Äîêóìåíò âçÿò èç êýøà ïîèñêîâîé ìàøèíû. Àäðåñ îðèãèíàëüíîãî äîêóìåíòà : http://www.stsci.edu/hst/acs/documents/isrs/isr0103.pdf Äàòà èçìåíåíèÿ: Mon Jan 8 23:20:36 2007 Äàòà èíäåêñèðîâàíèÿ: Sat Dec 22 23:21:42 2007 Êîäèðîâêà: Ïîèñêîâûå ñëîâà: ï ï ï ï ï ï ï ï ï ï ï ð ï ð ï ð ï

ST-ECF Instrument Science Report ACS 2001-003

ACS Grism Simulations using SLIM 1.0
N. Pirzkal, A. Pasquali, J. R. Walsh, R. N. Hook, W. Freudling, R. Albrecht, R.A.E. Fosbury April 4, 2001

ABSTRACT We introduce SLIM, a slitless spectroscopy simulator written in Python which can be used to simulate the ACS grism and prism modes. SLIM was designed to produce realistic, photometrically correct two-dimensional images from which spectra can be extracted. Here, we outline the features of SLIM and present some WFC and HRC grism simulations of emission line objects. We include emission line S/N estimates of a Seyfert 2 galaxy as a function of exposure time and observed V magnitude when observed with the WFC grism mode, as well as estimates of the S/N of the emission lines of a Cygnus A clone when observed using the HRC grism mode.

Introduction
A slitless spectroscopy simulator is required to create realistic data appropriate for the Advanced Camera for Surveys (ACS) slitless spectroscopic modes, and to produce simulated data to test slitless spectroscopy extraction software. Our basic requirement is to have a simulator that can, based on our current knowledge of the instrument, generate both geometrically and photometrically realistic images. The emphasis is made to keep the simulator simple and its tasks well defined. We have thus developed a simulator called SLIM which is able to disperse an input spectrum in a well-controlled manner. It does this in a very conservative manner by direct convolution and avoids using Fourier transforms. SLIM is designed to produce direct and grism images with 1 second exposure times which can be scaled up to the appropriate exposure times and have the correct background and noise added by the user.
The Space Telescope European Coordinating Facility. All Rights Reserved.


ST-ECF Instrument Science Report ACS 2001-003

SLIM Implementation
SLIM is implemented in Python, which has several advantages over more classical languages. Python is a free, well documented and well supported language which can be used with almost all known combinations of hardware and operating system. While a scripting language, Python has been shown to be very flexible and appropriate for both small and large projects. One thing that makes Python attractive for this project is the existence of many Python extension packages which provide useful high-level data types. For example, SLIM makes extensive use of the Python Numeric package which allows to store and manipulate images as sets of arbitrarily large arrays. Furthermore, input spectra and throughput curves are loaded as interpolated functions using the Python Scientific module. This allows them to be treated as a set of continuous functions, and not just discrete datasets, hence avoiding all the problems associated with non-uniform sampling of input spectra and throughput curves. The speed of an interpreted scripting language such as Python is somewhat lower than that of a low level C program. This drawback is, however, offset by the decreased development time that is achieved using a modern, feature rich, object oriented language which requires no compilation phase.

SLIM Interface
The program interface is kept as simple as possible. Every aspect of the simulator is controlled through the use of three simple text files containing the various simulation parameters (Examples are shown in the Appendix). The first one contains a list of input objects (2D Gaussian descriptions, positions, spectral types, redshifts, and reference magnitudes). The second one contains general simulation parameters such as telescope mirror size, CCD gain, pixel size, and a list of throughput files. Finally, the last configuration file contains a field-dependent polynomial description of the polynomial dispersion relation to apply to the input spectra. SLIM is designed to use the CDBS calibration database files and reads those directly, ensuring that as new ACS calibration files are made available, it will be possible to use these with SLIM. The only thing needed to generate a SLIM simulation is knowledge of the various CDBS files that are needed in order to account for the various elements in the optical train.

SLIM Main Features
Configured using simple text files (See examples in Appendix). Creates slitless simulations by dispersing a 1D input spectrum and, at each wavelength, convolving it with a Gaussian PSF of the appropriate shape and size to simulate the response of the instrument and the intrinsic shape of the object.

