The COSMOS-Web Lens Survey (COWLS) III: forecasts versus data

TL;DR

The paper adapts the lenspop forecasting framework to JWST COSMOS-Web by integrating the JAGUAR source catalog and COSMOS-Web observing specs, yielding a forecast of 107 detectable strong lenses and highlighting the potential to probe the epoch of reionisation via sources up to $z_{\rm s}\sim11$. It then compares these forecasts to the first COSMOS-Web lens catalogue (COWLS), finding good agreement in Einstein radii distributions for the larger high+medium confidence sample but notable discrepancies in lens redshifts and magnitudes, likely driven by selection effects and the use of a SDSS-based fundamental plane to predict lens magnitudes. The analysis demonstrates the utility and limitations of forward modelling for JWST-era lens surveys, identifies biases in both forecast and observed samples, and underscores the need for spectroscopic follow-up and potentially expanded high-redshift lens searches, including MIRI data. Overall, the work validates lenspop as a viable tool for JWST lens forecasts and highlights how JWST enables new lens discoveries at high redshift and with small Einstein radii, while pointing to refinements needed to reconcile magnitude and redshift distributions with observations. The findings have implications for planning future lens searches and for interpreting lens statistics in the context of galaxy evolution and cosmology."

Abstract

We compare forecasts for the abundance and properties of strong gravitational lenses in the COSMOS-Web survey, a $0.54$ deg$^2$ survey of the COSMOS field using the NIRCam and MIRI instruments aboard JWST, with the first catalogue of strong lens candidates identified in the observed NIRCam data, COWLS. We modify the lenspop package to produce a forecast for strong lensing in COSMOS-Web. We add a new mock galaxy catalogue to use as the source population, as well as the COSMOS-Web survey specifications, including the transmission data for the four NIRCam filters used. We forecast 107 strong lenses can be detected in COSMOS-Web across all bands, assuming complete subtraction of the lens galaxy light. The majority of the lenses are forecast to have small Einstein radii ($θ_{\rm E} < 1$ arcsecond) and lie at redshifts between $0 < z <2$, whilst the source redshift distribution peaks at $z\sim 3$ and has a long tail extending up to $z \sim 11$, unambiguously showing that strong lensing in JWST can probe the entirety of the epoch of reionisation. We compare our forecast with the distributions of Einstein radii, lens photometric redshifts, and lens and source magnitudes in the observed lenses, finding that whilst the forecast and observed Einstein radii distributions match, the redshifts and magnitudes do not. The observed lens redshift distribution peaks at a slightly lower redshift than the forecast one, whilst the lens magnitudes are systematically brighter in the observed data than in the forecast.

Figures (11)

• Figure 1: Multiband images of four simulated COSMOS-Web-like lenses using properties of the forecast population. From left to right, there is a near-complete Einstein ring, an extended arc, a single image (the counter-image being obscured by the lens light) and a distinctive blue lensed source. The colours are representative and are produced by combining the data simulated in three different NIRCam bands (F150W, F277W and F444W).
• Figure 2: The distribution of Einstein radii (top left), velocity dispersions (top right), lens redshifts (bottom left) and source redshifts (bottom right) for our forecast lenses. Each histogram is weighted so the total count sums to 107, the number of lenses we forecast for COSMOS-Web.
• Figure 3: Distribution of the source redshifts in our COSMOS-Web-like forecast systems (purple) and the redshifts of galaxies in the JAGUAR mock catalogue from which the source population was drawn (black).
• Figure 4: Distribution of the forecast unlensed source magnitude in the F277W band (red) and the unlensed source magnitude in the JAGUAR catalogue from which the forecast population was drawn (black).
• Figure 5: Our forecast distribution of Einstein radii (yellow) compared to the observed lenses (black). Left: the 17 spectacular lenses. Middle: the 59 high confidence lenses. Right: the 116 high and medium confidence lenses.