candidates for which spectra were taken, 84% turned out to be supernovae.

The spectroscopy used to classify the supernova candidates was carried out using five different telescopes. The spectra taken for this sample of supernovae that had peak brightness before 2013 May are summarized in Table 1. The scarcity of spectroscopy time was one limiting factor in the survey. When spectroscopy time was available, those candidates were chosen for spectroscopy that were deemed to be the youngest after supernova explosion. The spectra were classified using the SNID (Blondin & Tonry 2007) and GELATO (Harutyunyan et al. 2008) supernova classification programs. For good signal-to-noise spectra, the two were found to give consistent results. All of these supernova classifications were announced promptly in the online reports of the Astronomers Telegram (ATELs).

The 1 m Swope telescope at the Las Campanas Observatory was used to follow the SNe Ia with optical imaging to obtain their light curves. The SITe3 CCD camera, a 2048 × 3150 CCD array with 15 μm pixels that correspond to a plate scale of 0.435 arcsec pixel⁻¹, was used for these observations. To speed up the readout time, only the central 1200 × 1200 pixels were read out. Exposure times were typically 5–10 minutes near peak, longer for fainter supernovae and for supernovae long past peak, with nominally each of six filters, the $B,V$ (Bessell 1990), and the SDSS (Fukugita et al. 1996) $u,g,r$, and $i$. This telescope and instrument have been calibrated with great care over the course of previous CSP campaigns (Hamuy et al. 2006; Contreras et al. 2010; Stritzinger et al. 2011). The extinction coefficients and color terms have been carefully measured and shown to be extremely stable over many years. These color terms and extinction coefficients will be updated (K. Krisciunas et al. 2015, in preparation). The better than 1% calibration of this instrument and the typically 1 arcsec seeing at the Las Campanas site play an important role in obtaining light curves of the highest quality.

The 2.5 m du Pont telescope, also located at the Las Campanas Observatory in Chile, has been used both to take spectra to classify supernova candidates and to take the final template images of the host galaxy after the supernova has faded to allow the subtraction of the host galaxy light from the measured light curve. For the spectroscopy, the WFCCD spectrometer was used with a wavelength range of 3800–9200 Å. For template imaging the detector used is a SITE2 2048 × 2048 CCD with a plate scale of 0.259 arcsec pixel⁻¹. The du Pont is preferred to the Swope to take the generally fainter host galaxy images due to its larger size and better seeing. For these template exposures, the same physical filters were used as were used for the photometric observations with the Swope telescope, facilitating good subtractions. The du Pont telescope was also used with the RetroCam camera to take infrared images of a subset of these supernovae in the $Y,J$, and $H$ bands for another part of the CSP II low-redshift supernova project.

As our first step, images are bias-subtracted and flat-fielded, after which point an astrometric solution is applied. Linearity and exposure time variation corrections are also applied. PSF instrumental magnitudes (m_inst) are then measured for a number of selected field stars (local-sequence stars) common to all the images for a given target. We use an IRAF¹⁰ -based pipeline developed by the Carnegie Supernova Project. These local-sequence magnitudes are calibrated using catalog standards (by Landolt 1992, for the B and V filters and by Smith et al. 2002, for $u,g,r,i$) on at least three photometric nights to obtain measurements of the zero points. Photometric nights are selected by determining the zero points for all of the standard stars during the night and requiring that the scatter in these zero points be less then 0.03 mag for the $B,V,g,r,i$ bands and less then 0.08 mag for the u band. For this calibration, the Landolt and Smith standards are converted to the natural system of the Swope telescope. This means that we do not transform our observations to the systems of the standard catalogs, which would require the use of S-corrections (Stritzinger et al. 2002). Instead, we transform the catalog standards to the Swope system using stable color terms that have been well characterized from years of extensive observations of stellar standards with the Swope telescope (see Equations (1)–(6) of Contreras et al. 2010). For a given band pass, the transformation from m_inst to the Swope natural magnitude system (m_nat) is then given by

$$\begin{eqnarray*}&&{m}_{i,\mathrm{nat}}={m}_{i,\mathrm{inst}}-{k}_{j}{X}_{i}+{\mathrm{zpt}}_{j},\end{eqnarray*}$$

where zpt_j is the zero point measured for a given photometric night, X_i is the air mass of that field star, and k_j is the air mass extinction coefficient for a band pass j. We use extinction coefficients that have been measured for the Swope telescope site and found to be remarkably stable over the years (see Figure 3 of Contreras et al. 2010).

