EDE 2026 - GC 3/4

Clustering des galaxies observations 3/4 - Aspects expérimentaux

Ecole Rodolphe Cledassou 2026

Corentin Ravoux

LPCA, CNRS/IN2P3, Clermont-Ferrand

ACCEL² simulation

Ecole Rodolphe Cledassou 2026 - galaxy clustering: observation 3/4

Outline

1. Light path from source to detector

2. Galaxy survey properties

3. An overview of galaxy surveys

Ecole Rodolphe Cledassou 2026 - galaxy clustering: observation 3/4

Outline

1. Light path from source to detector

2. Galaxy survey properties

3. An overview of galaxy surveys

1. Light path from source to detector

Ecole Rodolphe Cledassou 2026 - galaxy clustering: observation 3/4

Life of a photon

Source

Cosmic web

Telescope

Atmosphere

Milky Way

Ecole Rodolphe Cledassou 2026 - galaxy clustering: observation 3/4

Life of a photon

Source

Ecole Rodolphe Cledassou 2026 - galaxy clustering: observation 3/4

Redshift

Source

Emitted spectra are redshifted due to Universe expansion and galaxy peculiar velocity

(1+z)=\left(1+z_{\exp }\right)\left(1+z_{\text {vpec }}\right)

Probe universe expansion

Probe cosmic web

Ecole Rodolphe Cledassou 2026 - galaxy clustering: observation 3/4

Life of a photon

Cosmic web

Ecole Rodolphe Cledassou 2026 - galaxy clustering: observation 3/4

Traveling the cosmic web

Galaxy/Quasar light travels over large distances through:
- Intergalactic medium (IGM)
- Other galaxies/clouds
All these elements leave an imprint on the spectrum (spectro-side) and the shape of the object (photo-side)

Cosmic web

Ecole Rodolphe Cledassou 2026 - galaxy clustering: observation 3/4

Life of a photon

Milky Way

Ecole Rodolphe Cledassou 2026 - galaxy clustering: observation 3/4

Absorption: dust maps

Milky way

Dust and gas affect the
transfer of radiation through a galaxy (scattering & absorption)
Impact generally expressed in term of color difference in B (blue) and V (visual) bands, called reddening:

Light that cross dust appears redder than emitted light

E(B-V) = (B-V)_{\text{obs}} - (B-V)_{0}

Johnson/Cousins UBVRI system

Ecole Rodolphe Cledassou 2026 - galaxy clustering: observation 3/4

Absorption: dust extinction law

Milky way

Total extinction in the visual band (V)

Extinction for a specific wavelength:

a and b functions have complex profiles, depending on the galaxy
Affects light both in the MW and in the host galaxy

A_V = R_V \times E(B-V)

A(\lambda) = A_V \times \left[ a\left(\frac{1}{\lambda}\right) + \frac{b\left(\frac{1}{\lambda}\right)}{R_V} \right]

= ratio of total to selective extinction

R_V

~ 3.1 for Milky way but can vary depending on dust property

Ecole Rodolphe Cledassou 2026 - galaxy clustering: observation 3/4

Absorption: Milky Way dust

Milky Way

E(B-V) dust extinction map in the Milky Way
Major contaminant of CMB surveys
For galaxy surveys: High values of E(B-V) can bias the clustering of pre-targeted galaxies

Additional absorption lines of specific elements: CaII H-K lines, Na doublet line..

Improved SFD map

Ecole Rodolphe Cledassou 2026 - galaxy clustering: observation 3/4

Emission: background

Milky Way

Background for JWST

Yellow = out of FoV
blue = galactic + zodiacal
red = self emission

Galactic emission:
- Synchrotron: e⁻ in galactic magnetic field
- Thermal free-free: e⁻ scatter off ions (bremsstrahlung)
- Dust, molecular emissions
Zodiacal emission: Infrared radiation by heated particles in our Solar System

