Effect of the choice of FDR threshold on the catastrophic failure rate after cleaning (upper panel), the retention (middle panel) and the capture rate (lower panel) on catalogues with a fixed S/N of 1.0 and 2.0 with flat noise; and on a mixed catalogue with a uniform distribution in S/N between 1 and 20, with pixel-dependent noise. Note the greater sacrifices required both in retention and capture rate in order to obtain the same catastrophic failure rate at an S/N of 1.0 compared to 2.0. Note also that we are able to obtain a 5.1% failure rate in our redshift estimates for the cleaned catalogue, a retention of 52.6%, and a capture rate of 76.2% with the catalogue at an S/N of 2 at an FDR threshold of 4.55%.