Before measuring supernova magnitudes, we remove the host-galaxy light by subtracting a host galaxy template image taken after the supernova has faded to a negligible level (>300 days after maximum light). These template images were deeper and taken with seeing conditions that match or exceed the seeing conditions for the light-curve images, and were typically taken at the larger du Pont telescope, and occasionally with the Swope telescope, both at the Las Campanas Observatory in Chile. In either case, the same physical filters are used as for the supernova light-curve exposures. Two algorithms were used to carry out these galaxy subtractions. One was a fast automated procedure using the HOTPANTS image subtraction program (Becker et al. 2004). All supernovae were run through this fast method. The subtracted images were examined to assess how well the subtraction worked. About two-thirds of the supernovae were judged to be successfully subtracted by this method. The remaining third were then processed by a slow, manually guided subtraction procedure to obtain satisfactory subtractions.

Using the procedure described above, light curves for each of the supernovae in the sample, using PSF fitting for aperture-independent photometry, were obtained in each of the filters in the Swope natural system. The light curves for supernova LSQ13ry are shown in Figure 3. The light curves for the remaining 30 supernovae in this sample are shown in Figure 11 in the appendix. The lines on these figures are the SALT2.4 fits. On the whole (but see comment in the last paragraph of Section 4.4), the light curves are well sampled and well fitted by the SALT2.4 templates (outlier points are ignored in these fits). All of the light curves in the sample are presented in numerical form in Table 2 (the first 10 lines are in the printed version, the remainder are available electronically).

Zoom In Zoom Out Reset image size

We use the SALT2.4 supernova template fitting program (Guy et al. 2007; Betoule et al. 2014) to fit the light curves to obtain the best estimate of the rest-frame B-band peak magnitude for each supernova, the width of its light curve, and its color. SALT2.4 has a large collection of supernova spectra at various redshifts and phases from previous surveys. It uses these templates to fit the light curves in several filter bands and performs the K corrections to convert the filter bands to the rest frame of the supernova. SALT2.4 also has the effective transmission curves of the filters used at several telescopes, including Swope, and thus is able to fit the light curves in the Swope natural system. The inputs to SALT2.4 are the light curves in several filters, the heliocentric redshift, and the Milky Way extinction for each supernova. The Milky Way extinctions are taken from the dust maps of Schlegel et al. (1998) with the Schlafly & Finkbeiner (2011) recalibration. The output of SALT2.4 is the peak magnitude ${{m}_{B}}^{*}$ in the rest-frame B band, the stretch factor x₁, the color c, and the time of the peak magnitude of the supernova. The color c equals the excess $B-V$ color over the natural $B-V$ of a typical SNe Ia in the SALT synthetic spectrum.

The values of these output parameters for our sample are summarized in Table 3. The distribution in the stretch parameter x₁ from SALT2.4 is given in Figure 4 and the distribution in the color parameter c is given in Figure 5. Superimposed on these figures are the x₁ and c distributions from other supernova samples, some at low redshift (0.01 < z < 0.1) and some at higher redshifts. The level of agreement between these samples is discussed in Section 5.1 below.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Table 3.  Parameters for the LSQ-CSP Supernova Sample^a

