In this paper, some aspects of the expected statistical properties of the noise for the future PLANCK satellite, both on data streams and on output maps, have been analysed. It has been shown that for the bolometer instrument of PLANCK, no significant striping on the maps is expected due to low-frequency noise only for the nominal knee frequency of the noise of 0.01 Hz, as a simple destriping method can remove low-frequency noise effects on maps. For the PLANCK HFI the increase in the noise rms on destriped maps is expected to be less than half a percent. Even if the low-frequency noise were much worse than expected, correction for its effects are possible with the scanning strategy of PLANCK, contrarily to what has been suggested by Wright (Wright 1996).
However, we believe that testing the extraction of cosmological information using simulated noise maps is probably one of the best way to quantify the exact consequences of instrumental effects such as striping, and continuing efforts should be made towards simulating realistic noise maps in the future.
An optimised scanning strategy must be devised by including simulations of most foreseen systematics. We believe that this optimum can be found within the orbit specifications for PLANCK, which minimise most of the potentially harmful systematic effects.
Valuable comments from G. Smoot and helpful discussions with F.-X. Désert, R. Gispert, J.-M. Lamarre and J.-L. Puget are gratefully acknowledged. This work was supported by a PhD research grant from CNES, the French national space agency.