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

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

cab.inta.csic.es
from cab.inta.csic.es More from this publisher
02.01.2015 Views

56 3 • IFS of a GEHR in NGC 6946 4.0 1e−13 ◦A−1 ) s−2 Flux (erg cm−2 3.5 3.0 2.5 2.0 Counts103 7500 8000 8500 9000 9500 10000 Wavelength (◦A) Figure 3.10: Upper panel: Spectrophotometric standard star BD+17 ◦ 4708. Bottom panel: Solid (blue) line represents a extrated 1D spectrum of the star BD+17 ◦ 4708 and dashed (green) line a smooth fit to the continuum. The atmospheric absorption features in the spectrum of the standard star are clearly seen along all the wavelength coverage. Notice the strong absorptions in the 8900 - 9700 Å range. The scale is logarithmic in order to enhance the mean features. absolute flux calibration in IFUs with filling factors less than 1. Assuming that the pointspread function (PSF) of the star can be represented as a 2-dimensional Gaussian, it can be divided in increasing rings apertures. The maximum is in the center, while the external ring would have a minimum flux. Therefore, most of the flux would be concentrated in the first rings. In the case of the one-pointing frame, the gap between the first ring (the first fiber), and the second one (a ring of 6 fibers, as it can be seen in left upper panel of Figure 3.8), can hold a important fraction of the flux of the star. This effect is clearly seen in the jump of the flux ratio curve derived for a one fiber extraction to the 7 fibers one. Next rings would have less weight in the total flux, as seen in the convergence of curves from 7 up to 37 fibers, taking into account each case independently. It is important to have a smooth final flux ratio curve. In general, this is accomplished by convolving the spectra with a Gaussian kernel to match solutions between the measured spectrum of the spectrophotometric standard star and its absolute value as obtained from tables, so that the final flux ratio (i.e. sensitivity function) is a continuous, smooth curve. Sometimes a smooth spline function of gaussian kernel is not enough, so another approach

3.3. Data Reduction 57 is required. Before comparing the 1D count-rate spectrum with the flux table calibration it is advisable to fit a continuum curve to the spectra. This is specially important in the red part of the spectra (8000-11000 Å range), where absorption features from the atmosphere can depress some parts of the spectra, introducing spurious features in the continuum of the standard star, and therefore in the final flux ratio curve. Figure 3.10 shows an example of the absorption effects in the spectrum of the standard star BD+17 ◦ 4708. The upper panel shows the absolute flux distribution taken from the standard star table. Notice the almost absence of features in the continuum. The lower panel shows in solid (blue) line the 1D spectrum (in counts) extracted following the steps described above and in dashed (green) a fitted continuum. It can be clearly seen the strong atmospheric water-vapour absorptions bands along the spectrum, specially in the range where the sulphur emission lines [Siii] λλ 9069,9532 Å lines are located. The fitted continuum is then used to obtain the flux ratio between the counts and the flux calibrated table. It is advisable to have several measurements of different standard stars along the night, or several exposures of the same star at different airmasses in order to have a representative sensitivity function. Then, an average sensitivity function is computed from those with the same instrumental setup. If the conditions of the atmosphere changed significantly from one star to another, a gray shift over the the sensitivity function can be performed. This is done by shifting the data so that the mean sensitivity of each star is the same as the star with the greatest mean sensitivity. This compensates for variable grey extinction due to clouds. Sometimes it is necessary to delete one star if its slope/shape is very different from others. Unfortunately, our set of data had only one observation of a standard star per night, so we had to derive a flux ratio for each wavelength range only from one curve. Despite this fact, after flux calibration, in the overlapping region of the spectra taken for both setups, the agreement in the average continuum level in a fiber-to-fiber comparison was about 5 per cent. 3.3.7 Sky-subtraction The emission from the Earth’s atmosphere contributes to the detected signal. Sky emission lines can be easily identified in the RSS data as vertical lines. Many IFUs provide special fibers placed with an offset from the observed object to obtain a clear sky spectrum. When these fibers are not available or the target is surrounded by contaminating emission from adjacent objects, the usual method consists on taking an additional exposure of the nearby sky after or before the exposure of the target itself, bearing in mind that there should not be any significant contaminating emission from other sources. If the sky is derived form an external sky-frame, the sky subtraction should be performed after the flux calibration. If it is derived from the same dataset, the substraction can be done before. If the target does not fill all the FOV of the IFU, E3D can be used in a similar way as for the sky subtraction of

