*chto@uchicago.edu

Synergies between Roman and Rubin: weak lensing point of view

Chun-Hao To*

Roman High Latitude Imaging Survey Project Infrastructure Team (HLIS-PIT)

  • We build pipelines to enable Roman's 3x2pt + cluster science.
    • Shear catalogs and calibrations (such as m, photo-z, ...). 
    • Lens and cluster catalogs and calibrations (such as weights, photo-z, ...). 
    • Associated products for the above, such as coadd images, PSF, theory pipeline...
  • See https://roman-hlis-cosmology.caltech.edu/page/products  for expected products and timeline. 
  • Wavelength coverage

    •  

Synergies between Roman and Rubin

https://rtn-011.lsst.io/

  • Wavelength coverage
    • Roman will help Rubin Photo-z

Synergies between Roman and Rubin

  • Wavelength coverage
    • Rubin will help Roman Photo-z

Synergies between Roman and Rubin

  • Wavelength coverage
    • Both Roman and Rubin need spec-z for photo-z. 

Synergies between Roman and Rubin

Link

Cosmos DDF

XMM DDF

20k-30k spectra with H depth of ~24.5 (AB), subsampled to have flatter mag distribution.

Contact:
Brett Andrews, Jeff Newman,
Dan Master.

  • Wavelength coverage
    • Photo-z. 

Synergies between Roman and Rubin

  • Wavelength coverage
    • Photo-z.
  • Resolution: 
    • Rubin: PSF ~0.7" FWHM 
    • Roman: ~0.1" FWHM 

Synergies between Roman and Rubin

We are expecting to get to this depth in COSMOS DDF ~ summer 2027

Link

Performance: Rubin

Rubin Y5

Roman Medium Tier

Rubin detection

Roman detection

  • Wavelength coverage

    • Photo-z.

  • Resolution:

    • Deblending.

    • Star-galaxy separation.

Synergies between Roman and Rubin

See also:
Jain+15, Rhodes+19, Eifler+19, Chary+19, Bechtol+23, Openuniverse+24, Bianco+24, and many more...

  • Forced photometry using Rubin coadd images based on Roman detections. 

One step toward pixel-level synergies between Roman and Rubin

What are the unique challenges for Roman photometry? 

  • The point spread function can be undersampled: Pixel size ~= PSF size (0.1")
    • Solution: specialized image co-addition algorithm (IMCOM) to ensure that the output PSF is exactly Gaussian and uniform across the footprint. 

See also: Rowe+11, Hirata+24, Yamamoto+24, Cao+25

 

Imcom PSF

Drizzle PSF

What are the unique challenges for Roman photometry? 

  • Ground-based telescope (LSST/DES):

     Sky background ~ 2000 e/p/s vs Read noise ~ 8.8 e/p
           Poisson shot noise.
  • Space-based telescope (Roman/JWST):

    Zodiacal+thermal ~ 0.76 e/p/s vs Read noise ~ 8.5 e/p
          readout noise is no longer negligible, which can be correlated across pixels (1/f noise). 

See also, Rauscher+22, Betti+ 24

Laliotis+ 25

ImDeStripe: method subtracting 1/f noise using multiple overlapping observations 

Even with this, correlation remains

Generated with remaining noise powerspectrum  

Impact on flux uncertainty

MonteCarlo Simulation 

Out-of-the-box Flux Uncertainty.

Photometry measurement

  • Unique challenges for Roman: undersampled PSF and correlated noise. 
  • Places we use photometry:
    • Selecting galaxy samples for clustering and weak lensing.
    • Dividing galaxy samples into multiple redshift bins.
    • Estimating redshift distribution for galaxies in each redshift bin.
  • Requirements:
    • Reliable color and color error estimations, whose performance is not correlated with observation properties (such as depth, crowdedness, observing condition, etc.).