2


ST-ECF Instrument Science Report ACS 2001-003 An arbitrary number of throughput files can be used to simulate any combination of mirrors, windows, filters, dispersive elements, and CCD's. Photometric (Vegamag) zero points are computed on-the-fly using the input list of filters, throughputs and a spectrum of Vega. Arbitrary, nth order polynomial descriptions of the dispersion relations can be used, allowing both near-linear (grism) and highly nonlinear (prism) simulations to be generated. Field dependence can be accounted for. Does not require any a-priori sampling of the input spectrum or throughput files with . Produces WFC simulations that agree within 1 percent with the ACS ETC. Can simulate multiple order grism observations. Handles WFC, HRC, and SBC. Easily adaptable to other instruments.





SLIM Inner Working
A list of 2D Gaussian objects to simulate is read from the input object list. This list contains the object positions in the direct image, descriptions of their ellipsoidal shapes, magnitudes (Vegamag), spectral types, and optional redshifts. For each object in this list, the associated input spectrum is loaded, redshifted as required, and scaled to the desired magnitude in the reference bandpass. Each object's input spectrum is then dispersed along a curve, as described by the polynomial dispersion relation description, and projected onto a grid which can be optionally a few times finer than the desired output dispersed image. At each point of the dispersed spectrum, a Gaussian PSF of the appropriate size (accounting for the size of the object and for the diffraction limit) is computed, scaled by the input spectrum flux, and added to the final output dispersed image. While this approach is very conservative and somewhat slow, it ensures that flux is conserved when spectra are dispersed. Direct PSF convolution avoids FFT-related artifacts and allows for the size of the PSF to be made wavelength-dependent. In addition to the steps described above, SLIM can optionally incorporate the effect of spectral fringing by multiplying each input spectrum by a wavelength-dependent fringing pattern. The fringing model used by SLIM is based on a model of the STIS fringing which was created by Eliot Malumuth.

3


ST-ECF Instrument Science Report ACS 2001-003

SLIM Examples
SLIM has been successfully used to create simulations of grism and prism observations using all three ACS channels (WFC, HRC, and SBC). Here, we present grism simulation of simulated HDF-N grism observations using the WFC and the HRC A simulated HDF-N field was generated using the HDF-N object catalog from FernÀndezSoto et al., 1999. This catalog lists 1067 HDF-N galaxies for which photometric redshifts have been determined and we selected from this catalog 410 objects for which V<26. As the ACS field of view is larger than the field of view of WFPC2, we reproduced the actual HDF-N at the center of a single 4096x4096 array, while the surrounding area was randomly filled. This was done by selecting objects from the input catalog, randomizing their position angle, and adjusting their redshift, size, and brightness by a few percent.The final number of objects in our HDF-N simulations is 2430. Each galaxy was assigned a separate redshifted template spectrum (from Kinney et al. (1996)) of the galaxy's listed morphological type (from FernÀndez-Soto et al., 1999). WFC Using this object input list, SLIM was used to generate three 1-second direct WFC images (F435W, F606W, and F814W) as well as one 1-second G800L grism WFC image. These images were then scaled up to integration times of 1000 seconds and the appropriate amount of background, dark-current, and readout noise were added using the MKNOISE routine in IRAF. A color composite created using the three simulated WFC direct images is shown in Fig. 1 and the corresponding simulated grism image in shown in Fig.2. The simulated grism image contains the zeroth, first, and second spectral orders and the dispersion direction makes an angle of 2 degrees with respect to the x-axis (Hartig, 2000). Figure 3 contains an enlarged portion of Fig. 2, showing an Sc, V=21.6, z=0.26 dispersed galaxy spectrum. HRC The same HDF-N object list and the procedure described above were also used to generate HRC direct and grism simulated images that have a 10 hours integration time. These are shown in Fig. 4 and 5. The dispersion angle is 45 degrees with respect to the x-axis. Note that while the field of view of the HRC (26"x29") is smaller than the one from the WFC (202"x202"), the HRC grism images provide twice the spectral resolution of the WFC (29 å/pixel vs 40 å/pixel).

4


ST-ECF Instrument Science Report ACS 2001-003

Figure 1: A color composite (F435W, F606W, F814W) of a simulated ACS/WFC image of an HDF-N like field.The true HDF-N distribution of objects was preserved near the center of this field. Integration time is 1000 seconds in each filter. The object pointed to by the arrow is an Sc type, V=21.6 galaxy at z=0.26.

