1. Lecture #2 Olivier Le Fèvre - LAM Cosmology Summer School 2014

2. Lecture plan 1. Designing a deep survey 2. Instruments for deep surveys 3. Observational methods 4. Data processing 5. Databases and information systems 6. Comparing surveys

3. Designing a survey  Science goals & strategy  Survey parameter space  Existing or new instrumentation ?  Examples  SDSS  VVDS  VIPERS  VUDS

4. How are galaxy surveys designed ? The ‘Wedding cake’ approach 4 Deep / small field Medium / large field Shallow / all-sky

5. Some Principles  Surveys need to be unbiased  Volume, luminosity/mass, type, environment…  Proper photometric catalogs  Statistically robust  Complete census  Selection function control  Apriori hypotheses  Large deep imaging surveys  Large samples  Multi-wavelength 5 2 types of surveys: photometric and spectroscopic

6. Science goals: the starting point  What are the science questions addressed by the survey ?  What are the measurements to be performed ?  What is the desired accuracy ? Star formation rate Merger rate Mass of galaxies in DM Halos Build-up of red sequence Morphology-density Growth rate … As a function of z…

7. Survey parameter space Survey Volume Depth Spectral resolution Nobj z- range Also: spatial resolution

8. Survey volume Single pointing footprint: Megacam @CFHT Whole sky tiling: Euclid

9. An instrument system is more efficient the larger the Etendue Etendue: the area of the entrance pupil times the solid angle the source E=A  Etendue (some instrumentation stuff…)

10. A: a key element in instrument systems  A = telescope collecting area   = telescope+instrument field of view  The larger the A , the more information can be accessed Telescope  1m4m CFHT8m VLT40m EELT Field of view1 deg²0.25 deg²0.125 deg²0.025 deg² AA 1111 These instrument systems have the same efficiency

11. Survey depth  Depends on  Telescope diameter  Instrument throughput (optical efficiency)  Exposure time  Detector noise  Background  Signal to noise S/N Source Background Detector Noise Det. Dark current Source N photons

12. Survey redshift range The redshift range will determine the wavelength range (and vice-versa)

13. Survey spectral resolution  Ability to separate spectral features  R= /d  The higher R, the better is the velocity resolution, or velocity accuracy  Choice depends on the spectral features you are interested into  Broad features (e.g. because of velocity dispersion) or narrow

14. Survey number of objects  A key number: 10 4 objects Why ?

15. 15 Nobj ?? ~10 5 !! –Study evolution vs. Luminosity, color (type), environnement –Minimise cosmic variance effects: survey several independant fields –Several time intervals to follow evolution –50 galaxies per measurement bin –Total number of galaxies: 50  10  3  3  4  7 > 100000 per bin mag.bin colors env. fields time steps

16. Science vs. Parameter space matrix Science Goals Survey parameters Area -range microns Spectral RMag. Lim.Nobj Goal 11 deg²0.36-125024.510000 Goal 20.5 deg²0.55-11000256000 Goal 33 deg²0.35-0.82502450000 … Compile all science goals into one single survey observing strategy

17. Examples of spectroscopic survey design Survey Survey Design Parameters Area -range microns z-rangeSpectral R Mag. Lim. Nobj SDSS-III10000 deg²0.36-0.90-0.520001810000 VVDS-Wide8 deg²0.55-10-1.525022.522500 VVDS-Deep1 deg²0.55-10-52502412500 VVDS- UltraDeep 500 arcmin²0.36-10-6.725024.751000 VIPERS25 deg²0.5-10.5-1.52502450000 VUDS1 deg²0.36-12-6+2502510000 Euclid15000 deg²0.95-1.80.8-230022 50  10 6

18. Which instrument for my survey ?  Imaging or spectroscopy ?  Need both !  Need more ?

19. 14 Instruments at the VLT (and VLTI, VISTA, VST)

20. Imaging cameras  Based on CCDs for the visible domain  Based on HgCdTe arrays for 1-5 microns  Other hybrid detectors in UV and to ~25 microns  Radio and sub-mm recievers  X-ray cameras  …. Key elements  Field of view  Wavelength domain  Spatial resolution  Throughput / Quantum efficiency