Object	R.A.	Decl.	zh	δz	zCMB	${m}_{B}$ a	$\delta {m}_{B}$ a	c	δc	x1	δx1	$E(B-V)$
LSQ11bk	04:20:44.25	−08:35:55.75	0.037	0.005	0.037	16.901	0.018	−0.097	0.017	1.086	0.040	0.100
LSQ11ot	05:15:48.34	06:46:39.36	0.027	0.0001	0.027	17.822	0.018	0.293	0.018	−0.073	0.036	0.167
LSQ11pn	05:16:41.54	06:29:29.40	0.033	0.0001	0.033	17.413	0.019	0.136	0.019	−3.016	0.073	0.152
LSQ12agq	10:17:41.67	−07:24:54.45	0.064	0.0001	0.066	18.494	0.022	0.078	0.020	0.267	0.071	0.038
LSQ12aor	10:55:17.64	−14:18:01.38	0.093	0.0001	0.094	19.447	0.021	−0.002	0.021	−2.528	0.106	0.042
LSQ12bld	13:42:44.03	08:05:33.74	0.083	0.0001	0.084	19.022	0.020	0.048	0.020	−0.847	0.070	0.023
LSQ12blp	13:36:05.59	−11:37:16.87	0.074	0.0001	0.075	18.333	0.030	−0.054	0.025	−0.357	0.093	0.049
LSQ12btn	09:21:30.47	−09:41:29.86	0.054	0.0001	0.055	18.242	0.019	0.075	0.017	−1.578	0.054	0.033
LSQ12ca	05:31:03.62	−19:47:59.28	0.098	0.0001	0.098	19.122	0.021	−0.016	0.018	−0.010	0.160	0.043
LSQ12cdl	12:53:39.96	−18:30:26.16	0.111	0.001	0.112	19.171	0.022	−0.108	0.021	0.554	0.267	0.050
LSQ12fuk	04:58:15.88	−16:17:58.03	0.020	0.001	0.020	15.842	0.029	0.102	0.025	0.923	0.042	0.073
LSQ12fvl	05:00:50.04	−38:39:11.51	0.056	0.0001	0.056	18.668	0.027	0.214	0.026	−3.177	0.133	0.018
LSQ12fxd	05:22:17.02	−25:35:47.01	0.031	0.0001	0.031	16.315	0.024	−0.024	0.020	0.896	0.073	0.022
LSQ12gdj	23:54:43.32	−25:40:34.09	0.030	0.0001	0.029	15.831	0.026	−0.085	0.023	0.980	0.028	0.020
LSQ12gef	01:40:33.70	18:30:36.38	0.065	0.005	0.064	18.975	0.027	0.327	0.025	0.976	0.207	0.053
LSQ12gln	05:22:59.41	−33:27:51.32	0.112	0.005	0.112	18.962	0.024	−0.123	0.022	0.552	0.121	0.020
LSQ12gpw	03:12:58.24	−11:42:40.13	0.058	0.005	0.057	17.432	0.026	−0.032	0.025	2.386	0.240	0.063
LSQ12gxj	02:52:57.38	01:36:24.25	0.036	0.0001	0.035	18.087	0.027	0.262	0.024	0.896	0.048	0.059
LSQ12gyc	02:45:50.07	−17:55:45.74	0.093	0.001	0.092	18.768	0.025	−0.058	0.024	0.945	0.251	0.021
LSQ12gzm	02:40:43.61	−34:44:25.87	0.100	0.0001	0.099	19.514	0.027	0.021	0.029	0.356	0.344	0.021
LSQ12hjm	03:10:28.72	−16:29:37.08	0.070	0.005	0.069	18.266	0.025	−0.111	0.023	−0.358	0.069	0.025
LSQ12hno	03:42:43.25	−02:40:09.76	0.048	0.0001	0.047	17.920	0.027	0.056	0.024	0.187	0.155	0.106
LSQ12hvj	11:07:38.62	−29:42:40.96	0.071	0.0001	0.072	18.403	0.025	0.043	0.023	0.741	0.088	0.047
LSQ12hxx	03:19:44.23	−27:00:25.68	0.069	0.0001	0.069	18.193	0.025	−0.012	0.023	0.541	0.113	0.016
LSQ12hzj	09:59:12.43	−09:00:08.25	0.029	0.001	0.030	16.481	0.027	−0.112	0.023	−0.406	0.038	0.058
LSQ12hzs	04:01:53.21	−26:39:50.15	0.072	0.0001	0.072	18.810	0.025	0.094	0.023	0.236	0.085	0.023
LSQ13abo	14:59:21.20	−17:09:09.34	0.067	0.0001	0.068	18.835	0.029	0.170	0.025	−0.922	0.135	0.093
LSQ13aiz	13:15:14.81	−17:57:55.65	0.009	0.0001	0.009	13.770	0.018	0.040	0.017	−0.117	0.044	0.082
LSQ13pf	13:48:14.35	−11:38:38.58	0.085	0.005	0.086	19.758	0.027	0.316	0.026	−1.380	0.246	0.057
LSQ13ry	10:32:48.00	04:11:51.75	0.030	0.0001	0.031	16.330	0.021	−0.296	0.021	−1.025	0.032	0.042
LSQ13vy	16:06:55.85	03:00:15.23	0.032	0.005	0.032	17.715	0.018	0.090	0.017	−1.164	0.035	0.069

Note.

^aColumns 4 and 5 give the heliocentric redshift with its error, column 6 is the redshift with respect to the CMB. Columns 7–12 give the values output by SALT2.4: the rest-frame blue-band magnitude and its error, the supernova color and its error, and the stretch factor and its error. The last column gives the Milky Way extinction color parameter.

