In the light of an increased understanding of the properties of ISOCAM and of the underlying galaxy population, some alterations were made to the observing strategy used for the HDF-S with respect to that used for the HDF-N.
Eight rasters were taken, one with the LW2 and one with the LW3 filter at
each of four positions, with raster centres offsetted by a fraction of
the pixel field of view to improve spatial resolution of the final map
obtained by their coaddition. The pixel field of view at 6.7
m was
changed from 3 arcsec to 6 arcsec, matching that used at 15
m, since
it had been realized that HDF-N images at 6.7
m were not confusion
limited, as previosuly expected, and that the improvement in
ratio
and areal coverage obtained by moving to larger pixels was expected to
compensate the loss in resolution. The other significant change was to
increase the raster step size, which had two effects. First, the step size
is now larger than the PSF so that consecutive pointings have no longer
correlated signal, making the removal of time-correlated noise much easier.
Second, the larger pixel area meant that the full survey area could be
covered by each individual
raster, with the complete survey
being made up by stacking four independent rasters in each passband and
thus increasing the redundancy.
Figure 2.3 shows the final ISO maps at 6.7 and 15
m.
A number of 35 ISO sources are detected, of which 32 have associations
with optical objects. Eight of them are stars, one is definitely an AGN,
a second seems likely to be an AGN while the remaining 22 appear to be
normal or starburst galaxies forming stars at rates of
yr
, with a median value of
yr
. However,
the small volume of the survey implies that estimates of the star formation
rate density may suffer from a large sampling variance and that the results
therefore do not place robust constraints on the global SFR.