21. CFHT in Hawaii Visible cameras: CFHT 3.6m+Megacam MegaCam: 256 millions pixels ParameterValue Field of view1 deg² -range 0.33-1 microns Pixel scale0.2 arcsec Filtersugriz

22. IR cameras: on 4m VISTA at ESO ParameterValue Field of view0.6 deg² -range 0.8-2.5 microns Pixel scale0.34 arcsec FiltersYJHK

23. HST imaging  The best resolution  The best sensitivity  The smallest field ParameterValue Field of view11 arcmin² -range 0.35-1 microns Pixel scale0.05 arcsec FiltersUbvriz-like ParameterValue Field of view4.6 arcmin² -range 0.8-1.7 microns Pixel scale0.13 arcsec FilterszYJH ACS WFC3

24. Optical filters, are interference filters, can be quite complex multi-layer coatings. Optical filters selectively transmits light in a given bandpass, while blocking the remainder. Many VLT Instruments (FORS1+2, ISAAC, NACO, VIMOS, VISIR) are camera which provide direct imagery through sets of filters. Throughput for imaging cameras

25. Limiting magnitudes: imaging Example of the COSMOS survey

26. MOS: multi-object spectrographs  A key invention for Cosmology !  Principle: observe more than one object at once  Multiplex N obj  The multiplex is like having N obj telescopes each observing 1 object  Different types of MOS  Multi-slit: better sky subtraction  Multi-fiber: wide field  Multi-IFU: velocity fields Key elements  Field of view  Wavelength domain  Spectral resolution  Multiplex  Throughput

27. Spectra, one by one E. Hubble 27

28. Multi-object spectroscopy Deep multi-color imaging Multi-object spectroscopy Target selection Today MOS have N obj >> 100 Multiplies the efficiency of your telescope by N obj ! 28

29. Multi-Object Spectrograph have become the work-horse of many observatories  In all major observatories: CFHT-MOS/SIS, Keck-LRIS, VLT-FORS, GMOS, Keck- DEIMOS, VLT-VIMOS, IMACS …  Now going to the IR: MOSFIRE, VLT-KMOS VIMOS DEIMOS FORS2 IMACS 29

30. VIMOS on the VLT VIMOS was build by a consortium of french+italian institutes, led by LAM ParameterValue Field of view220 arcmin² AperturesSlit mask -range 0.36-1 microns Pixel scale0.2 arcsec FiltersUgriz Spectral R250-2500 Number of slits~600

31. VIMOS  4 tons  80cm beam Optical train

32. Vertical trace: one galaxy spectrum Horizontal lines: night sky emission VIMOS: 1000 spectra in one shot

33. DEIMOS on Keck ParameterValue Field of view80 arcmin² AperturesSlit mask -range 0.42-1 microns Pixel scale0.1 arcsec FiltersBVRIZ Spectral R1500-5000 Number of slits~120

34. SDSS spectrograph ParameterValue Field of view7 deg² AperturesFibers -range 0.38-92 microns Fiber size3 arcsec Spectral R2000 Number of fibers ~600

35. MOS in the IR: MOSFIRE on Keck ParameterValue Field of view45 arcmin² AperturesMoveable slits -range 0.8-2.5 microns Pixel scale0.1 arcsec FiltersYJHK Spectral R2000-5000 Number of slits45

36. KMOS: a multiple IR IFU on the VLT ParameterValue Field of view40 arcmin² AperturesMoveable IFU -range 0.8-2.5 microns Pixel scale0.2 arcsec Spectral R3000-4000 Number of slits24 IFU: Integral Field Unit

37. 10kpc MASSIV survey at z~1.5 37 Integral field spectroscopy: velocity fields

38. Instrument design and development  Instrument making is fundamental to astrophysics  Relies on new & improved technology  Optics, detectors, mechanics, control (active)  Space technology  Software: data processing, databases  Professional project development  Skilled instrument scientists and specialty engineers  Project management  Expensive telescopes (~1B€) and instruments (~15M€ ground-based / ~150M€ space-based) 38

39. Instrument development cycle  Define science goals: science requirements  Survey volume, number of objects, redshift  Derive technical requirements  Field of view, wavelength, resolution, throughput  Global performances  Produce strawman opto-mechanical design  Identify new technology developments: grating, detectors,…  Produce prototypes  Manufacture all parts  Assembly, integration and tests  Measure performances, calibrate  First light 39 T 0 T 0 +1y T 0 +2y T 0 +4y T 0 +5y T 0 +6-7y SPACE instruments: 2x longer !