Download table as:  ASCIITypeset image

The SALT2.4 fits to the light curves are shown as the lines on Figures 3 and 11. A close inspection shows that in a few cases, such as LSQ11ot for example, the fit to the i band is not very good. Since the B and V light curves in these cases are well sampled, removing the i band from the SALT2.4 fit makes a negligible difference, less then a third of a standard deviation in the case of LSQ11ot.

The distance modulus is defined as the difference between the apparent and the absolute magnitude of an object. Since the absolute magnitude is defined at a distance of 10 pc from the object, the distance modulus is given by

$$\begin{eqnarray*}&&\mu =5.0{\mathrm{log}}_{10}({d}_{{\rm{L}}}/10\;\;\mathrm{pc}),\end{eqnarray*}$$

where d_L is the luminosity distance (in units of parsecs) which depends on the cosmological parameters as well as the Hubble constant H₀. In the comparison of the distance moduli of our sample with the Hubble curve, i.e., the expected distance modulus μ_H versus z curve (the curve on Figure 8 below), we use the cosmological parameters from the latest compilation of all of the presently available data (Anderson et al. 2012) which are ${{\rm{\Omega }}}_{{\rm{m}}}=0.270\pm 0.012$, ${{\rm{\Omega }}}_{\mathrm{DE}}=0.740\pm 0.013$, ${{\rm{\Omega }}}_{{\rm{k}}}\;=-0.010\pm 0.005$, and ${w}_{0}=-0.93\pm 0.16$, and ${w}_{{\rm{a}}}\;=-1.39\pm 0.96$, although at the low redshifts of our sample the dependence on these parameters is negligible. For the Hubble constant, we use ${H}_{0}=70.8\pm 1.4$ km⁻¹ s⁻¹ Mpc⁻¹.

The distance modulus μ of each supernova is determined using the equation

$$\begin{eqnarray*}&&\mu ={m}_{B}*-M+\alpha {x}_{1}-\beta {\rm{c}},\end{eqnarray*}$$

where ${m}_{B}*$, x₁, and c are the SALT2.4 output values as defined above. M is the absolute B magnitude of SNe Ia, and α and β are the stretch and the color correction coefficients that are determined in a fit to the Hubble curve (μ versus z) as described below. The determination of M also depends on H₀; thus, the value of M we quote below is appropriate for the value of H₀ we use in calculating the distance moduli. The values of α and β vary somewhat depending on the choice of the width and color of a "typical" SNe Ia with respect to which the stretch factor x₁ and the color c are calculated by SALT2.4.

We evaluate M, α, and β in the following way. Using some starting values of these parameters, we calculate the distance modulus μ_i for each supernova and plot it versus its CMB redshift, i.e., construct the Hubble diagram. We then vary the parameters M, α, and β to minimize the ${X}^{2}$ of the points around the Hubble curve. We define the ${X}^{2}$ as

$$\begin{eqnarray*}&&{\chi }^{2}=\sum _{i}\displaystyle \frac{{({\mu }_{i}-{\mu }_{{\rm{H}}})}^{2}}{{\sigma }_{i}^{2}},\end{eqnarray*}$$

where the sum is over the number of supernova with distance moduli μ_i, and μ_H is the distance modulus expected at the redshift of the supernova as discussed above. The σ_i are the errors on the distance modulus of supernova i and have three ingredients, ${\sigma }_{i}^{2}={\sigma }_{\mathrm{meas}}^{2}+{\sigma }_{\mathrm{sys}}^{2}+{\sigma }_{\mathrm{int}}^{2}$, where σ_meas is the measurement error on the corrected B-band magnitude, σ_sys is a systematic error, and σ_int is the intrinsic spread in the supernova magnitudes. The error due to the uncertainties on the redshifts is negligible, as well as errors due to peculiar velocities for supernovae with redshifts above 0.01. In analyses to measure cosmological parameters (see for example Conley et al. 2011; Betoule et al. 2014, and Rest et al. 2014), supernovae below a redshift of 0.01 are usually eliminated to ensure that the supernovae are in the Hubble flow. None of our sample is below a redshift of 0.01.

The measurement error on each supernova is calculated using the 3 × 3 correlated error matrix C_ij in the variables ${m}_{B}*$, x₁, and c, which is an output from the SALT2.4 fit. The measurement error on the distance modulus is then taken to be

$$\begin{eqnarray*}&&{\sigma }_{\mathrm{meas}}^{2}={{\boldsymbol{V}}}^{{\rm{T}}}{\boldsymbol{C}}V,\end{eqnarray*}$$

with the vector ${{\boldsymbol{V}}}^{{\rm{T}}}=(1,\alpha ,-\beta$). The distribution in σ_meas is shown in Figure 6. The average measurement error is slightly over 4% for this sample.

