Clustering redshift distribution calibration of weak lensing surveys using the DESI-DR1 spectroscopic dataset

arXiv:2512.15963 · astro-ph.CO · Submitted 2025-12-17 · Read on arXiv

Listen

Radio episode about this paper

Transcript

Introduction to the show: ident: Astrophysics Radio. Generated commentary on the latest astrophysics papers.

Vera: We're looking at the paper: "Clustering redshift distribution calibration of weak lensing surveys using the DESI-DR1 spectroscopic dataset".

Jocelyn: Here's the gist of the paper. We estimate the source redshift distribution of current weak lensing surveys by applying the clustering-based redshift calibration technique, using the galaxy redshift sample provided by the Dark Energy Spectroscopic Instrument Data Release 1 (DESI-DR1) The clustering redshifts enabled by DESI’s unprecedented spectroscopic sample provide a robust, complementary, and independent constraint capable of reducing one of the dominant systematic uncertainties in weak lensing cosmology Pruning redundant reasoning tokens The study applies the clustering-z technique to calibrate the redshift distributions of tomographic weak lensing samples from DES, KiDS, and HSC using DESI-DR1 spectroscopy as a reference sample The methodology involves cross-correlating Bright Galaxy Survey (BGS), Luminous Red Galaxies (LRGs), and Emission Line Galaxies (ELGs) from DESI within the redshift range 0.1 < z < 1.6 with overlapping tomographic source samples from DES, KiDS, and HSC surveys The analysis tests the stability of the clustering-redshift signal to fitting scale, reference-sample choice, and the evolution of source galaxy bias while explicitly modeling and marginalizing over magnification contributions which become non-negligible at z ≳ 1 How the training signal is built The clustering-z approach is based on the cross-correlation signal between a sample with unknown redshift probability distribution, pu(z), and a reference sample for which the redshift distribution, pr(z), is accurately measured through spectroscopy The angular cross-power spectrum is modeled using the Limber approximation to relate the angular correlation function to the underlying cosmological model parameters like linear galaxy bias bu(z) and matter power spectrum Pm(k, z) To account for magnification bias, a likelihood explicitly includes magnification terms in theoretical templates by introducing additional free parameters connected to the galaxy–magnification angular power spectra which are then marginalized over in the inference Where the gains come from The study utilizes three primary reference populations from DESI-DR1: the Bright Galaxy Survey (BGS) at low redshift, Luminous Red Galaxies (LRGs) at intermediate redshifts, and Emission Line Galaxies (ELGs) at higher redshifts The analysis measures angular auto-correlation functions wrr(θ) for the DESI reference samples in narrow redshift bins to determine the clustering amplitude Arr It also measures angular cross-correlation functions wur(θ) between the DESI reference samples and each of the tomographic sub-samples of every weak lensing survey The posterior estimates for the product bu(zr)pu(zr) are derived by fitting these correlation amplitudes, accounting for magnification contributions p and q Validation and Results The methodology is validated using mock catalogues sampled from the Buzzard N-body simulations to test pipeline robustness against realistic galaxy populations, source weights, and photometric redshift errors The analysis shows good agreement between the error estimates obtained from mock-to-mock scatter, jackknife resampling, and analytic Gaussian covariance predictions For real data, the clustering-z constraints are consistent with the fiducial redshift distributions of DES-Y3, KiDS-1000, HSC-Y1 and HSC-Y3 largely within quoted uncertainties The analysis demonstrates that magnification contributes non-negligibly at z ≳ 1, particularly for HSC and the upper tomographic bins of DES and KiDS Marginalizing over source bias evolution Since the source galaxy bias bu(z) is not independently determined from the cross-correlation measurement, recovering the underlying redshift distribution pu(z) requires marginalizing over the possible redshift evolution of bu(z) This is done using a smooth linear parameterization, bu(z) = b0 + b1z, which captures the expected monotonic increase of galaxy bias with redshift The resulting uncertainty budget includes statistical noise in the clustering measurement, marginalisation over magnification parameters (p, q), and uncertainty in the redshift evolution of the source bias The present implementation introduces an uncertainty contribution associated with marginalising over source bias evolution which is the dominant term in the final error budget at high redshift The techniques developed are directly applicable to Stage-IV surveys such as LSST and Euclid where spectroscopic crosscorrelations will remain essential for achieving the precision required for dark energy and modified gravity constraints The analysis presented here forms the basis of the redshift calibration pipeline for future DESI-DR2 joint 3×2-pt inference where redshift calibration requirements approach the sub-percent level

Title and authors: Vera: Let's talk about the title and who wrote this, "Clustering redshift distribution calibration of weak lensing surveys using the DESI-DR1 spectroscopic dataset." It sounds very technical, but it’s just saying they are using clustering to fix redshifts for weak lensing data with the DESI spectroscopic sample.

