The J-PAS survey: The effect of photometric redshift errors on cosmic voids

We investigated the impact of photometric redshift errors in the ongoing Javalambre Physics of the Accelerating Universe Astrophysical Survey (J-PAS) on void identification and properties using a watershed-based method, aiming to assess the recovery of individual voids and the overall void environment. We created galaxy mock catalogues for redshift z = 0.1 using the IllustrisTNG300-1 simulation, defining two datasets: an $ideal$ sample ($m_r < 21$ mag) and a $perturbed$ sample with the Z-coordinate errors mimicking J-PAS's line-of-sight errors, derived from the precursor miniJPAS survey data. We identified voids using ZOBOV, a watershed algorithm. We found 1065 voids in the $ideal$ sample and 2558 voids in the $perturbed$ sample. The $perturbed$ sample voids have, on average, smaller sizes and denser interiors. We filtered out voids based on density and radius in order to eliminate overdense and small spurious instances. The stacked density profile of filtered voids in the $perturbed$ sample remains close to the average density even at the boundary peak, indicating a strong blurring of structures by the redshift errors. The number of $ideal$ sample voids for which at least $50\%$ of the volume is recovered by a void in the $perturbed$ sample is 53 (29 for the filtered sample). The volume occupied by these voids is less than $10\%$ of the simulation volume. Merging voids in the $perturbed$ sample marginally improves the recovery. The overall volumes defined as voids in the two samples have an overlap of $80\%$, making up $61\%$ of the simulation box volume. While some statistical properties of voids might be recovered sufficiently well, the watershed algorithms may not be optimal for recovering the large-scale structure voids if applied straight to photometric redshift survey data.