Parasitic Light in NGST instruments: the accuracy of photometric redshifts and the effect of filter leaks in the visible and near IR camera

A detailed analysis of NGST multiband photometry applied to the reference case of the study of high-redshift galaxies has been carried out with simulations based on galaxy SEDs derived from the currently available empirical and model templates and on plausible standard filter-sets. In order to correctly identify star forming galaxies in the redshift range 5 2 and avoid confusion with other SEDs it is mandatory to have photometric information in optical bands, besides a standard IR filter-set like F110 F160 K L M. In particular by adding the V606, I814 and z-Gunn filters a good discrimination is obtained above z>5 for star forming galaxies and z>1 for early-types. The case for an extension of the NGST wavelength domain to the optical range is therefore strongly supported by this analysis. The effects of leaks in the filter blocking have also been investigated. In spite of rather pessimistic assumptions (a constant leak at a level of 10^{-4} of the peak transmission over the whole spectral range or a leak of Gaussian shape placed at 1.5 times the effective wavelength of the filter with an amplitude of 10^{-3} of the peak transmission and a width of 5% of the central wavelength) the effects are not dramatic: the accuracy of the determination of photometric redshifts with a standard V606 I814 z-Gunn F110 F160 K L M filter set does not significantly deteriorate for a sample limited to M_AB~30. In the range 5.5<z<12.5 a rather high accuracy with a typical error |z_sim-z_phot|<=0.2 is achieved. With the filter blocking achieved by the present standard technology, extended and continuous spectral coverage appears to be the driving factor to maximize the scientific output.