Jocelyn: The authors are a big group of researchers from different places, including people involved with the Bright Galaxy Survey and Luminous Red Galaxies from DESI itself. They bring together observational astronomy and cosmology expertise here.

Subrahmanyan: That combination is key because they’re tackling this problem using clustering cross-correlation signals, which directly relate to how matter is distributed in the universe at different depths. It suggests a systematic way to disentangle the source redshift from cosmological parameters like the matter power spectrum.

Vera: So, what's the core idea they are proposing? They are using cross-correlations between known spectroscopic samples and unknown weak lensing tomographic samples to infer those unknown redshifts.

The paper's summary: Jocelyn: The summary says they apply this clustering technique to calibrate the redshift distributions for three different weak lensing surveys: DES, KiDS, and HSC. They use the DESI-DR1 spectroscopy as their reference sample to do this calibration.

Vera: It sounds like they are cross-correlating specific galaxy types from DESI—like the Bright Galaxy Survey, Luminous Red Galaxies, and Emission Line Galaxies—with the tomographic source samples from those three surveys.

Subrahmanyan: That’s the methodology: they test this signal across various fitting scales and reference samples to check if their redshift estimation holds up. They are also modeling things like magnification contributions that get important when we look at redshifts higher than one point, which is a crucial detail <ref:2512.15963#pg1>.

Jocelyn: To put it simply, they’re using the DESI data to build a robust way to assign depths to galaxies in those other surveys so we can measure cosmology accurately.

The paper's improvements: Vera: Now, the paper points out some specific ways they’ve improved this approach. They explicitly introduce modeling magnification contributions by using additional free parameters connected to the galaxy–magnification angular power spectra, which they then marginalize over in their final inference.

Subrahmanyan: That’s smart because magnification bias becomes non-negligible at redshifts greater than one especially for surveys like HSC and the upper bins of DES and KiDS <ref:2512.15963#pg1>. By modeling it this way, they are making the calibration more self-consistent with those higher redshift effects.

Jocelyn: They also have a more complete uncertainty budget now. It includes statistical noise from the clustering measurement, marginalization over those magnification parameters p and q, and uncertainty in how the source galaxy bias evolves with redshift.

Vera: That's really important because they explicitly state that the marginalization over source bias evolution is becoming the dominant term in their error budget at high redshift. It’s a very conservative way to estimate what we don't know.

Conclusion: Jocelyn: So, to wrap things up, these techniques are being tested against mock catalogues from the Buzzard N-body simulations to make sure the pipeline is robust against real galaxy populations and photometric redshift errors.

Vera: And for real data, the constraints they get from this clustering-z method are consistent with what we already expect from other redshift determinations like DES-Y3, KiDS-one thousand HSC-Y1 and HSC-Y3 <ref:2512.15963#pg1>.

Subrahmanyan: The implication here is that this clustering approach provides a strong constraint on the unknown redshift distributions for future surveys. It’s directly applicable to Stage-IV surveys like LSST and Euclid where these spectroscopic crosscorrelations will remain essential for achieving the precision needed for dark energy and modified gravity constraints.

Jocelyn: And they’ve even laid the groundwork for a joint inference pipeline with DESI-DR2, aiming for sub-percent level redshift calibration requirements. It shows how essential this type of spectroscopic reference data is moving forward in cosmology.

Queensland University of Technology · Centre for Astrophysics & Supercomputing Swinburne University of Technology · Lawrence Berkeley National Laboratory · Department of Physics Boston University · Dipartimento di Fisica ’Aldo Pontremoli Università di Milano · INAF-Osservatorio Astronomico di Brera Milano, Italy · Department of Physics & Astronomy University College London United Kingdom · Institut d’Estudis Espacials de Catalunya Institute of Space Sciences ICE-CSIC Barcelona Spain · Department of Physics and Astronomy The University of Utah · Institute de Física Universidad Nacional Autónoma de México · Department of Astronomy & Astrophysics University of Toronto Canada · Department of Physics and Astronomy and Pittsburgh Particle Physics Astrophysics, and Cosmology Center University of Pittsburgh USA · Department of Physics The Ohio State University USA · University of California Berkeley USA

astro-ph.CO

Submitted: 2025-12-17

Updated: 2026-11-05

Comments: Published in the Open Journal of Astrophysics, 20 pages, 10 figures

Code: https://github.com/rmjarvis/TreeCorr

License: http://creativecommons.org/licenses/by/4.0/

Importance score: 91/100

The gist: We estimate the source redshift distribution of current weak lensing surveys by applying the clusteringbased redshift calibration technique…

