G Aviv

where kx is the telescope-plus-detector efficiency at A, Fc is the continuum flux in the line filter, Tl is the transmission of the line filter, /1 is the line flux, and Avc and Avl are the bandwidths of the continuum filter and the line filter, respectively.

Figure 8.6. An image of the spectrophotometric standard star BD1747. The star is marked with a circle.

Figure 8.6. An image of the spectrophotometric standard star BD1747. The star is marked with a circle.

The number of counts per second of the calibration star is

and that for the continuum filter is

where fx is the monochromatic flux of the spectrophotometric standard star (Figure 8.6). Then

Fc Fc

268 S. Pascual and B. C'edres

With this we are able to obtain /l.

Oke & Gunn (1983) give the following monochromatic magnitude for the calibration stars:

With a few manipulations,

Moreover,

F} Ff where FG and F^ are the counts per second at the object and at this calibration star, respectively, and are the air masses of the object and the calibration star, respectively, and K\ is the atmospheric extinction for wavelength A. After making a few substitutions, we obtain

10-0.4[mAB+48.6-Ki(xg-x\)] £o _ a10"0.4[mAB+48.6-Kc(xg-xC)] £o

where a is a factor that takes into account the differences between the continuum filter and the line filter. The method for the determination of this factor is explained in the following section.

8.2.4 Continuum subtraction To obtain the flux of the line alone, the continuum contribution needs to be wiped out. This is done by using an image with a filter centered near the target line but not affected by it, so as to sample only the underlying continuum. One method for obtaining this is to multiply the continuum image by a factor near the unit (factor a in Equation (8.11)). This factor can be determined in two ways: by making the emission of the field stars null after subtraction, or by assuming that the inter-arm emission is null after the subtraction in the case of galaxies. The first method could lead to an overestimation or underestimation of the real value of the factor a due to differences between the spectral distribution of the field stars and that of the object. The second method would wipe out the diffuse emission from the object (as seen in Figure 8.7). So, depending on the aim of the research, one method would be more suitable than the other. Nevertheless, special care has to be taken with both methods in order to make the correction. The fluxes for several field stars (more than three) must be obtained, using aperture photometry, in the first method. In the second method, statistics must be collected for small boxes in the inter-arm zones for the line image and for the same inter-arm zones for the continuum image.

8.3. Imaging in the airglow-windows range

Fringing is one of the issues that appear when observing in the broad bands R, I, and Z, and in the narrow bands. In this section we explain a method for removing fringing using IRAF tasks. We also explain a simple method for selecting emission-line galaxies

FIGURE 8.7. Ha images of NGC 4395 before (left panel) and after (right panel) continuum subtraction.

FIGURE 8.7. Ha images of NGC 4395 before (left panel) and after (right panel) continuum subtraction.

using the narrow-broad color excess. The selection process is studied in detail in Pascual et al. (2006).

8.3.1 Fringing

Fringing is produced by the multiple reflection and interference of light with the chargecoupled-device (CCD) substrate. When a monochromatic coherent light beam undergoes multiple reflection between two flat-parallel surfaces, the phase difference produces interference, either constructive or destructive, depending on the relationship between the wavelength and the distance between the two layers. If these are parallel, the resulting pattern is a series of bright and dark rings (Newton's rings). The effect arises when the wavelength is comparable to the thickness of the CCD. This implies wavelengths redward of —6000 A and up to the limit of the CCDs, around —10 000 A. In a real CCD, where the two layers are not parallel, the patterns are irregular.

When, instead of a monochromatic light, a wide range of wavelengths is used, constructive and destructive interferences tend to vanish locally. It is in narrow-band photometry, or when strong emission lines (nearly monochromatic) appear (such as those of the night-sky spectrum), that fringing is important. Note that a dome-flat image can have fringing if the illuminating lamp has emission lines or the filter is narrow. Figure 8.8 shows a sample I-band science image with a clearly visible fringe pattern. The effect is cumulative. This fringing must be removed by building a master fringe image. A method for doing this is explained below.

8.3.2 Fringing removal using the package mscred This section describes how to remove fringing using tasks within the IRAF package mscred (Valdes 1998). The basic steps for removing fringing are creating a template of the fringe pattern, creating a mask that isolates the background in each exposure, creating a background map when there are background gradients across the exposure, determining the scale factor that best matches the fringe template to the exposure, and subtracting the scaled fringe template from the exposure.

Figure 8.8. A science image containing the fringe pattern.

A fringe template is constructed from all the sky exposures which exhibit the same fringe pattern. The best result is obtained using dark-sky exposures where the fields have been dithered so that every pixel has several exposures that are uncontami-nated by sources. The exposures are put together to make a fringe template using the mscred.combine task. The images are scaled to a common level and objects are excluded by rejection and masking techniques.

Object masks to be excluded during combining can be created with task objmasks from package nproto. Since making object masks (and background maps) for each exposure is needed for automatic computation and removal of the fringing, making the masks for creating the fringe template does not add unnecessary processing. In any case, the sky map is needed only if there is a significant sky gradient. The mean sky level is automatically accounted for during the fringe-removal step.

The following are typical parameters of the task objmasks:

cc> nproto np> epar objmasks images = @type_object objmasks= obj_//@type_object

(omtype = numbers)

(convolv= block 3 3)

List of images or MEF files List of output object masks Object mask type List of input/output sky maps

Convolution kernel Sigma threshold above sky Sigma threshold below sky

The text file type_object is processed and new object masks and background maps are produced for each input image.

The template fringing is produced using the mscred.combine task. The critical parameter here is masktyp. The value of the parameter should be lobjmask. This instructs combine to use the header keyword OBJMASK that objmasks inserts into the images.

The object masks and sky maps produced by objmasks are used in the task rmfringe to automatically determine the fringe scaling and subtract the pattern. The most important parameters of the task are input = @filter_BB

fringe = FringeBB

masks = lobjmasks

List of input images List of output corrected images Fringe or list of fringe patterns List of object/bad data masks

(backgro= sky_//@filter_BB ) Lisk of input image backgrounds

The task does not always remove the fringing successfully. An alternative task that can be used to remove the fringing with good results is irmfringe. The basis of the task is to find the scale factor interactively by visual inspection.

This task is prepared to work with multi-extension fits files; in order to use it with single-extension fits, this syntax is used:

input = output = template=

r85132[0] List of input mosaic exposures r85132_f[0] List of output mosaic exposures FringeBB[0] Template mosaic exposure

8.4. Selection of candidates and line flux

Candidate line-emitting objects were selected using their excess narrow versus broad flux on a plot of tonb versus I — tonb .

Catalogs of objects present in the images can be constructed with any object-detection software. Here we use SExtractor (Bertin & Arnouts 1996). The common approach to object detection is to use the double-image mode : the narrow-band frame is used as a reference image for detection and then the flux is measured both in the narrow- and in the broad-band image. Other methods can be used to construct a segmentation image where the object detection is performed. The broad- and narrow-band images can be added or combined (Szalay et al. 1999) to allow the recovery of objects present in only one of the images.

Narrow-band imaging not only allows object detection, but also makes it possible to estimate the flux and equivalent width of the emission line with simple assumptions.

The flux density in each filter can be expressed as the sum of the line flux and the continuum flux density (the line is covered by both filters):

with f C the continuum flux, fL the line flux, AB and AN the broad- and narrow-bandfilter effective widths and fB and fN the flux densities in the two filters. Then the line flux, continuum flux, and equivalent width can be expressed as follows:

Was this article helpful?

0 0

Post a comment