Ecole Rodolphe Cledassou 2026 - galaxy clustering: observation 3/4

Life of a photon

Atmosphere

Ecole Rodolphe Cledassou 2026 - galaxy clustering: observation 3/4

Atmosphere

For ground based surveys, need to account for atmosphere effects

Ecole Rodolphe Cledassou 2026 - galaxy clustering: observation 3/4

Some astronomy definitions

Transparency/Opacity: transmission of the atmosphere on the line of sight.
Sky level: Sky brightness from natural and artificial light sources (mag.arcsec⁻²)
Airmass: Amount of air in the line-of-sight (normalized to zenith).
Seeing: Blurring caused by atmosphere, quantified by FWHM of a star image (in arcsec)
Often normalized to a nominal atmosphere on the considered observation site

Atmosphere

Ecole Rodolphe Cledassou 2026 - galaxy clustering: observation 3/4

Atmospheric absorption

Atmosphere

Atmospheric absorption determines the wavelength range of a survey

Ecole Rodolphe Cledassou 2026 - galaxy clustering: observation 3/4

Atmospheric absorption

Atmosphere

Absorptions caused by different molecules in the atmosphere
Scattering caused by molecules (Rayleigh) and aerosols (particulate matter, Mie)
Some absorptions vary during the night (water, aerosols)

Ecole Rodolphe Cledassou 2026 - galaxy clustering: observation 3/4

Atmospheric emissions

Atmosphere

Airglow: natural upper atmosphere emission
Light pollution
Moon light especially in the blue (u band)
Molecular emissions

Sky emission spectrum taken by DESI

Ecole Rodolphe Cledassou 2026 - galaxy clustering: observation 3/4

Atmosphere continuum

Atmosphere

In the optical and NIR, emission dominated by airglow (OH in particular)
Thermal radiation from atmosphere and telescope dominates above 2.2 μm

Noll et al. 2012

Sky model in Cerro Paranal

Ecole Rodolphe Cledassou 2026 - galaxy clustering: observation 3/4

Atmosphere sky lines

Atmosphere

Sky emission highly dominated by OH radicals (continuum + lines)

Ecole Rodolphe Cledassou 2026 - galaxy clustering: observation 3/4

Life of a photon

Telescope

Ecole Rodolphe Cledassou 2026 - galaxy clustering: observation 3/4

Instrument

Telescope design (refracting, reflecting, both)
Lenses for focal plane focusing and optical corrections
Telescope responses (sometimes including atmosphere):
- Throughput: Total transmission of the telescope as a function of wavelength
- PSF (Point Spread Function): image of a point-like source (with or without seeing)

Telescope

Jee et al. 2011

Detector: CCD, CMOS, IR... see J-Ch. Cuillandre Y2-obs

Ecole Rodolphe Cledassou 2026 - galaxy clustering: observation 3/4

Outline

1. Light path from source to detector

2. Galaxy survey properties

3. An overview of galaxy surveys

2. Galaxy survey properties

Ecole Rodolphe Cledassou 2026 - galaxy clustering: observation 3/4

Surveys

Generally, surveys tend to be either:
- Static or dynamic (transient)
- Wide & shallow or small & deep
- Photometric or spectroscopic
Very recent surveys achieve several functions at once
A survey covers a fixed wavelength range (from gamma to radio)
In galaxy surveys for this course: optical/NIR, wide, static surveys

Ecole Rodolphe Cledassou 2026 - galaxy clustering: observation 3/4

Properties: field of view

Field of view (Ω): area of the sky visible through the telescope camera
Depends on telescope optics (design, focal length, lenses, correctors) and camera size
Collecting area surface (A): larger mirror = higher photon flux
Etendue (E=AΩ): characterizes the quantity of accessible information on the sky

Ecole Rodolphe Cledassou 2026 - galaxy clustering: observation 3/4

Properties: volume of the survey

Footprint: observed size on the sky at the end of the survey
Expressed in healpix format (hierarchical equal area isolatitude pixelation)
Density: Number of objects observed per square degree
Redshift range dz
Volume ∝ Footprint × comoving radial range
Python package: healpy

N_{\mathrm{pix}} = 12 N_{\mathrm{side}}^2

Ecole Rodolphe Cledassou 2026 - galaxy clustering: observation 3/4

Properties: depth

Magnitude: logarithmic scale for object fluxes for one given filter
Apparent magnitude (m): magnitude on the detector (including extinction)
Absolute magnitude (M): magnitude at 10 parsec
Zero point calibration:
- Vega system: Vega star magnitude null for all wavelength
- AB system: Reference to a constant flux object (3631 janskys)

