PHOTOMETRIC REDSHIFT OUTLIERS AND THEIR IMPACT ON LSST WEAK LENSING COSMOLOGY: INSIGHTS FROM A GAUSSIAN ISLAND MODEL

Abstract

Next-generation cosmological surveys, such as the Vera C. Rubin Observatory's Legacy Survey of Space and Time (LSST), will deliver unprecedented statistical power to probe fundamental aspects of the Universe, including the nature of dark energy and dark matter. However, achieving this precision requires careful control of systematic uncertainties. One critical challenge is the presence of catastrophic photometric redshift (photo-z) outliers - galaxies for which the photometric redshift estimate deviates substantially from the true (spectroscopic) redshift. These outliers can skew the inferred redshift distributions used in weak lensing analyses, biasing constraints on cosmological parameters. We investigate how catastrophic photo-z outliers affect key cosmological parameters derived from LSST-like weak lensing data. Using a "Gaussian Island Model" to represent these outliers as distinct islands in redshift space, we find that even small fractions of certain outlier populations can significantly bias constraints on matter density (Ωm), time evolution of dark energy (w), and the amplitude of matter fluctuations (S8). Our analysis shows that "false low-z" outliers (high z, low z_ph) induce more pronounced biases than "false high-z" galaxies (low z, high z_ph), making accurate estimates of the false low-z outlier fraction especially critical. This sensitivity underscores the need for careful outlier characterization, improved photo-z calibration, and strategic spectroscopic follow-up. By better understanding and mitigating these catastrophic outliers, we can preserve the integrity of cosmological inferences and fully exploit the scientific potential of next-generation weak lensing surveys.