5


ST-ECF Instrument Science Report ACS 2001-003

Figure 2: A simulated 1000 second WFC grism observation of the HDF-N. This is the same field that is shown in Fig 1. The object pointed to by the arrow is an Sc type, V=21.6 galaxy at z=0.26, and it is shown in greater details in Fig. 3.

6


ST-ECF Instrument Science Report ACS 2001-003

Figure 3: The raw dispersed spectrum of the z=0.26,V=21.6, Sc type galaxy from Fig. 2. Emission lines are clearly visible.

7


ST-ECF Instrument Science Report ACS 2001-003

Figure 4: A color composite (F435W, F606W, F814W) using SLIM to simulate 10 hour ACS/HRC exposures of a sub-part of the HDF-N like field shown in Fig. 1. The field of view is 26"x29". The object pointed to by the arrow is a V=24.1 starburst type galaxy at z=1.6.

8


ST-ECF Instrument Science Report ACS 2001-003

Figure 5: A SLIM simulation of a 10 hour HRC grism observation of the HDF-N. This is the same field that is shown in Fig. 4. The arrow points to the dispersed spectrum of the V=24.1, z=1.6 galaxy from Fig. 4.

WFC GRISM SPECTROSCOPY
The spectrum of NGC 1068, a Seyfert 2 galaxy and a prototype of the AGN class, was used to simulate WFC grism spectra of objects at different redshifts and observed V magnitudes. Its spectrum is shown in Fig. 6 in units of rest wavelength and with a resolution of 0.5 å/pix. The prominent emissions of [OIII] 4959, 5007 å and H can be identified, while features due to [NeIV] 2439, MgII 2798, [NeV] 3426, [OII] doublet at = 3727 å and [NeII] 3869 can be recognized shortward of 4000 å. This spectrum was used to generate simulations of NGC 1068-like objects at redshifts of z=0.7 and z=1.6, and for a range of V magnitudes from V=20 to V=28. All objects have an elliptical shape, with a size of 0.1"x0.25", with the orientation of the semi-major axis

9


ST-ECF Instrument Science Report ACS 2001-003 making an angle of PA=45o with respect to the x-axis. These simulations were scaled to integrations times of 1000 seconds, 1800 seconds, 1, 2, 3, and 4 hours, and appropriate levels of background, read-out noise, and dark current were added using the MKNOISE routine in IRAF. Individual spectra were then extracted from the simulated images. The extraction was performed both with and then without doing any background subtraction using the IRAF routine APALL. All S/N estimates quoted hereafter were derived using the non-background subtracted spectra. Figure 7 contains examples of background-subtracted, 1st order simulated spectra of NGC 1068-like objects, in a 1000 second exposure, and if at z=0.7 (right column) or at z=1.6 (left column). These extracted spectra were neither flux nor wavelength calibrated and their shape reflects the effect of the G800L grism response function; the units are Counts vs Pixels from the object position in the direct image.

REDSHIFT 0.7 A redshift of 0.7 moves the optical blue portion of the Seyfert 2 spectrum into the wavelength range covered by the ACS grism. It therefore becomes possible to detect the H and the [OIII] 4959, 5007 å lines in the extracted 1st order spectra of Fig. 7 (left column).These features are however blended together by the low spectral resolution of the grism but due to the relatively strong flux of the [OIII] doublet, the H + [OIII] blend can be detected down to V=25 with a S/N of ~5 in a 1000 second exposure. To better quantify the line detectability of the WFC grism we have computed the S/N ratio in the H + [OIII] blend as a function of observed V magnitude and exposure time. The line S/N ratios, as measured in non background-subtracted spectra, are plotted in Fig. 8 as a function of exposure time for two observed magnitudes (V=25 and 26). It can be seen that the H + [OIII] emission of a source at V=25 can be detected with a S/N of ~8 in an 1800 second exposure, while a one hour integration is required to observe the same feature with a S/N of ~3 in the spectrum of a galaxy at V=26.