56 3 • IFS <strong>of</strong> a GEHR <strong>in</strong> NGC 6946<br />

4.0<br />

1e−13<br />

◦A−1<br />

)<br />

s−2<br />

Flux (erg cm−2<br />

3.5<br />

3.0<br />

2.5<br />

2.0<br />

Counts103<br />

7500 8000 8500 9000 9500 10000<br />

Wavelength (◦A)<br />

Figure 3.10: Upper panel: Spectrophotometric standard star BD+17 ◦ 4708. Bottom panel: Solid<br />

(blue) l<strong>in</strong>e represents a extr<strong>at</strong>ed 1D spectrum <strong>of</strong> the star BD+17 ◦ 4708 and dashed (green) l<strong>in</strong>e a<br />

smooth fit to the cont<strong>in</strong>uum. The <strong>at</strong>mospheric absorption fe<strong>at</strong>ures <strong>in</strong> the spectrum <strong>of</strong> the standard<br />

star are clearly seen along all the wavelength coverage. Notice the strong absorptions <strong>in</strong> the 8900 -<br />

9700 Å range. The scale is logarithmic <strong>in</strong> order to enhance the mean fe<strong>at</strong>ures.<br />

absolute flux calibr<strong>at</strong>ion <strong>in</strong> IFUs with fill<strong>in</strong>g factors less than 1. Assum<strong>in</strong>g th<strong>at</strong> the po<strong>in</strong>tspread<br />

function (PSF) <strong>of</strong> the star can be represented as a 2-dimensional Gaussian, it can be<br />

divided <strong>in</strong> <strong>in</strong>creas<strong>in</strong>g r<strong>in</strong>gs apertures. The maximum is <strong>in</strong> the center, while the external r<strong>in</strong>g<br />

would have a m<strong>in</strong>imum flux. Therefore, most <strong>of</strong> the flux would be concentr<strong>at</strong>ed <strong>in</strong> the first<br />

r<strong>in</strong>gs. In the case <strong>of</strong> the one-po<strong>in</strong>t<strong>in</strong>g frame, the gap between the first r<strong>in</strong>g (the first fiber),<br />

and the second one (a r<strong>in</strong>g <strong>of</strong> 6 fibers, as it can be seen <strong>in</strong> left upper panel <strong>of</strong> Figure 3.8),<br />

can hold a important fraction <strong>of</strong> the flux <strong>of</strong> the star. This effect is clearly seen <strong>in</strong> the jump<br />

<strong>of</strong> the flux r<strong>at</strong>io curve derived for a one fiber extraction to the 7 fibers one. Next r<strong>in</strong>gs would<br />

have less weight <strong>in</strong> the total flux, as seen <strong>in</strong> the convergence <strong>of</strong> curves from 7 up to 37 fibers,<br />

tak<strong>in</strong>g <strong>in</strong>to account each case <strong>in</strong>dependently.<br />

It is important to have a smooth f<strong>in</strong>al flux r<strong>at</strong>io curve. In general, this is accomplished<br />

by convolv<strong>in</strong>g the spectra with a Gaussian kernel to m<strong>at</strong>ch solutions between the measured<br />

spectrum <strong>of</strong> the spectrophotometric standard star and its absolute value as obta<strong>in</strong>ed from<br />

tables, so th<strong>at</strong> the f<strong>in</strong>al flux r<strong>at</strong>io (i.e. sensitivity function) is a cont<strong>in</strong>uous, smooth curve.<br />

Sometimes a smooth spl<strong>in</strong>e function <strong>of</strong> gaussian kernel is not enough, so another approach

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!