Key concepts

Clustering-z Technique
This method calibrates unknown redshift distributions by cross-correlating a sample with an unknown probability distribution, pu(z), against a reference sample whose distribution, pr(z), is known from spectroscopy. It uses angular cross-power spectra modeled via the Limber approximation to link clustering signals to cosmological parameters.
Magnification Bias
This effect occurs when the magnification of distant galaxies changes their observed flux, leading to an apparent increase in the number of detected sources at certain magnitudes. The analysis explicitly models this by including terms in theoretical templates related to galaxy–magnification angular power spectra and marginalizing over these parameters.
Source Bias Evolution
Since the source galaxy bias, bu(z), is not directly measured, recovering the true redshift distribution requires modeling its evolution. A smooth linear parameterization, bu(z) = b0 + b1z, is used to capture how this bias changes with redshift to ensure stable results across different bins.
Uncertainty Budget Dominance
The final error budget for the calibration pipeline is dominated by the uncertainty introduced when marginalizing over the evolution of source bias at high redshifts. This indicates that accurately modeling how galaxy properties change with distance is crucial for achieving high precision in future surveys.

Terminology

Summary

We estimate the source redshift distribution of current weak lensing surveys by applying the clusteringbased redshift calibration technique, using the galaxy redshift sample provided by the Dark Energy Spectroscopic Instrument Data Release 1 (DESI-DR1) The clustering redshifts enabled by DESI’s unprecedented spectroscopic sample provides a robust, complementary, and independent constraint capable of reducing one of the dominant systematic uncertainties in weak lensing cosmology

Methodology for Redshift Inference

The technique employed is the clustering-z or “cross-correlation redshifts” method, which infers the redshift distribution of a photometric sample through angular cross-correlations with a spectroscopic reference sample of known redshifts This approach relies on modeling the angular cross-power spectrum between an unknown sample with probability distribution pu(z) and a reference sample with distribution pr(z) using the Limber approximation The angular galaxy cross-correlation function wur(θ) is related to this power spectrum via an integral over multipoles To produce these models, the authors integrate the angular power spectrum across multipoles 1 ≤ l ≤ 10 5 The lens-source (reference-unknown) cross-correlation function is expressed as wur(θ) = pu(zr) bu(zr) br(zr) ∆z wm(θ), where wm(θ) is the matter auto-correlation function

Incorporating Systematics and Magnification

Magnification bias arises because weak gravitational lensing by large-scale structure alters both the observed fluxes and apparent sizes of galaxies, potentially introducing correlations between physically unassociated populations at different redshifts The contribution of magnification to the angular cross-correlation function is given by wur(θ) = brαu wmag ru (θ) + buαr wmag ur (θ), where p and q are parameters representing the product of the galaxy bias and number-count slopes The authors model this contribution by introducing additional free parameters connected to the galaxy–magnification angular power spectra, which they marginalise over in the inference

Data Sources and Validation

The study utilizes spectroscopic redshifts from the DESI-DR1 dataset, specifically using samples from the Bright Galaxy Survey (BGS), Luminous Red Galaxies (LRGs), and Emission Line Galaxies (ELGs) as reference populations The analysis applies this technique to calibrate photometric redshift distributions of tomographic weak lensing samples from DES, KiDS, and HSC surveys Validation is performed using mock catalogues sampled from the Buzzard N-body simulations The mocks are tuned to match the redshift distribution, photometric-redshift scatter, source weights and tomographic binning of the KiDS-1000, DES-Y3 and HSC-Y1 datasets

Recovery of Source Bias Evolution

Since the source galaxy bias bu(z) is not independently determined from the cross-correlation measurement, recovering the underlying redshift distribution pu(z) requires marginalising over the possible redshift evolution of bu(z) This is done using a smooth linear parameterization, bu(z) = b0 + b1z, which captures the expected monotonic increase of galaxy bias with redshift The resulting uncertainty budget includes statistical noise in the clustering measurement, marginalisation over magnification parameters (p, q), and uncertainty in the redshift evolution of the source bias