40. Example: a MOS for the EELT DIORAMAS  Phase A completed in 2011  Call for tender: mid-2014  7y development  First light 2022 ? 40 Multi-slit

42. Observational methods  Sample selection  Observations preparation and follow-up  Measuring the sample selection function

43. Sample selection  Magnitude or flux selection  Color selection  Color-color selection  Photometric redshift selection

44. Magnitude / flux selection Observe all galaxies brighter than a limit Here: I AB ≤24

45. Color selection  Apply a color cut  Color=difference between two photometric bands  Here (magenta) select the red galaxies with Mu-Mr>1.4  Can add a magnitude selection on top (green): select all red and bright galaxies

46. Color-color selection  Select objects in a part of a color-color diagram  Most known: Lyman- break galaxy selection (LBG)  Here is shown a gzK diagram to select galaxies at z~2

47. Lyman-break galaxy selection  Use predicted galaxy tracks vs. Redshift to isolate galaxies in color-color space Different types of galaxies z=0 z=2.7 z=3.4

48. MOS observations preparation and follow-up  Produce a reference photometric catalog based on your selection  The “parent catalog”  Produce a list of objects to be observed in MOS  Based on geometric constraints of the instrument  Produce “Observing blocks” to be executed at the observatory

50. Observing blocks at VLT

51. Resulting observations

52. Measuring the selection function  Masking bad regions  Target Sampling rate  Limiting magnitude  Limiting magnitude (at 5  or at 90% completness)

53. Measuring the selection function  Spectroscopic success rate  As a function of magnitude  As a function of redshift  …

54. Past and present deep spectroscopic surveys SurveyInstrumentredshift# galaxies 2dFGRS2dF/AAT0<z<0.5220000 SDSSSDSS/Apache Point0<z<0.5930000 CFRS – 1995CFHT-MOS0<z<1.2600 LBG – 1999KECK-LRIS2.5<z<4.51000 DEEP2, 2005+KECK-DEIMOS0.7<z<1.450000 VVDS, 2005+VLT-VIMOS0<z<550000 zCOSMOS, 2007+VLT-VIMOS0<z<1.2 1.4<z<3 20000 10000 VIPERS, 2009+VLT-VIMOS0.5<z<1.2100000 VUDS, 2010+VLT-VIMOS2.5<z<6.710000 GOODSVLT FORS20<z<7.11000 And more ! 54

55. Survey examples: SDSS the Sloan Digital Sky Survey  Small telescope but large A   Low redshift 0<z<0.5 but large area/volume Survey parameterValue Telescope diameter2.5m Field of view7 deg² Area10000 deg² Redshift range0<z<0.5 -range 3800-9200A DepthR~18 R2000 Nspec1 million

56. VVDS: the VIMOS VLT Deep Survey  Magnitude selection Survey parameterValue Telescope diameter8.2m Field of view220 arcmin² Area8.7 deg² Redshift range0<z<6.7 -range 3600-9500A DepthI AB =24.75 R230 Nspec35 000

57. VIPERS: the VIMOS Public Extragalactic Survey  Magnitude AND color selection Survey parameterValue Telescope diameter8.2m Field of view220 arcmin² Area25 deg² Redshift range0.5<z<1.5 -range 5500-9500A Depthi AB =22.5 R230 Nspec100 000

58. 4. Data processing  Imaging and spectroscopy generate hundreds of Gb of data per night  Process from raw uncalibrated data to instrument-corrected and calibrated data  A very important step  Data processing packages available for each instrument  Need to invest time before using them to the best

59. 5. Information system, Databases  Data volume from surveys is huge, many Tb  Information system: the management of all survey data  Data flow  Includes all steps: from design to observations, to data processing, to final measurements  Easy access to data  Query oriented  Long term access, and reference  Virtual observatory compatible

60. Follow the observations and data processing

63. 6. Comparing surveys  Not so easy because the selection functions can be very different  Look at the parameter space  Nobj vs. Area and vs. Magnitude  Area vs. Magnitude  Nobj and Area vs. Redshift  …

67. Who’s the best survey ??