Description

As part of the ST-ECF support for the spectroscopic modes of Wide Field Camera 3 (WFC3), the slitless spectroscopy group of the ST-ECF has developed a dedicated simulation package applicable to WFC3. Whilst the package was initiated for exploitation of WFC3 slitless grism modes, it is equally applicable to other slitless spectroscopy modes of HST, such as ACS and NICMOS. An option of the simulator is to produce a direct image through a selected filter to match the slitless spectrum image.

The simulation package uses the same components as used by aXe for the extraction of slitless spectra and thus aims at spectrophotometric integrity - useful for observation design but essential for the quantitative assessment of slitless data. The simulation package will help HST users during their Phase I and Phase II proposal preparation to:

• gain a 2D-impression of the layout of the target field and the likely problems to be encountered in slitless spectroscopy, e.g. crowding, spectral overlap, presence of bright spectra;
• check the exposure time to determine a given signal-to-noise in a 2D spectrum. The spectroscopic ETC's at the STScI provide more flexible determination of S/N for an individual 1D spectrum;
• by adjustment, determine the optimal pointing and roll angle for a specific target field, such as to minimize the contamination of spectra of interest;
• learn what to expect (in terms of spectral resolution and sensitivity) from slitless spectra, and thus choose the best instrument and configuration;
• and incidentally become familiar working with and reducing slitless spectroscopic data with the aXe package.

While the primary motivation to start the aXeSIM project was Servicing Mission 4 and the installation of WFC3 with its three grisms, aXeSIM will also work for all the ACS grism and prism modes (WFC/G800L, HRC/G800L, HRC/PR200L, SBC/PR130L, SBC/PR110). Including these slitless modes simply requires including the necessary configuration files specifying the instrument-specific aspects of the slitless spectra; these files are already available from the ST-ECF web (http://www.stecf.org/instruments/ACSgrism/calibration).

The simulation package also finds application to analysis of existing slitless spectroscopy datasets. The simulations can be extracted identically to slitless spectral data, allowing quantitative assessment of detected spectra and spectral features, such as emission lines. Tasks such as matching spectra against templates, convolved with the actual object size and at the grism spectral resolution, determining the detection limits for spectra of given types and measuring cross-contamination between spectra are readily achieved. More extensive studies such as assessing completeness limits of survey observations for a variety of object classes can also be performed.

aXeSIM is made available in two ways. It is distributed via the ST-ECF homepage as a PyRAF/IRAF module and is made available via a web interface. Though based on identical software, the web interface targets new, less experienced or occasional users and offers a subset of the options existing in aXeSIM.

In addition to the software we will also provide from our webpages the necessary configuration and sensitivity files (for WFC3 initially based on ground calibration data) to run aXeSIM. The core part of these files (trace- and dispersion description) will coincide with the configuration and calibration files necessary for the reduction of slitless data with the 'traditional' aXe extraction software (see [2], [3] and [4]).

User-provided input for the PyRAF module

The user will have to prepare only a single list, called Model Object Table , with the descriptions of the simulation objects as a minimal input. Then the simulations will be done by executing one PyRAF task simdata.

With this minimal input, the Model Object Table contains columns with all the necessary shape and spectral information to completely characterize one object per row. The table format is based on the SExtractor ASCII table format, and the minimal input for the table contains the columns:

-

NUMBER: the object identifier; int

-

X_IMAGE: the x-position of the object [pix]; float

-

Y_IMAGE: the y-position of the object [pix]; float

-

A_IMAGE: the major axis rms [pix]; float

-

B_IMAGE: the minor axis rms [pix]; float

-

THETA_IMAGE: the position angle of the major axis [deg]; float

-

MAG_?<number>*: the AB-magnitude at the wavelength; float 5#5number6#6 [nm]

In addition the user has the opportunity to perform more detailed simulations by e.g. using high resolution spectral templates at different redshifts and flux levels, image templates or total passband curves

For more detailed simulations, the object list needs additional columns such as:

-

SPECTEMP: the index of the model spectrum to use; int

-

MODIMAGE: the index of the image template to use; int

-

Z: redshift of the object; float

The columns SPECTEMP and MODIMAGE refer to the spectral templates and image templates to be used in the simulation, respectively. As a first step, the template is translated to the redshift given in the column Z.

The output of aXeSIM

The output of the aXeSIM simulation task will always consist of a simulated 2D slitless dispersed image with the spectra of the simulated objects. For grism images, the simulations can include several dispersed orders.

If provided with a proper total passband of a filter, aXeSIM will also produce the direct image associated to the slitless image. The total passband curve contains the total system throughput (mirror + instrument + detector + ...) as a function of wavelength for the direct image to be simulated.

The set of direct image - slitless image will be identical to an ideal observed direct image - slitless image pair in standard ACS/WFC3 slitless observations. This means that a positional offset, which might be applied in the observations by default (as it is the case for e.g. ACS data with HRC/PR200L) will also be present in the simulated images.

Sky background (provided by the user either as single value or background image file in [e/s]) and random noise (readout- and photon-noise from background and objects) will be added to the output images.

It is possible to close the loop by performing a simple aXe extraction at the end of the simulations in aXeSIM.
Alternatively, the user can use SExtractor on the direct images and then perform a standard extraction using the 'original' aXe software.
Making the spectral extractions on the simulated data is very important to, for example, check the detectability of spectral features as a function of the signal-to-noise ratio or to make completeness and reliability tests for scientific publications.

All final output of aXeSIM is written to the directory to which the environmental variable AXE_OUTSIM_PATH is pointing to (see Sect. 6).

Limitations of aXeSIM

There are a number of effects and features in real data that are currently not included in the aXeSIM simulations:

• non-linear detector effects such as saturation or blooming are not simulated;
• aXeSIM neither adds dark current nor takes dark current into account in the noise model;
• aXeSIM does not include a detector or large scale flatfield in the simulations;
• images simulated with aXeSIM do not have intra pixel sensitivity variations as images from real detectors (e.g. NICMOS) do;

Users can compensate some of these aXeSIM deficiencies, e.g. by adding the dark current to the sky background, or applying non-linear effects on aXeSIM images.

The PyRAF module

Similar to the aXe extraction package [2] (available in PyRAF under stsdas.hst_calib.acs.axe) the aXeSIM package for PyRAF has three layers:

• at the bottom layer, executables compiled from C-programs provide the fast processing speed necessary for computing the pixel values in the simulated images. Most of this code does already exist as part of the aXe extraction package (in 'quantitative contamination', see [5]). This guarantees the symmetry between the simulations and the extractions. Moreover this 'legacy' code has been used for some years and therefore has proven to be without major bugs or problems.
• a middle layer with python scripts connects the various C-executables and performs less computationally expensive preparatory work
• a very thin upper layer makes the python scripts accessible from within PyRAF.

This concept is currently followed in all software published in the stsci_python package [6].