Using Cross-Correlations to Calibrate Lensing Source Redshift Distributions: Improving Cosmological Constraints from Upcoming Weak Lensing Surveys (1306.0534v2)

Abstract: Cross-correlations between the galaxy number density in a lensing source sample and that in an overlapping spectroscopic sample can in principle be used to calibrate the lensing source redshift distribution. In this paper, we study in detail to what extent this cross-correlation method can mitigate loss of cosmological information in upcoming weak lensing surveys (combined with a CMB prior) due to lack of knowledge of the source distribution. We consider a scenario where photometric redshifts are available, and find that, unless the photometric redshift distribution p(z_{ph}|z) is calibrated very accurately a priori (bias and scatter known to ~0.002 for, e.g., EUCLID), the additional constraint on p(z_{ph}|z) from the cross correlation technique to a large extent restores the cosmological information originally lost due to the uncertainty in dn/dz(z). Considering only the gain in photo-z accuracy and not the additional cosmological information, enhancements of the dark energy figure of merit of up to a factor of 4 (40) can be achieved for a SuMIRe (Subaru Measurement of Images and Redshifts, the combination of the Hyper Suprime Cam lensing survey and the Prime Focus Spectrograph redshift survey)-like (EUCLID-like) combination of lensing and redshift surveys. However, the success of the method is strongly sensitive to our knowledge of the galaxy bias evolution in the source sample. If this bias is modeled by a free parameter in each of a large number of redshift bins, we find that a prior of order 0.01 is needed on b_i \sqrt{\Delta z} in each redshift slice (where \Delta z is the bin width and b_i the value of the galaxy bias in the i-th source bin) to optimize the gains from the cross-correlation method (i.e. to approach the cosmology constraints attainable if the bias were known exactly).[abridged]