Conclusion on Calibration Performance

The clustering-z posteriors are consistent with the directly-calibrated redshift distributions of DES-Y3, KiDS-1000, HSC-Y1 and HSC-Y3, largely within quoted uncertainties The analysis demonstrates that magnification contributes non-negligibly at z ≳ 1, particularly for HSC and the upper tomographic bins of DES and KiDS The techniques developed are directly applicable to Stage-IV surveys such as LSST and Euclid, where spectroscopic crosscorrelations will remain essential for achieving the precision required for dark energy and modified gravity constraints

Table 1 Summary of Results

Table 1 summarizes the mean redshift of the measured bu(zr)pu(zr) distributions and fiducial model, for each survey and tomographic bin It reports the mean redshift of the recovered clustering-z estimator ¯zmeas and its propagated uncertainty σ(¯zmeas), along with the normalized shifts σ∆z ≡ (¯zmeas − z¯fid)/σ(¯zmeas) The results show that the recovered means are typically within a statistical error margin of the true values

Figure 4 Analysis of Bias Evolution

Figure 4 shows the evolution of the source bias bu(zr) across surveys, as determined from Buzzard mock catalogues for each imaging survey The rise of bu(zr) with redshift reflects the increasing typical halo mass of galaxies selected at higher redshift The trends are smooth and consistent with expectations from galaxy–halo connection models

Figure 6 and Figure 8 Posterior Estimates

Figure 6 and Figure 8 show the posterior estimates of the bias–weighted source redshift distribution bu(zr) pu(zr) in each DES-Y3 tomographic bin, with 1σ uncertainties The red points correspond to the baseline analysis with magnification contributions included, and blue points show the result when magnification is excluded

Figure 7 and Figure 9 Posterior Estimates

Figure 7 and Figure 9 show posterior estimates of the bias-weighted source redshift distribution bu(zr) pu(zr) in each KiDS-1000 tomographic bin, with 1σ uncertainties The black dashed curve shows the survey’s fiducial redshift distribution, and the red and blue points corresponding to analyses with and without magnification, respectively

Figure 4 Analysis of Component Decomposition

Figure 4 shows component decomposition of the angular cross-correlation wur(θ) between DES-Y3 source galaxies and the DESI spectroscopic reference sample The curves show the best-fit clustering term Aurwm(θ) (the orange lines), the magnification contributions p wmag ru (the green lines) and q wmag ur (the red lines), and their sum (the black lines)

Figure 5 Validation using Mocks

Figure 5 shows the clustering-z validation using mock catalogues for DESI and DES-Y3 The red points show the raw clustering-based estimate of bu(zr) pu(zr), measured from wur(θ) in discrete redshift slices The reasonable agreement between the points and the solid black curve demonstrates that the method accurately recovers the source bias–weighted distribution when applied to realistic mock data

Figure 7 and Figure 8 Cross-Correlation Visualization

Figure 7 and Figure 8 show example cross-correlation measurements wur(θ) between DES-Y3 source galaxies and the DESI spectroscopic reference sample The points are measurements with jackknife errors; the black line is the best-fitting model Aurwm + p wmag ru + q wmag ur The variation of wur(θ) across bins traces the redshift distribution of the unknown sample, together with the evolution of the clustering amplitude with redshift and the scale dependence of the magnification contributions p and q

Figure 10 Comparison of Tracer Consistency

Figure 10 compares the recovered bu(zr) pu(zr) distributions when using LRGs (blue) or ELGs (red) as the DESI spectroscopic reference sample, in each DES-Y3 tomographic bin The comparison between the two schemes reveals a consistent picture In every dataset, the inferred bu(zr) posteriors are stable across all tomographic bins, with the two configurations overlap at the intermediate redshifts where the LRG sample dominates

Figure 6 and Figure 7 Analysis of HSC Results

Figure 8 and Figure 9 show posterior estimates of the bias-weighted source redshift distribution bu(zr) pu(zr) in each HSC-Y1 tomographic bin, with 1σ uncertainties The agreement with the template is generally good The mean-redshift shifts reported in Table 2 show that all bins are within σ∆z < 1

Table 2 Summary of Redshift Recovery

Table 2 reports the mean redshift of the measured bu(zr)pu(zr) distributions and fiducial model, for each survey and tomographic bin The difference between the bias-weighted and unweighted fiducial means is small (z¯(bupu)fid − z¯(pu)fid ≲ 0.04), and the recovered means are typically within a statistical error margin of the true values

