1

SDSS-II Supernova Survey

Josh FriemanLeopoldina Dark Energy Conference

October 8, 2008

See also: poster by Hubert Lampeitl, talk by Bob Nichol

2

SN Models and Observations•SN cosmology based on a purely empirical approach (Phillips)

•SN observations over the last decade have strengthened evidence for cosmic acceleration, but dark energy constraints now dominated by systematic errors

•SNe will be one of 3 dark energy probes pursued by JDEM

•Reaching JDEM level of precision for SNe will require improved control of systematics

•Improved SN modeling, better empirical approaches to estimating SN distances, and better data are all important weapons in the arsenal to reduce systematics

•Current empirical distance estimators are limited by the paucity of high-quality input/training data. The situation is improving (CfA, CSP, KAIT, SNF, SDSS), but we need better, homogeneous data at low/intermediate redshifts and a systematic approach to ingesting them to build better empirical estimators. Will current ground-based SN surveys deliver what we need for JDEM?

4

Cosmic AccelerationDiscovery from High-redshiftSNe Ia

SNe at z~0.5 are 25% fainter than in an open Universe with same value of m

= 0.7 = 0.m = 1.Technological Redshift

Desert:Possible photometric offsets between low- and high-redshift data

Desert still there 10 years later

5

SDSS II Supernova Survey Goals• Obtain few hundred high-quality SNe Ia light curves in the

`redshift desert’ z~0.05-0.4 for continuous Hubble diagram

• Spectroscopic follow-up for redshifts, SN typing, and to study diversity of SN features

• Probe Dark Energy and systematics in redshift range complementary to other surveys

• Well-observed, homogeneous sample to anchor Hubble diagram & train distance estimators

• Large survey volume: rare & peculiar SNe, probe outliers of population to test SN models

Frieman, et al (2008); Sako, et al (2008)

Spectroscopic follow-up telescopes

R. Miquel, M. Molla

CfA team P. Challis, G. Narayan, R. Kirshner

Search Template Difference

g

r

i

Searching For Supernovae• 2005

– 118,693 objects scanned

– 10,937 unique candidates

– 130 confirmed Ia

• 2006– 14,430 scanned

– 3,694 candidates

– 193 confirmed Ia

• 2007 13,613 scanned 3,962 candidates

– 175 confirmed Ia

•Positional match to remove movers•Insert fake SNe to monitor efficiency

B.

Dild

ay

Redshift Distribution for SNe Ia

and counting

Well-sampled, multi-band light curves, including measurements before peak light

SDSSSNLight-curves

Holtzman et al (2008)

Spectroscopic Target Selection2 Epochs

SN Ia Fit

SN Ibc Fit

SN II Fit

Sako etal 2008

Spectroscopic Target Selection2 Epochs

SN Ia Fit

SN Ibc Fit

SN II Fit

31 Epochs

SN Ia Fit

SN Ibc Fit

SN II Fit

Fit with template library

Classification>90%accurate after 2-3 epochs

Redshifts 5-10% accurate

Sako etal 2008

SN and Host Spectroscopy

MDM 2.4m

NOT 2.6m

APO 3.5m

NTT 3.6m

KPNO 4m

WHT 4.2m

Subaru 8.2m

HET 9.2m

Keck 10m

Magellan 6m

TNG 3.5m

SALT 10m

2005+2006

SDSS SN Ia Spectra

~1000 spectra taken over 3 seasonsZheng et al (2008)

Fitting SN light curves I: MLCS2k2• Multicolor Light Curve Shape (Riess et al '98; Jha et al '07)

• Model SN light curves as a single parameter family, trained on low-z UBVRI data from the literature

• Assumes SN color variations are due to dust extinction, subject to prior

fit parameters

time-dependent model “vectors”

Time of maximum distance modulus dust law extinction stretch/decline rate

P(Av)

MLCS2k2 model templates

∆ = -0.3: bright, broad

∆ = +1.2: faint, narrow

Jha et al, 2007

18

Fitting SN Light curves II: SALT2

• Fit each light curve using rest-frame spectral surfaces*:

• Transform to observer frame:

• Light curves fit individually, but distances only estimated globally:

*Not trained just on low-redshift data; distances are cosmology-dependent, flat priors on model parameters

Global fit parameters, determined along with cosmological parameters

color term

Guy et al

light-curve shape

19

Light Curve Fitting with MLCS2k2 and SALT2

Monte Carlo Simulations match data distributions

Use actual observing conditions (local sky, zero-points, PSF, etc)

Model Spectroscopic & Photometric Efficiency

Redshift distribution for all SNe passing photometric selection cuts (spectroscopically complete sample)

Data

Need to model biases due to what’s missing

Difficult to model spectroscopic selection

Extract AV Distribution from SDSS

(no prior)

Extract RV distribution from SDSS SN data

• MLCS previously used Milky Way avg RV=3.1

• Lower RV more consistent with SALT2 color law

• Not conventional dust

€

RV =AV

E(B −V )≈ 2

Preliminary Cosmology Results

Kessler, Becker, et al. 2008

w

open

25

Issues with rest-frame U band

•Data vs. SALT2 Model Residuals•Similar Low-z vs. High-z discrepancy seen in MLCS•MLCS trained only on Low-z, SALT2 model dominated by SNLS•Similar differences seen in rest-frame UV spectra (Foley et al)

epoch

26

SN Ia vs. Host Galaxy Properties: I

Smith et al

Bright SN Luminosity/Decline Rate Faint

27

SN Ia vs. Host Galaxy Properties: II

Smith et al

Color/reddening

Is reddening local to the SN environment?

28

SN Ia vs. Host Galaxy Properties: III

Smith et al

PreliminaryTwo SN IaPopulations?

Implications for SNcosmology:host-galaxypopulationevolution

Future: Improved SN Ia Distances

TrainFitters

Fit Cosmology