A statistical study of weak lensing by triaxial dark matter haloes: consequences for parameter estimation

Measuring the distribution of mass on galaxy cluster scales is a crucial test of the cold dark matter (CDM) model, providing constraints on the nature of dark matter. Recent work investigating mass distributions of individual galaxy clusters using gravitational lensing has illuminated potential inconsistencies between the predictions of structure formation models relating halo mass to concentration and those relationships as measured in massive clusters. However, such analyses typically employ only simple spherical halo models with canonical Navarro, Frenk & White (NFW) slopes, while the haloes formed in simulations show a range of more complex features. Here we investigate the impact of such expected deviations from the canonical NFW halo profile on mass and parameter estimation using weak gravitational lensing on massive cluster scales. Changes from the canonical NFW profile slopes are found to affect parameter estimation. However, the most important deviation is halo triaxiality because it is impossible even with fiducial weak lensing data to fully resolve the 3D structure of the halo due to lensing's sensitivity only to projected mass. Significant elongation of the halo along the line of sight can cause the mass and concentration to be overestimated by as much as 50 per cent and by a factor of 2, respectively, while foreshortening has the opposite effect. Additionally, triaxial haloes in certain orientations are much better lenses than their spherical counterparts of the same mass, indicating that clusters chosen for study because of evident lensing are likely to be drawn from the high-triaxiality end of the halo shape distribution; cluster samples chosen with no shear bias return correct average parameter values. While the effects of triaxiality alone may not be enough to fully explain the very high concentrations reported for some clusters, such as Abell 1689, they go a long way in easing the tensions between observations and the predictions of the CDM paradigm.