A 1/6° model simulation of the Atlantic ocean forced with daily fluxes from ECMWF (re-analysis 1979–1993 and analysis 1994–1999) has been carried out within the Clipper project. A storage strategy which filters out inertial oscillations is defined: five-day mean fields are continuously stored at five-day intervals. It is shown that aliasing errors on the monthly mean meridional heat transport (MHT, a second-order moment) are negligible in that case. These errors are of the order of 0.8 PW in the tropics in the case of a sampling strategy based on instantaneous fields stored every five days, even in the case where step-like variations in the forcing are avoided by an interpolation of the daily wind stress to the model time step. It is also shown that aliasing errors on the annual mean MHT can be as large as 0.2 PW in the tropics in the case of sub-sampling with instantaneous fields.