M = m + 5 - 5 \log _{10}(d)

m_1-m_2=-2.5 \log _{10}\left(\frac{F_1}{F_2}\right)

m_{\mathrm{Vega}}(\mathrm{Obj})=-2.5 \log \left[\frac{\int f_\lambda(\mathrm{Obj}) W_\lambda d \lambda}{\int f_\lambda(\mathrm { Vega }) W_\lambda d \lambda}\right]

m_{\mathrm{AB}}(\mathrm{Obj})= -2.5\log\left[\frac{\int f_\lambda({\rm Obj})W_\nu d\lambda}{\int f_\lambda^{\rm AB}W_\nu d\lambda}\right]

= filter response/window

W_\lambda

Survey depth: magnitude (redshift) of the faintest (farthest) object observed
Signal-to-noise ratio (SNR): characterize the quality of one observation

Ecole Rodolphe Cledassou 2026 - galaxy clustering: observation 3/4

Calibration images

Zero/Bias

Dark

Flat

Arc

All surveys need calibration images
Zero/Bias: response to a minimum exposure time image
- Measure the CCD intrinsic readout noise
Dark: Image with the same exposure time as science image without any light
- Use to characterize noise sources

Ecole Rodolphe Cledassou 2026 - galaxy clustering: observation 3/4

Calibration images

Flat: Focal plane illuminated with a known smooth spectrum lamp
- Use to correct relative CCD responses under illumination
Arc: Image produced by known emission lines
- Use for wavelength calibration (only for spectra)
Master: A stack/coadd of several calibration images

Zero/Bias

Dark

Flat

Arc

Ecole Rodolphe Cledassou 2026 - galaxy clustering: observation 3/4

Photometric vs Spectroscopic surveys

Source: DESI communication figures

Measure the spectra of objects (selected or not)
Examples: SDSS, 6dFGS, DESI, 4MOST, Euclid NISP (spectro)

Measure the object flux through a filter
Examples: DES, ZTF, CFHT, LSST, Euclid VIS, Euclid NISP (photo)

Same detection method (CCD, IR)

Ecole Rodolphe Cledassou 2026 - galaxy clustering: observation 3/4

Photometric surveys: filters, filters and filters

Filter: Precise transmission shape and width vs wavelength
"Standard" filters: Johnson/Cousins (UBVRI), Bessell, Gunn, IR (JHK...)
Full/Synthetic filter: Filter x throughput x quantum efficiency x standard atmosphere
Images taken with different filters (fast changer for recent surveys)

Ecole Rodolphe Cledassou 2026 - galaxy clustering: observation 3/4

Photometric surveys: filters, filters and filters

A lot of different systems of filters are used...
Practical website for filter profiles:
- https://astroquery.readthedocs.io/en/latest/svo_fps/svo_fps.html
- https://speclite.readthedocs.io/en/latest/filters.html

Ecole Rodolphe Cledassou 2026 - galaxy clustering: observation 3/4

Photometric surveys: photometric redshift

Only a partial part of the spectra is measured
Classification and redshift methods:
- Template fitting
- Machine learning
- Additional environment information
Outliers: redshift & object type confusion
Recent precision:
Need to be calibrated with spectroscopic redshift

\sigma(z) \sim 10^{-2} (1+z)

\sigma(z) \sim 10^{-4} / 10^{-5}

Lee et al. 2024

Ecole Rodolphe Cledassou 2026 - galaxy clustering: observation 3/4

Photometric surveys: image calibration

Photometry: providing a precise estimate of object magnitudes/brightness
After the CCD, e⁻ are converted into digital counts, or Analog-Digital Units (ADUs) by ADC (Analog To Digital Converter)
Photon transfer curve: ADU variance of uniform exposures vs flux
Gain: N input e⁻ / N output ADU
Measuring stability of gain and saturation is essential !

