Observational biases in flux magnification measurements

Abstract: Flux magnification is an interesting complement to shear-based lensing measurements, especially at high redshift where sources are harder to resolve. One measures either changes in the source density (magnification bias) or in the shape of the flux distribution (e.g. magnitude-shift). The interpretation of these measurements relies on theoretical estimates of how the observables change under magnification. Here we present simulations to create multi-band photometric mock catalogues of Lyman-break galaxies in a CFHTLenS-like survey that include several observational effects that can change these relations, making simple theoretical estimates unusable. In particular, we show how the magnification bias can be affected by photometric noise, colour selection, and dust extinction. We find that a simple measurement of the slope of the number-counts is not sufficient for the precise interpretation of virtually all observations of magnification bias. We also explore how sensitive the shift in the mean magnitude of a source sample in different photometric bands is to magnification including the same observational effects. Again we find significant deviations from simple analytical estimates. We also discover a wavelength-dependence of the magnitude-shift effect when applied to a colour-selected noisy source sample. Such an effect can mimic the reddening by dust in the lens. It has to be disentangled from the dust extinction before the magnitude-shift/colour-excess can be used to measure the distribution of either dark matter or extragalactic dust. Using simulations like the ones presented here these observational effects can be studied and eventually removed from observations making precise measurements of flux magnification possible.