Conclusion on Error Budget

The total error budget on the recovered bu(z) pu(z) – including jackknife covariance and marginalisation over magnification nuisance parameters – is consistent with the mock-to-mock variance The present implementation introduces an uncertainty contribution associated with marginalising over source bias evolution, which is the dominant term in the final error budget at high redshift Future work will explore strategies to further constrain bu(z), for example by combining clustering-z with shear–density cross-correlations, galaxy–galaxy lensing, or halo-occupation based forwardmodelling The analysis presented here forms the basis of the redshift calibration pipeline for future DESI-DR2 joint 3×2-pt inference The techniques developed in this work are also directly applicable to Stage-IV surveys such as LSST and Euclid

Table 1 Summary of Final Results

Table 1 lists the mean redshift of the true bias-weighted distribution ¯z(bupu)fid, the underlying source distribution ¯z(pu)fid (i.e. with bu = 1); the mean redshift of the recovered clustering-z estimator ¯zmeas and its propagated uncertainty σ(¯zmeas), and the normalized shifts σ∆z ≡ (¯zmeas − z¯fid)/σ(¯zmeas) The difference between the bias-weighted and unweighted fiducial means is small (z¯(bupu)fid − z¯(pu)fid ≲ 0.04), and the recovered means are typically within a statistical error margin of the true values

Table 2 Summary of Survey Performance

Table 2 reports the mean redshift of the measured bu(zr)pu(zr) distributions and fiducial model, for each survey and tomographic bin The shifts σ∆z show that the clustering-z pipeline recovers unbiased centroids for the distributions within the statistical error range for most tomographic bins The largest departures reaching ∼1.5–2σ only in the highest-redshift HSC-Y1 bin

Table 6 Summary of Posterior Estimates

Figure 6 shows the posterior estimates of the bias–weighted source redshift distribution bu(zr) pu(zr) in each DES-Y3 tomographic bin, with 1σ uncertainties The red points correspond to the baseline analysis with magnification contributions included, and blue points show the result when magnification is excluded

Table 7 Summary of KiDS-1000 Results

Figure 7 shows posterior estimates of the bias-weighted source redshift distribution bu(zr) pu(zr) in each KiDS-1000 tomographic bin, with 1σ uncertainties All σ∆z values are within ∼ 1σ agreement of the fiducial model

Table 8 Summary of HSC-Y1 Results

Figure 8 shows posterior estimates of the bias-weighted source redshift distribution bu(zr) pu(zr) in each HSC-Y1 tomographic bin, with 1σ uncertainties The shape match to the model is good and consistent with this

Table 9 Summary of HSC-Y3 Results

Figure 9 shows posterior estimates of the bias-weighted source redshift distribution bu(zr) pu(zr) in each HSC-Y3 tomographic bin, with 1σ uncertainties The agreement with the template is generally good

Table 2 Summary of Survey Performance (Final)

Table 2 reports the mean redshift of the measured bu(zr)pu(zr) distributions and fiducial model, for each survey and tomographic bin The shifts σ∆z show that the clustering-z pipeline recovers unbiased centroids for the distributions within the statistical error range for most tomographic bins The largest departures reaching ∼1.5–2σ only in the highest-redshift HSC-Y1 bin

Figure 6 Analysis of Mock Validation

Figure 6 shows the posterior estimates of the bias–weighted source redshift distribution bu(zr) pu(zr) in each DES-Y3 tomographic bin, with 1σ uncertainties The shaded coloured band shows the final clustering-z estimate after smoothing, together with the propagated uncertainty from jackknife resampling and after marginalization over the magnification nuisance parameters p and q

Figure 4 Analysis of Mock Validation

Figure 4 shows the mean source-galaxy clustering bias bu(zr) inferred from the mock catalogues for each imaging survey The rise of bu(zr) with redshift reflects the increasing typical halo mass of galaxies selected at higher redshift The trends are smooth and consistent with expectations from galaxy–halo connection models

Figure 3 Component Decomposition

Figure 3 shows the component decomposition of the angular cross-correlation wur(θ) between DES-Y3 source galaxies and the DESI spectroscopic reference sample The markers show the measured correlations with jackknife uncertainties The curves show the best-fit clustering term Aurwm(θ) (the orange lines), the magnification contributions p wmag ru (the green lines) and q wmag ur (the red lines), and their sum (the black lines) The variation of wur(θ) across bins traces the redshift distribution of the unknown sample, together with the evolution of the clustering amplitude with redshift and the scale dependence of the magnification contributions p and q

