Normal and streamwise components of the velocity fields of turbulent flow in a channel at low Reynolds numbers have been measured with laser-Doppler techniques. The experiments duplicate the conditions used in current direct numerical simulations of channel flow, and good, but not exact, agreement is found for singlepoint moments through fourth order. In order to eliminate LDV velocity bias and to measure velocity spectra, the mean time interval between LDV signals was adjusted to be much smaller than the smallest turbulence time scale. Spectra of the streamwise and normal components of velocity at locations spanning the channel are presented. I IntroductionThis paper presents measurements, made with a two-component laser-Doppler velocimeter, of the streamwise and normal velocity fields for turbulent flow in a rectangular channel. They were obtained at low velocites, so that the Reynolds number, based on the half channel height and the bulk velocity, was 2,777 or 2,457, and the dimensionless half width of the channel was H + = 178.6 or H + = 158.5, respectively. The motivation for this work was to examine the accuracy of the direct numerical simulation of three dimensional, time dependent turbulent flow in a channel at Re~ = 2,800 by Kim et al. (1987) and to obtain new measurements of the power spectral densities of both streamwise and normal turbulent velocity components.The study reported in this paper is part of a Ph.D. thesis by M. Niederschulte (1989). Measurements were made in fully developed flow 200 channel heights downstream from the entrance of a 5.08 x 61 cm rectangular channel by using a two-component laser Doppler velocimeter. Certain aspects of the LDV and the experimental procedure were tailored to the demands of this flow and make this experiment unique. Firstly, the optics were designed to permit accurate measurements at distances as close to the wall as y+ =0.6 (0.16 ram) with spatial resolution better than 0.035 mm in the direction normal to the wall. This was accomplished by using a threebeam LDV of the type described by Adrian (1975), combined with 6.27 : 1 expansion of the laser beams, as described in Buckles et al. (1984). Secondly, the fluid and the scattering medium in the fluid were also managed carefully to avoid the uncertainties associated with LDV velocity bias, and to produce data whose rate was large enough to permit measurement of the power spectra of the relatively low turbulence intensity flow. Carefully cleaned water was seeded with 808 nm polystyrene particles having a specific gravity of 1.05 and standard distribution of 0.5 %. The concentration of this very uniform scattering medium was chosen, in accord with the discussion in Adrian (1983), to be large enough to yield time intervals between particle arrivals that were smaller than the smallest turbulence time scale, but not so small as to create ambiguity (or "phase") noise. On average, the mean separation between particles in the streamwise direction was less than 0.2 ram. The resulting signal was a nearly continuous repre...