SlimFarmer

  • Optimized for weak lensing and photometric galaxy and galaxy cluster clustering analyses.
  • Support forced photometry on Rubin coadd.  
  • Mostly follow the Farmer (Weaver+23):
    • Detection: SEP on Y106, J129, H158 mean-combined maps.
    • Photometry: Multi-object model fitting (MOF) using the Tractor.
  • The following improvements:
    • Correlated noise correction using noise realization fields.
    • Fine-tuned model decision tree for Roman image.
    • Noise from astronomical objects estimated using the modeled photometry.

Astronomy 101: How to calculate flux error? 

\text{Noise} \approx \sqrt{N_{\text{galaxy}} + N_{\text{sky}} + N_{\text{readout}}^2}
  • The Farmer and LSST-DM: From the image itself
  • Fitvd: Estimate from residual map.  
  • Slimfarmer:
    • Estimated from the modeled flux at each optimization step.
    • The noise is scaled using residual map (empirical way to correct additional noise in DES)

Performance validation: DC25 Sim

RA

DEC

  • Input: OpenUniverse24.
     
  • A series of imaging process pipelines from raw data to coadded images with a uniform PSF. 

Performance validation: DC25 Sim

Performance validation: DC25 Sim

Roman DC25

Rubin LSST Y5, OpSim v3.2

Performance: Roman Color

Out-of-the-box uncertainty

Our method

Color Error

Performance: Crowdedness 

More Crowded

Less Crowded

Color Error

Forced photomety

  • Rubin coadd: Obtain the coadd image from the Butler, which contains WCS.
     
  • Forced photometry:
    • Take groups of Roman detections (in ra, dec). 
    • Rotate them to Rubin coadd using WCS (done in Tractor)
    • Vary their fluxes and positions with a position prior of 0.06" (0.3 of the pixel size). 
    • Jointly fit each group of galaxies. 

Performance: Forced Rubin Color

Color Error

Forced Rubin Color

  • Multi-object fitting (MOF): Fit overlapping objects jointly
  • Single-object fitting (SOF): Fit each object seperately

Roman MOF

Rubin MOF

Roman MOF

Rubin SOF

Roman SOF

Rubin SOF

Roman SOF

Rubin MOF

More Crowded

Less Crowded

More Crowded

Less Crowded

More Crowded

Less Crowded

More Crowded

Less Crowded

  • Flexzboost with perfect spec-z samples to isolate the impact of photometry measurement.
     
  • Single object fitting will lead to environment-dependent photo-z performance.

 

 

Impact on downstream analysis

More Crowded

Less Crowded

  • Flexzboost with perfect spec-z samples to isolate the impact of photometry measurement.
     
  • Single object fitting will lead to environment-dependent photo-z performance.

 

 

Impact on downstream analysis

More Crowded

Less Crowded

Impact on downstream analysis

$ ceci ../yaml/test_all.yaml

Roman-SOMPZ: Built on rail-sompz

  • Uncertainty quantification.
  • Much improved structure.

One single command
 

photometry → n(z) + uncertainty

Roman-SOMPZ Team

Implications: Photo-z binning

  • Ignoring correlated noise has a small impact on redshift bin assignments.
  • LSST photometry  is important (even with caveats of simulations).

Conclusions

  • Slimfarmer is a simple photometry pipeline that only does detection and photometry 
    • Correlated noise 
    • Multi-object fitting (MOF)
    • Forced Rubin photometry 
    • Tuned to work on Roman-like images
  • Using SlimFarmer, DC25 sim, and Roman-SOMPZ, we find:
    • MOF is only important for absolute photometry but not colors
    • Failure to account for correlated noise can underestimate flux uncertainties by several factors, but has only a small impact on SOMPZ.
    • Rubin photometry will be important for Roman weak lensing photo-z. 

Strong synergies between Roman and Rubin

  • Multiwavelength coverage:

    • Photo-z.
       

  • Resolutions:

    • Deblending.

    • Star-galaxy separation.

One step toward joint pixel processing

  • Slimfarmer is built to perform Roman photometry and Rubin forced photometry.
     

  • Correlated Noise: Failure to account for this will significantly underestimate flux uncertainties, but has minimal impact on photoz.
    ​

  • Multi-object fitting is needed to obtain unbiased colors.