The PAU Survey: Dissecting strong-lensing systems with narrow-band photometric redshifts

Wide-field imaging surveys such as Euclid will identify large samples of strong-lens candidates, whose scientific use requires redshifts for both deflectors and lensed sources. Source redshifts are challenging because faint arcs are blended with foreground light. We test lens-model-informed photometry on eight known strong lenses, using broad-band models to define component regions, point-reflection subtraction to reduce contamination, and PAU Survey narrow bands combined with five HSC or CFHT broad bands. BCNz estimates separate lens and source redshift distributions without imposing their expected ordering. All eight lens redshift peaks agree with spectroscopy within |δz|≤0.014, where δz=(zb-zspec)/(1+zspec). Four source peaks, all at zₛˢᵖᵉᶜ≤1.23, agree within |δz|≤0.025. Symmetric subtraction reduces the median source |δz| from 0.336 to 0.071. At |δz|<0.05, narrow- plus broad-band photometry yields eight lens and four source matches, versus three and two using EAZY with broad bands alone. These results demonstrate component redshift recovery in blended systems and support redshift-based lens-candidate assessment alongside morphology and colour. Larger narrow-band surveys, including J-PAS, could extend this approach, pending uncertainty calibration and blind candidate tests.