The Perot-Fabry reducing software package ADHOC, developed at Marseille's Observatory (Boulesteix 1993) was used.
In the first step individual channel images were added in order to build a
data cube and observation cubes were the reduced into a wavelength
calibrated cube (x, y, 
). In the second step, the analyze of the
detailed profile of the observed line within the free spectral range of the
interferometer was done for each pixel. Then night sky and continuum
subtractions are computed and velocity, monochromatic and continuum images
were extracted.
First step is to correct each image of the data cube from dark and flat field. It was done using images obtained observing the dome. After the flat field and dark corrections it is necessary to remove cosmic rays and correct the cube from transparency fluctuations and seeing variations. We had considerable changes in air mass during our exposure times often longer than 2 hours. This correction was made using field stars fluxes. To minimize seeing variations during one observation, we have smoothed each scan with a gaussian function with a FWHM equal to the worse seeing that we have in the data cube in order to have the same seeing in each channel. Then this clean cube is proceeded in the same manner as for IPCS observations.
The phase map, obtained from the calibration cubes, is used to calibrate in
wavelength observation cubes. The accuracy of the zero wavelength is a
fraction of a channel width (
) over the whole field.
In the red spectra (H
, [NII] 6584 Å), it is necessary to
subtract the night sky OH lines passing through the narrow filter which are
often of the same intensity as the gas emission lines we observe. This is
done by determining the OH profile from areas of the field sufficiently far
from the center of the galaxy to be supposed free from any galactic emission
(Laval et al. 1987).
Even if the width of the filter is large compared to the free spectral range a modulation can appear on observed spectra, due mainly to the strong continuum of elliptical galaxies. Observation of a continuum lamp was used to correct from this effect.
From the individual spectra calibrated in wavelength the continuum map was
easily derived and represents for each pixel the value of the mean of the
three faintest channels, in order to avoid channel noise effect. To obtain
monochromatic images is more delicate because, in the case of multiple
components profiles, different maps can be produced for each component. In
this case, it is first necessary to decompose multiple profiles in order to
build cubes corresponding to each component separately. A gaussian algorithm
fitting is used to decompose complex profiles.
Then monochromatic maps are obtained by integrating the monochromatic profile of each component. Velocity maps are computed from the position of the emission line. For interferometer #1, some uncertainty remains, due to the size of the smallest free spectral range and the possible confusion of the orders. Ambiguity is resolved by using other observations (interferometer #2, previous long slit or HI observations).
Monochromatic maps have a one pixel resolution in the center of the
galaxies. Spectral profiles have been binned in outer parts (to 
pixels)
in order to get a sufficient signal to noise ratio to measure and decompose
components.