A spatially resolved study of ionized regions in galaxies at different ...

3.3. Data Reduction 61
strong absorption is clearly seen particularly at 9300 Å with corrections around a factor
of 2. Taking into account that the theoretical ratio between the sulphur lines is I(9532Å)
≃ 2.44 · I(9069Å), it is immediately obvious that the intensity of the [Siii] 9532Å line is
heavily cut up by atmospheric absorptions bands, since both sulphur lines are of the same
intensity before correction. After applying the correction factor both lines are pushed to
higher intensities. Nevertheless, the total intensity of the 9532 Å line is not recovered, since
the ratio between them is still well below the theoretical one. This ratio is not accomplished
at any fiber, meaning that the absorption in the range from 9300 Å is stronger. Thus, we must
rely on the 9069Å line when deriving temperatures or using empirical calibrators, applying
the theoretical ratio to obtain the intensity of the 9532 Å line. Taking into account the low
correction factors obtained near the location of the 9069 Å line, we can assume that the
intensity recovered is good to an estimated accuracy of about 7 per cent.
3.3.10 Correction for differential atmospheric refraction
The refraction of light by the Earth’s atmosphere is wavelength-dependent. This effect
is called differential atmospheric refraction (DAR), and results in the image of an object to
appear at different positions in the telescope focal plane depending on the wavelength of
observation. The amplitude of the shift due to DAR is a function of zenith distance, causing
that the red and blue ends of an object spectrum will appear at different spatial positions.
The effect is important when fiber-to-fiber intensities of different wavelengths are combined,
but negligible if spatially integrated intensities are used.
IFUs have the advantage over traditional long-slit spectroscopy of determining and correcting
for DAR (Arribas et al., 1999), once the data is totally reduced and converted to
datacube format. Any DAR correction requires to resample the data spatially, which produces
a loss of the initial spatial configuration of the spaxels, changing from one spectrum
(wavelength versus flux) per spaxel (spatial coordinate) to a 2D flux image at each wavelength
point.
There are several ways of correcting for DAR. One approach is to apply a theoretical
correction based on the Filippenko (1982) formulae, in which an image offset vector for each
wavelength bin is used 12 . Each image, at every wavelength, is thereafter shifted with respect
to a pre-defined reference wavelength, using a (fractional) bi-linear interpolation procedure.
Doing this the intensity in each spaxel becomes a function of four spaxels - introducing an
unavoidable smoothing. Nevertheless, in many cases the empirical correction works better
than the theoretical one. The reason is that theory assumes that the layers of the atmosphere
with equal refraction index are flat and perpendicular to the zenith of the telescope. However,
it is known that this is only a first order approximation, since these layers depend strongly
12 Parallactic angle (orientation of the IFU with respect to the downward direction), and physical conditions
at the time of the observations like humidity, pressure, and temperature are needed.