REDSHIFT 1.6 At a redshift of 1.6, the near UV part of the Seyfert 2 restframe spectrum becomes accessible to the ACS grism. Consequently, emission lines like [NeIV] 2439, MgII 2798 and [NeV] 3426 can be seen in the 1st order spectra shown in Fig. 7 (right column). The integration times are still 1000 seconds and the spectra shown in Fig. 7 have been background-subtracted. Since these lines are fainter than the optical [OIII] doublet threshold for NUV lines at z=1.6 and for a S/N ratio of ~4 exposure. We have determined the S/N ratio in the line for function of exposure time and observed V magnitude. The 9. The MgII 2798 line can be detected in the spectrum of at 5000 å, the grism detection occurs at V=23 in 1000 second the MgII 2798 feature as a results are summarized in Fig. galaxies at V=22 and 23 with a

10


ST-ECF Instrument Science Report ACS 2001-003 S/N value of ~9 and ~4 respectively, in a1800 second exposure. In a one hour exposure, the same emission line can be observed in a source at V=24 with a S/N of ~3.

4000

6000 Wavelength (A)

8000

Figure 6: The spectrum of NGC 1068, a Seyfert 2 galaxy, used to simulate WFC grism spectra of objects at different redshifts and observed V magnitudes, is shown.

11


ST-ECF Instrument Science Report ACS 2001-003

Figure 7: WFC grism SLIM simulations of Seyfert 2 galaxies at various V magnitudes and redshifts, for an integration time of 1000 seconds. These are background-subtracted, 1st order spectra redshifted to Z = 0.7 (left column) and Z = 1.6 (right column).

12


ST-ECF Instrument Science Report ACS 2001-003

Figure 8: S/N of the H + [OIII] blended lines of a Seyfert 2 galaxy as a function of exposure time and for the observed magnitudes V=25 and 26.

Figure 9: S/N ratio in line for the MgII 2798 feature as a function of exposure time and observed V magnitude.

13


ST-ECF Instrument Science Report ACS 2001-003

HRC GRISM SPECTROSCOPY
A number of HRC grism simulations were created using the spectrum of Cygnus A, one of the most powerful radio galaxies at low redshift also harboring an active nucleus and shown in Figure 10. The contribution of the underlaying elliptical galaxy has been subtracted so that only scattered light can be observed. The spectrum is characterized by nebular emission lines among which the strongest are the [OIII] doublet at 5007 å and the H line. We have redshifted the spectrum of Cygnus A by 0.3 and scaled it to observed magnitudes ranging from V=22 to V=27 in order to mimic local emission line galaxies. This luminosity range was chosen to avoid having the strongest emission features in the spectrum from saturating the CCD. We simulated 6 objects, elliptical in shape, with a size of 0.05" x 0.12", and with a major axis at PA=20o with respect to the x-axis. The simulations were scaled to the same exposure times that were used for our simulations of NGC 1068, and realistic background, read-out, and dark current levels were added to each image using the MKNOISE IRAF routine. The simulated spectra were once again extracted both with and without background-subtraction turned on. The extraction had to be performed by first rotating the grism image by 45o clockwise in order to align the dispersion axis along the x-axis, and by then adding up the rows (in the spatial direction) containing the dispersed spectra. These steps were necessary because the APALL routine in IRAF was not able to properly trace spectra tilted by 45o with respect to the x-axis.

Figure 11 shows spectra extracted from 1000 second simulations (left column); they are background subtracted but neither flux not wavelength calibrated. Their units are Counts vs Pixels from the object position in the direct image. The combination of the selected redshift and of the grism spectral range allows us to identify the [OIII] doublet at 5007 å blended with the H line, the [OI] 6300 line, the blend of H with [NII] 6584, and the [SII] doublet at 6730 å, down to V=26 and with a S/N of ~6. The S/N ratio in the line has been measured in the non background-subtracted spectra for both the H + [OIII] blend at 5007 å and for the H + [NII] + [SII] blend as a function of the integration time and of observed V magnitude. These results are summarized in Fig, 12. While the line S/N value for the [OIII] and the H blends is larger than 20 for targets brighter than V=25 in an 1800 second exposure, it decreases down to ~6 at V=26. Both lines can be detected with a S/N ratio of ~5 in a 1 hour exposure (right column of Fig. 11) of a galaxy at V=27.

14


ST-ECF Instrument Science Report ACS 2001-003

0 4000 5000 6000 Wavelength (A) 7000 8000

Figure 10: Optical spectrum of Cygnus A, one of the most powerful radio galaxy at low redshift, with a resolution of 2.4 å/pix.