Photons on CCD follow Poisson law

(full well capacity)

Ecole Rodolphe Cledassou 2026 - galaxy clustering: observation 3/4

Spectroscopy

Diffraction grating: periodic structure
Interference condition: path difference between AB and A'B‘ is equal to an integer multiple of λ

m \lambda= a (n_1 \sin \alpha+n_2 \sin \beta)

Wavelength dependent dispersion
Zeroth order corresponds to the specular reflection off the grating

n_2 \sin(\beta) = - n_1 \sin(\alpha)

= distance between adjacent grooves

\beta(\lambda)

Ecole Rodolphe Cledassou 2026 - galaxy clustering: observation 3/4

Spectroscopy

Pattern of the grating have an impact on the diffraction scheme
Adjusting the blaze angle can maximize efficiency in one order and minimize / cancel it in the other orders
Littrow configuration:
Achieve maximum grating efficiency in a given diffraction order m

\beta \simeq \alpha = \theta_{\mathrm{B}}

Blazed grating

m \lambda= 2 a \sin \theta_{\mathrm{B}}

Ecole Rodolphe Cledassou 2026 - galaxy clustering: observation 3/4

Spectroscopy: dispersion

Grism: Transmission grating attached to a prism
Quasi-Littrow configuration - no variable distortion
Inefficient for number grooves > 600/mm due to groove shadowing and other effects

Ecole Rodolphe Cledassou 2026 - galaxy clustering: observation 3/4

Spectroscopy: dispersion

Volume Phase Holographic gratings work by a mechanism similar to Bragg diffraction (crystal diffraction)
- Periodic refractive index modulations in a volume of gelatin
- Interference pattern created by two coherent laser beams inside a photosensitive medium
Transmission gratings with very low fringe spacings can be made

Glass

Ecole Rodolphe Cledassou 2026 - galaxy clustering: observation 3/4

Spectroscopy: spectral resolution

Ability to distinguish spectral features separated by Δλ
Resolving power:
- Low resolution
- High-resolution
Choice depends on the spectral features
- Broad features (e.g. because of velocity dispersion) or narrow (object properties)
Higher resolution need larger exposure time for the same SNR

R = \frac{\lambda}{\Delta \lambda}

R \sim 100

R \sim 50000

R \sim 5000

R \sim 50000

Ecole Rodolphe Cledassou 2026 - galaxy clustering: observation 3/4

Spectroscopy: type of spectrograph

Slitless spectrograph (no mask, sky background)

Multi-slit spectrograph (focal plane mask)

Multi-fiber spectrograph (optical fibers)

Integral field spectrograph

See "Spectroscopie IR" Y2-OBS Y. Copin

Ecole Rodolphe Cledassou 2026 - galaxy clustering: observation 3/4

Spectroscopy: target selection

For multi-object spectrographs: select objects to measure from photometric surveys
Magnitude cuts: select brighter objects
Color (magnitude difference) cuts:
- Select a given population of object (LRG, ELG, QSO, LBG)
- Homogeneous sample for clustering measurements
- Reject stars and unwanted objects

Chaussidon et al. 2022

Galaxies

Stars

Zhou et al. 2022

Ecole Rodolphe Cledassou 2026 - galaxy clustering: observation 3/4

Outline

1. Light path from source to detector

2. Galaxy survey properties

3. An overview of galaxy surveys

Ecole Rodolphe Cledassou 2026 - galaxy clustering: observation 3/4

Galaxy observations

964: Abd al-Rahman al-Soufi (Persian astronomer) observes first galaxies (Andromeda, Large Magellanic Cloud)
1750/1755: Thomas Wright (English astronomer) & Emmanuel Kant postulate that nebulae are other galaxies like the Milky Way
- "Une nébuleuse n’est pas un unique et énorme soleil, mais un système de nombreux soleils, rassemblés en raison de leur distance dans un espace si étroit, que leur lumière, qui serait imperceptible pour chacun d’eux isolément, parvient, grâce à leur innombrable quantité, à produire une blancheur pâle et uniforme.” – Emmanuel Kant, Histoire générale de la nature et théorie du ciel, 1755

1910-20's:
- Leavitt & Hubble observations confirmed the existence of other galaxies
- Hubble & Lemaitre discovers expansion

Ecole Rodolphe Cledassou 2026 - galaxy clustering: observation 3/4

Photometric surveys - "Stage I"

