How Well Can Weak Lensing Measure the Mass of Galaxy Clusters?

The technique of weak-lensing aperture mass densitometry, so called the zeta-statistic, has recently been popular in actual observations for measurement of individual cluster mass. It has however been anticipated that the line-of-sight projection by foreground and background matter can adversely affect the cluster mass determination with not only substantial error dispersion but also a sizable positive systematic bias. Additionally, the finite number of background galaxies even at a reasonable observing depth can also introduce Poisson noise to the mass estimate. In this paper, we quantitatively investigate the degree of errors separately contributed by the two sources to the mass determination of those galaxy clusters with M_{200}>10^{14}M_{\odot}. We find that the aperture mass of zeta-statistic turns out to be a mass estimator of much reduced systematic bias, due to the cancellation by the positively biased local background mass sheet. However, the error dispersion of M_{200} arising from both projection effect and Poisson noise is found to be still sizable (40%-90%), even for the shear-selected, clean sample where multiple clusters located within a suitable projected aperture are removed. We also investigate how to remedy this large-error problem in weak lensing measurements, and propose a plausible alternative mass estimator, M(<theta_{1000}), an aperture mass measured within about half the virial radius. The aperture mass M(<theta_{1000}) is free of bias and has a substantially reduced error dispersion, 39% for the worst case of high-z, low-mass clusters, that can be smaller than the error dispersion of M_{200} as much as a factor 3.