Zoom In Zoom Out Reset image size

The systematic error σ_sys includes the systematic errors in the flux calibrations (partly due to the systematic errors in the luminosities of the standard stars used in the calibrations) and systematic errors introduced by the galaxy subtractions, which we believe to be the dominant part. Based on the quality of the subtracted images and the consistency of the result of different subtraction algorithms, we estimate the systematic error to be σ_sys = 0.03 mag. The limited analysis discussed here to estimate the values of M, α, and β is not very sensitive to the value of the systematic errors.

To estimate M, α, and β, fits of the distance modulus of each supernova in this sample to the Hubble curve (μ versus z) are carried out in an iterative fashion. In the first fit, all 31 supernovae in the sample are used and the intrinsic spread is fixed at an arbitrary starting value of σ_int = 0.17 (the exact starting value of σ_int is unimportant since it will be varied later to get its best-fit value). The values of M, α, and β are varied to minimize the ${X}^{2}$ of the residuals with respect to the Hubble curve. We calculate the number of standard deviations that the distance modulus of each supernova is from the Hubble curve. The distribution in the absolute value of these standard deviations is shown in Figure 7. The reduced ${X}^{2}$ for this fit (the ${X}^{2}$ divided by the number of degrees of freedom) was 1.7. Three of the supernovae (LSQ11ot, LSQ12gxj, and LSQ13vy) were more than three standard deviations from the curve (see Figure 7). Statistically, we expect 1 in a 1000 entries to be beyond 3 standard deviations, and so we consider these 3 supernovae to be outliers. When these 3 outliers are removed from the fit, the value of the intrinsic spread σ_int is varied to make the value of the reduced ${X}^{2}$ equal to 1. In this final fit, the errors due to the uncertainties on M, α, and β (listed in Table 4) are added in quadrature. This procedure yields the best value of the intrinsic spread σ_int = 0.14.

Zoom In Zoom Out Reset image size

All three of the outliers, LSQ11ot, LSQ12gxj, and LSQ13vy, are sub-luminous by about 0.6 mag with respect to the rest of the sample. They have significant Na i D absorption lines in their spectra which can be interpreted as a sign of absorption along their line of sight (see for example Poznanski et al. 2012), although Na i D absorbtion may not be a reliable indicator of absorbtion (Phillips et al. 2013). LSQ11ot and LSQ12gxj are the most reddened in the sample, suggesting an imperfection in the color–magnitude relation. In any case, we are not throwing these three supernovae away. We are merely not using them in the fit to estimate M, α, β, and the intrinsic spread for this sample. They remain in Table 3 and their light curves are in the appendix to be included or not as seen fit in any future analysis for the cosmological parameters.

The results of the ${X}^{2}$ fit to the Hubble curve described in Section 4.5 are given in Table 4. With these parameters, the distance moduli of the supernovae are calculated and plotted versus their CMB redshift, producing the Hubble diagram for this sample shown in Figure 8 below. The rms spread of the points around the Hubble curve is 15.6%. The error bars in this figure include the measurement errors and the intrinsic spread of the supernovae added in quadrature, combined with the errors due to the uncertainty of the parameters M, α, and β.

Zoom In Zoom Out Reset image size

Examining our resulting distributions in the SALT2.4 output stretch parameter x₁ and the color parameter c, we find that the ranges in x₁ and c for this sample are quite consistent with the ranges obtained in other larger samples of supernovae. In the x₁ distribution in Figure 4, we superimpose the x₁ distributions from Figure 11(b) of the most recent analysis of the cosmological parameters with the PanSTARRS supernova sample (Rest et al. 2014) where they collect all of the available low-z (0.01 < z < 1.0) sample, 197 supernovae, and also show the distribution for the 113 higher-redshift supernovae from the PanSTARRS sample. The comparison curves are normalized to the same area as the LSQ-CSP II sample. The distribution in the color parameter c from the Rest et al. (2014) paper for the low-z sample and the distribution for the PanSTARRS sample are shown superimposed on Figure 5. In addition, we show the c distributions from the SNLS sample of 242 higher-redshift supernovae and the SDSS sample of 93 supernovae from Figure 1 of Conley et al. (2011) superimposed on Figure 5 above. The comparison curves are normalized to the area of the LSQ-CSP II sample.