POSS I (1949 - 1958)
UKST (1974 - 1999)
POSS II (1985 - 2000)
First digital surveys:
- Digitized Sky Survey (1994) scanning of POSS I/II and UKST
- 2MASS (1997 - 2001): 3-band NIR photometric survey (also considered "stage II")

photographic plate of POSS I

copy-paste from the mighty Arnaud

Ecole Rodolphe Cledassou 2026 - galaxy clustering: observation 3/4

Photometric surveys - "Stage II/III"

Deep surveys:
- Hubble space telescope (1990-)
- VLT (1998-)
Wide surveys:
- DES (2013-2019)
- HSC (2014-)
- Pan-STARRS (2008-)
- CFHT (1979-)
- VISTA VHS (2009-2015)

COSMOS

VLT

UNIONS

DES Imager

Ecole Rodolphe Cledassou 2026 - galaxy clustering: observation 3/4

Spectroscopic surveys - "Stage I"

First spectrographic surveys:
- CfA I (1977 - 1982), ~2k redshifts, structures formed by galaxies
- CfA II (1985 - 1995), 18 000 redshifts, Great Wall discovery

copy-paste from the mighty Arnaud

Ecole Rodolphe Cledassou 2026 - galaxy clustering: observation 3/4

Spectroscopic surveys - "Stage II"

SDSS-I SDSS-II (1998-2008)
2dFGS 6dFGS (1997-2009)
WiggleZ (2006-2011)
Goal: BAO detection

6dFGS map

SDSS

WiggleZ BAO

Blake et al. 2011

Ecole Rodolphe Cledassou 2026 - galaxy clustering: observation 3/4

Spectroscopic surveys - "Stage III"

SDSS BOSS/eBOSS (2008-2020)
HETDEX (2017-2024)
Goal: improving BAO

IFU

Ecole Rodolphe Cledassou 2026 - galaxy clustering: observation 3/4

Moore's law for spectroscopic galaxy survey

III

copy-paste from the mighty Arnaud

Ecole Rodolphe Cledassou 2026 - galaxy clustering: observation 3/4

Present experiments: "Stage IV"

Ground photo:
- LSST (2026-)
Ground spectro:
- DESI (2021-)
- 4MOST (2025-)
- MUST (2028-)
- PFS Subaru (2025-)
Space (spectro/photo):
- Euclid (2023-)
- JWST (2021-)
- Roman (2026-)

Ecole Rodolphe Cledassou 2026 - galaxy clustering: observation 3/4

The Dark Energy Spectroscopic Instrument (DESI)

Ecole Rodolphe Cledassou 2026 - galaxy clustering: observation 3/4

4 m telescope at Kitt Peak Observatory
2021 - 2028
5 years of data secured

DESI collaboration 2022

DESI instrument

Ecole Rodolphe Cledassou 2026 - galaxy clustering: observation 3/4

DESI survey forecast

BGS: Bright Galaxy Survey
LRG: Luminous Red Galaxy
ELG: Emission Line Galaxy
QSO: Quasar
- z > 2.1: with a Lyman-α forest
- z < 2.1: as tracers

63 million redshifts in 7 years

over 17,000 deg²

Primary science: BAO and RSD measurements

Ecole Rodolphe Cledassou 2026 - galaxy clustering: observation 3/4

DESI BAO

DESI collaboration 2024

Ecole Rodolphe Cledassou 2026 - galaxy clustering: observation 3/4

DESI focal plane

5 μm positioning in few sec. for 5000 targets
All fibers connected to the spectrographs

Silber et al. 2022

Ecole Rodolphe Cledassou 2026 - galaxy clustering: observation 3/4

DESI spectrographs

10 Spectrograph modules with cryostats containing 3 CCD

DESI collaboration 2022

Ecole Rodolphe Cledassou 2026 - galaxy clustering: observation 3/4

DESI science image

Raw image over one DESI CCD of one spectrograph

Identify some features

Ecole Rodolphe Cledassou 2026 - galaxy clustering: observation 3/4

DESI science image

Fibers side by side

Wavelength

Cosmics

Atmosphere emission lines

Extracted spectrum

Ecole Rodolphe Cledassou 2026 - galaxy clustering: observation 3/4

DESI calibration images

Zero/Bias

Dark

Flat

Arc

DESI calibration images

Minimal exposure time with dome closed

Science exposure time with dome closed

Telescope pointed to a uniformly illuminated screen in the dome

Screen in the dome illuminated with calibration emission lamps

Ecole Rodolphe Cledassou 2026 - galaxy clustering: observation 3/4

DESI spectral extraction

Raw image calibrated with bias/flat/dark/arc
Cosmics and bad pixels detected and flagged by software
Spectra extracted with the "spectroperfectionism" algorithm:

