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

64 3 • IFS of a GEHR in NGC 6946
simple stellar populations (SSPs, or instantaneous burst) synthetic spectra to the continuum
of the object and subtract the underlying population, giving a final spectra which, in theory,
should only contain the gas emission spectra. Each spectrum from the IFU sample is fitted
with synthetic SSP models. As explained in Sánchez et al. (2007b), models were created
using the GISSEL code (Bruzual and Charlot, 2003), assuming a Salpeter IMF, for different
ages and metallicities. FIT3D has 72 models covering a discrete grid of 12 ages (5, 25, 100,
290, 640 Myr, 0.9, 1.4, 2.5, 5, 11, 13 and 17 Gyr), and six metallicities (Z = 0.0001, 0.0004,
0.004, 0.008, 0.02 and 0.05). In order to reduce the number of free parameters, we only used
a template of 3 metallicities, namely 0.02, 0.008 and 0.004, with the whole set of ages. The
set of metallicities were chosen to have an interval which contained the published metallicity
of the region. Each spectrum then was fitted to each of the 36 models. FIT3D re-samples the
model to the resolution of the data, convolves it with a certain velocity dispersion, and scales
it to match the data set by a χ 2 minimization scheme. In order to fit the continuum, it is
necessary to create a mask file giving wavelength intervals of the spectra containing emission
lines, features (Wolf-Rayet bumps), and strong residuals from the sky subtraction and other
artifacts. We only performed the fitting on the blue mosaic, i.e., only the spectral region at
wavelengths bluer than 7000 Å was used. At redder wavelengths the strong nightsky emission
lines and the telluric absorptions have strong residuals, so they were masked. Nevertheless,
the absorption of the underlying population to the emission lines in the red spectral region
is negligible.
To illustrate the fit results, we show in the upper panels of Figure 3.13 the spectrum of the
spaxel 635 (green line) with the best fit found by FIT3D (blue line) in the ranges 4200-5100
Å and 6150-6850 Å. The measured intensities of the most prominent Balmer emission lines
of the subtracted spectrum (free of underlying population) result within the errors as compared
with some test measurements using splot from IRAF, adopting a pseudo-continuum
as explained in Hägele et al. (2006). In both upper panels of Figure 3.13 we can appreciate
that small errors in the fit of the underlying stellar population could become a great
unquantifiable error in the emission line fluxes of the fainter emission lines. Nevertheless,
for the strongest emission lines, the differences between the measurements done after the
subtraction of the FIT3D fit and those using the pseudo-continuum approximation are well
below the observational errors, giving almost the same result. It must be highlighted that in
this type of objects the absorption lines, used to fit the underlying stellar populations, are
mostly affected by the presence of emission lines. There are a few that are not affected, such
as some calcium and magnesium absorption lines. To make the FIT3D fit it is necessary
to mask the contaminated lines, thus the fit is based on the continuum shape and takes
into account only a few absorption lines. Hence, it is not surprising that the FIT3D results
are not so good for this kind of objects. This result can be extended to other fitting codes
such us STARLIGHT. It must be noted too that this methodology is very advantageous for
statistical and comparative studies and when dealing with a large number of spectra, as in