2007A&A...461...81O

We apply a matched-filter cluster detection algorithm to the Canada-France-Hawaii Telescope Legacy Survey (CFHTLS) i-band data for the Deep-1, Deep-2, Deep-3 and Deep-4 fields covering a total of 4 square degrees. To test the implemented procedure we carry out simulations for assessing the frequency of noise peaks as well as estimate the recovery efficiency. We estimate that up to z∼0.7 the catalogue is essentially complete for clusters of richness class R>1. The recovered redshifts are in general overestimated by Δz=0.1 with a scatter of σ_Δz∼0.1, except at redshifts z>1 where the estimated redshifts are systematically underestimated. The constructed cluster candidate catalogue contains 162 detections over an effective area of 3.112 square degrees corresponding to a density of ∼52.1 per square degree. The median estimated redshift of the candidates is z=0.6. The estimated noise frequency is 16.9±5.4 detections per square degree. From visual inspection we identify systems that show a clear concentration of galaxies with similar colour. These systems have a density of ∼20 per square degree.