\vec{p}=\mathbf{A} \vec{f}+\vec{n}

Camera signal

Flux

Noise

Spectral noise evaluated with calibration images
Spectrum level normalized using sky fibers (fixed, or broken positioners) and standard stars

A = matrix measured with arc images, containing full 2D PSF modeling

Ecole Rodolphe Cledassou 2026 - galaxy clustering: observation 3/4

DESI spectral extraction

Object type and

redshift obtained by:

Visual inspection
Template fitting
Machine learning

\left(\mathrm{RA},\mathrm{DEC},z\right)

\left(\mathrm{RA},\mathrm{DEC},z,f_{\lambda}\right)

for galaxies

for Lyman-alpha forest

Correlation function and power spectrum measurements

Ecole Rodolphe Cledassou 2026 - galaxy clustering: observation 3/4

Legacy Survey of Space and Time (LSST)

Ecole Rodolphe Cledassou 2026 - galaxy clustering: observation 3/4

LSST acronyms

Vera Rubin Observatory (VRO)

Legacy Survey of Space and Time (LSST)

Dark Energy Science Collaboration (DESC)

+ 7 (Transients, SL, AGN...)

Observatory/Telescope

Survey

Science collaborations

Ecole Rodolphe Cledassou 2026 - galaxy clustering: observation 3/4

Rubin observatory

30 m diameter dome

Telescope mount assembly

AuxTel: 1.2m atmospheric telescope

Cerro Pachon

Service and maintenance facility

Control room

Ecole Rodolphe Cledassou 2026 - galaxy clustering: observation 3/4

Rubin observatory: mirrors

Modified three-mirror Paul-Baker design: M3 and M1 polished from a single substrate
8.4 m primary mirror
9.6 square degrees field of view

Ecole Rodolphe Cledassou 2026 - galaxy clustering: observation 3/4

Rubin observatory: mechanical structure

Telescope Mount Assembly (~15 m)
Can move 3.5° in 4 seconds
Mirrors mounted on hexapods and correctors: counteract the changing distortions produced by changes in the gravitational orientation

Ecole Rodolphe Cledassou 2026 - galaxy clustering: observation 3/4

Rubin observatory: camera

Largest digital camera in the world (3.2 Gpixels)
2s readout time
Integrated cryostat modules (~-120°C)
3060 kg

Ecole Rodolphe Cledassou 2026 - galaxy clustering: observation 3/4

Rubin observatory: camera

189 4k x 4k science CCD
Corner sensor for guiding and wave-front analysis
CCD packed by raft (3x3) with dedicated electronic treatment system

Ecole Rodolphe Cledassou 2026 - galaxy clustering: observation 3/4

Rubin observatory

Integrated filter wheel and filter changer (French contribution)
6-band ugrizy survey

Ecole Rodolphe Cledassou 2026 - galaxy clustering: observation 3/4

LSST survey

10-year photometric survey of half of the sky (~ 20 000 deg²)
Started in june 2026
Complete coverage every 4 nights during 10 years (static and transient survey)
15 TB data per night, 100 PB in total, 10 million alerts per night in average
Deliverables:
- 20 billion galaxies with photometric redshifts up to z=3
- 17 billion resolved stars
- ~ 1 million type Ia supernovae (with ~ 300,000 cosmological SNe Ia)

Ecole Rodolphe Cledassou 2026 - galaxy clustering: observation 3/4

LSST survey: observation strategy

Strategy elaborated for 10 years to satisfy 8 science collaborations and a lot of science cases
Full survey simulation night by night
Creation of precise metrics
Example: full SNIa simulation and impact of rolling cadence
5 Deep Drilling fields for photo-z & WL calibration, high redshift SNIa ...