Figure 4 Analysis in Detail

Figure 4 shows how we determine source bias bu(zr) for each imag- survey The amplitude Auu was obtained by minimizing χ 2(Auu) = ∆⃗w TC−1∆⃗w, with flat priors Auu ∈ (0, 20) The source bias was then determined as bu(zr) = ⟨A 1/2 uu (zr)⟩reg

Figure 4 Analysis of Bias-Weighted Distribution

Figure 4 shows the evolution of the source bias bu(zr) across surveys, as determined from the Buzzard mock catalogues for each imaging survey The solid line represents the average over the ten mock regions, and the shaded band indicates the region-to-region scatter

Figure 7 Analysis of Cross-Correlation Components

Figure 7 shows example cross-correlation measurements wur(θ) between DES-Y3 source galaxies and the DESI spectroscopic reference sample The points are measurements with jackknife errors; the black line is the best-fitting model Aurwm + p wmag ru + q wmag ur The variation of wur(θ) across bins traces the redshift distribution of the unknown sample, together with the evolution of the clustering amplitude with redshift and the scale dependence of the magnification contributions p and q

Table 2 Summary for HSC Results

Table 2 reports the mean redshift of the measured bu(zr)pu(zr) distributions and fiducial model, for each survey and tomographic bin The shifts σ∆z show that all bins are within σ∆z < 1 The largest departures reaching ∼1.

Green — the paper says it, and the page is printed. Red — no passage carries it, so treat that sentence as unverified. The same check runs on your own text: check a summary against its source, $5.

Improvements for AI systems

  1. Improve source redshift distribution recovery by explicitly modeling magnification contributions by introducing additional free parameters connected to the galaxy–magnification angular power spectra, which we marginalise over in the inference. This allows for a selfconsistent way to propagate magnification-induced systematics into the redshift calibration errors, especially at higher redshifts where this contribution becomes non-negligible.

  2. Enhance systematic uncertainty estimation by using a complex uncertainty budget that includes (i) statistical noise in the clustering measurement, (ii) marginalisation over magnification parameters (p, q), and (iii) uncertainty in the redshift evolution of the source bias. This ensures that results are conservative and internally calibrated uncertainty estimates, particularly noting that the present implementation introduces an uncertainty contribution associated with marginalising over source bias evolution. This is the dominant term in the final error budget at high redshift.

  3. Strengthen model robustness by testing the sensitivity of clustering-z results to different scale cuts, such as a large-scale configuration using only separations above 5 h −1Mpc and a mid-scale configuration defined by Rmin = 3 h −1Mpc and Rmax = 20 h −1Mpc. This allows the AI system to assess whether the inferred redshift distributions are stable across varying assumptions about galaxy bias scaling at different scales.

  4. Improve tracer consistency assessment by testing the internal consistency of clustering-z results using different spectroscopic tracers, such as comparing the inferred distributions of bupu when adopting the default tracer selection: tracer = BGS, z 1.1 against an alternative configuration where the LRG/ELG transition occurs at z = 0.8.

  5. Refine source bias evolution modeling by using a smooth linear parameterization for the source galaxy bias, specifically bu(z) = b0 + b1z, which captures the expected monotonic increase of galaxy bias with redshift while preventing unphysical bin-to-bin fluctuations in low signal-to-noise regimes.

Abstract

We estimate the source redshift distribution of current weak lensing surveys by applying the clustering-based redshift calibration technique, using the galaxy redshift sample provided by the Dark Energy Spectroscopic Instrument Data Release 1 (DESI-DR1). We cross-correlate the Bright Galaxy Survey (BGS), Luminous Red Galaxies (LRGs) and Emission Line Galaxies (ELGs) from DESI, within the redshift range 0.1 < z < 1.6, with overlapping tomographic source samples from the Dark Energy Survey (DES), Kilo-Degree Survey (KiDS), and Hyper Suprime-Cam (HSC) survey. Using realistic mock catalogues, we test the stability of the clustering redshift signal to fitting scale, reference-sample choice, and the evolution of source galaxy bias. We calibrate a source galaxy bias correction from simulations with conservative (factor-of-two inflated) priors to account for modelling uncertainties, and we explicitly model and marginalise over magnification contributions, which become non-negligible at z 1 due to the depth of the DESI ELG sample. We then compare the resulting bias-weighted redshift distributions to, finding agreement within uncertainties for all surveys and tomographic bins. Our results demonstrate that clustering redshifts enabled by DESI's unprecedented spectroscopic sample provides a robust, complementary, and independent constraint capable of reducing one of the dominant systematic uncertainties in weak lensing cosmology.

Sources

Related papers