15


ST-ECF Instrument Science Report ACS 2001-003

Figure 11: Extracted spectra from HRC grism simulations of a Cygnus A-type objects with magnitudes ranging from V=22 to V=27. The SLIM 1st order spectra are background-subtracted. Exposure times are 1000 seconds (left column) and 2 hours (right column).

16


ST-ECF Instrument Science Report ACS 2001-003

Figure 12: S/N of the [OIII], and Ha + [NII] + [SII] (blend) features in HRC grism observations of Cygnus A-type objects as a function of integration time and observed magnitudes V=26 and 27.

Conclusions
SLIM, a slitless spectroscopy simulator was written to allow us to investigate the expected properties of the various ACS grism and prism modes. SLIM has been successfully used to generate 2D HRC and WFC grism images and we simulated observation of emission line galaxies to quantitatively estimate the S/N of typical emission features in objects of various observed V magnitudes. SLIM is available for download at http://www.stecf.org/ software/. Future planned improvements for SLIM include the use of Tiny Tim simulated ACS PSFs instead of simple Gaussian PSFs, and to allow one to input realistic images of objects instead of simple 2D Gaussian descriptions.

References
FernÀndez-Soto, A., Lanzetta, K. M., Yahil, A., 1999, ApJ, 513, 34 Hartig, G., 2000, Personal communication Kinney, A. L., Calzetti, D., Bohlin, R. C., McQuade, K., Storchi-Bergmann, T., Schmitt, H. R. 1996, ApJ, 467, 38

17


ST-ECF Instrument Science Report ACS 2001-003

Appendix
Sample input object list This file contains the list of objects, described as simple 2D Gaussians. An optional redshift can be included. (For description of the meaning of the various parameters, please refer to the SLIM 1.0 distribution.)
# obj_x 400 400 400 obj_y angle size_x size_y mag 300 0 .1 .1 10 400 40 .3 .1 14.5 500 90 5 5 1. spec_type Z ngc7009_fornax 1.12 flat flat

Sample configuration file This file describes the various optical throughputs to take into account, some detector specific parameters, and the reference magnitude system. (For description of the meaning of the various parameters, please refer to the SLIM 1.0 distribution.)
# WFC Configuration file INSTRUME = WFC DISFILTER = GL800 # Pixel size in " PSIZE = 0.050 MSIZE = 2.4 GAIN = 1.0 # Directory containing CDBS files CALIBD = /xxx/calib/acs/ CALIBD = /xxx/calib/ota/ CALIBD = /xxx/calib/nonhst/ # Directory containing the input spectra SPECD = /xxx/starsp SPECD = /xxx/ESO_ETC/ # The various grisms/prisms DISP=acs_g800l0_001.fits,acs_g800l_002.fits,acs_g800l2_001.fits # The filter in which the input object magnitude system are given REFFILTER = johnson_v_003_syn.fits # Name of filter to compute direct image flux DIRFILTER=acs_f606w_003_syn.fits # Reference spectrum REFSPEC = vega.spc # WFC specific throughputs F = acs_wfc_ccd1_008_syn.fits # HRC QE F = hst_ota_007_syn.fits # HST OTA F = acs_wfc_im123_004_syn.fits # Mirrors F = acs_wfc_ebe_win12f_005_syn.fits # Window # Prism/Grism file DISPFILE = /xxx/slim-calib/ACS_WFC_1_GL800.txt

18


ST-ECF Instrument Science Report ACS 2001-003 Sample dispersion description This file describes the polynomial description of the dispersion relation and its field dependence (None in this example). (For description of the meaning of the various parameters, please refer to the SLIM 1.0 distribution.)
XOFF_0 -132.2 YOFF_0 0 THETA_0 45. DXMIN_0_0 0 DXMAX_0 10 DLDP_0_0 0 DLDP_0_1 726.74 XOFF_1 -132.2 YOFF_1 0 THETA_1 45. DXMIN_1 0 DXMAX_1 300 DLDP_1_0 0 DLDP_1_1 25 XOFF_2 -132.2 YOFF_2 0 THETA_2 45. DXMIN_2 0 DXMAX_2 600 DLDP_2_0 0 DLDP_2_1 12.44

19