A Kolmogorov–Smirnov test was carried out comparing the present sample and the average of the other samples. The confidence that the samples are consistent was 95% for the x₁ distribution and 58% for the c distribution, both acceptable confidence levels. The discrepancy in the c distribution comes from the reddened tail in the last two bins of Figure 5. We note that two of the three outliers to the Hubble diagram of Figure 8, discussed in Section 4.5 above, are in this tail. Removing these two outliers from the distribution, the consistency between the c distributions from the Kolmogorov–Smirnov test improves to 90%.

The fitted values of M, α, and β listed in Table 4 are compared to the measurement of these parameters for other samples of supernovae in Table 5 below.

There is some spread in the M, α, and β parameters not only with the data sets, but also with the particular analysis of the data. For example there is some variation in these parameters from the SNLS data between the papers of Conley et al. (2011), Sullivan et al. (2011), and Betoule et al. (2014). Sullivan et al. (2010) also point out that there is a variation of M with host galaxy properties such as stellar mass. From Table 5, we see that the values obtained in this paper are within the range of the M and β parameters from the other data samples, but is low compared to the others in β, although within one standard deviation of the Union2 data set. This parameter characterizes the correlation between the magnitude of supernovae and their color. This correlation is not well understood at this time. It may be due to extinction of the supernova light between the supernova and Earth, or due to intrinsic properties of the supernova itself. The color of the supernova itself may or may not be correlated to its magnitude, as discussed in Mariner et al. (2011) and Scolnic et al. (2014).

Sullivan et al. (2010) used 476 supernovae to find a dependence of M on the galaxy mass, and Scolnic et al. (2014) used 518 supernovae to see an effect of the assumptions about the source of the supernova color variation on β. The limited statistics of the present sample of 31 supernovae are not sufficient to contribute to the study of these effects.

The residuals of our data sample in the Hubble diagram of Figure 8 have an rms of 0.156 mag, and the intrinsic spread of the supernovae is σ_int = 0.14. These numbers are compared to the results from other data samples in Table 6.

The Hubble diagram residual rms and the intrinsic spread σ_int are within the range of the other samples in Table 6. Another interesting comparison can be made with the values for the rms and σ_int that are given in Table 7 of Amanullah et al. (2010) for the 16 different supernova data sets, 557 supernovae spanning redshifts from 0.01 to 1.1, that make up the Union2 data set. These numbers are plotted in Figure 9. The values for these two quantities from the present paper are the crosshatched entries on this figure. They are well within the distribution of these parameters.

Zoom In Zoom Out Reset image size

The plan of this collaboration is to continue this effort, possibly with other observatories joining, to reach a sample of 300 low-redshift SNe Ia in the next few years. The combined total of the presently published nearby supernovae of sufficient quality to use in a cosmological fit is less then 200 (Amanullah et al. 2010; Betoule et al. 2014; Rest et al. 2014). The other ongoing nearby supernova searches have not published any samples yet. Their expectations are as follow: SNfactory (Aldering et al. 2002) expect 300, PTF (Maguire et al. 2014) expect 300, and Sky Mapper (Keller et al. 2007) expect 200 nearby supernovae. Such a combined sample, even with some expected overlap between the samples, should approach roughly 1000 carefully observed and analyzed supernovae, with some infrared observations by the Carnegie Supernova Project (M. M. Philips et al. 2015, in preparation). Such a sample will be valuable as a low-redshift anchor for high-redshift supernova surveys such as SNLS, ESSENCE, DES, LSST, and WFIRST. To quantify the effect of such a large sample of nearby supernovae, the WFIRST Science Definition Team (Spergel et al. 2015) has estimated the increase in the figure of Merit (defined as the reciprocal of the area of the w_a versus w₀ error ellipse) as a function of the size of the nearby sample. The result, included here as Figure 10, shows an increase of a factor of three in the FoM by virtue of a nearby sample of 1000 supernovae.

Zoom In Zoom Out Reset image size

In addition to anchoring the Hubble diagram, which will eventually be systematics limited, a large sample of nearby supernova will be important to sharpen our understanding of the nature of supernovae as cosmological distance indicators (see for example Maguire et al. 2012; Silverman et al. 2013). For example, there are indications (Fakhouri et al. 2012) that dividing both the low- and the high-redshift samples into sub-classes of Type Ia SNe (comparing twins to twins, so to speak) will reduce systematic uncertainties. A large sample of nearby objects will be needed to obtain sufficient statistics in the individual sub-classes.