The population of galaxy-galaxy strong lenses in forthcoming optical imaging surveys

TL;DR

This paper forecasts the number of galaxy-galaxy strong lenses that upcoming optical surveys like DES, LSST, and Euclid could discover, based on simulated populations and realistic observational models.

Contribution

It introduces a new simulation framework for predicting strong lens discoveries in future surveys, validated against existing data and accounting for survey-specific detection efficiencies.

Findings

Forecasts 2400 lenses in DES, 120,000 in LSST, and 170,000 in Euclid.

Finds that doubling detection thresholds halves the number of discoverable lenses.

Uncertainties are mainly due to high-redshift source populations, with tens of percent errors.

Abstract

Ongoing and future imaging surveys represent significant improvements in depth, area and seeing compared to current data-sets. These improvements offer the opportunity to discover up to three orders of magnitude more galaxy-galaxy strong lenses than are currently known. In this work we forecast the number of lenses discoverable in forthcoming surveys and simulate their properties. We generate a population of statistically realistic strong lenses and simulate observations of this population for the Dark Energy Survey (DES), Large Synoptic Survey Telescope (LSST) and Euclid surveys. We verify our model against the galaxy-scale lens search of the Canada-France-Hawaii Telescope Legacy Survey (CFHTLS), predicting 250 discoverable lenses compared to 220 found by Gavazzi et al (2014). The predicted Einstein radius distribution is also remarkably similar to that found by Sonnenfeld et al (2013). For future surveys we find that, assuming Poisson limited lens galaxy subtraction, searches in DES, LSST and Euclid datasets should discover 2400, 120000, and 170000 galaxy-galaxy strong lenses respectively. Finders using blue minus red (g-i) difference imaging for lens subtraction can discover 1300 and 62000 lenses in DES and LSST. The uncertainties on the model are dominated by the high redshift source population which typically gives fractional errors on the discoverable lens number at the tens of percent level. We find that doubling the signal-to-noise ratio required for a lens to be detectable, approximately halves the number of detectable lenses in each survey, indicating the importance of understanding the selection function and sensitivity of future lens finders in interpreting strong lens statistics. We make our population forecasting and simulated observation codes publicly available so that the selection function of strong lens finders can easily be calibrated.