This product comprises mock observations of the VELA suite of cosmological simulations, demonstrating a physical model of galaxy formation over cosmic history. The Vela-Sunrise mock images are synthetic ultra high resolution space telescope images spanning a range of halo masses and assembly histories across cosmic time. The dataset spans the cosmic time evolution of 35 galaxies over 10-50 timesteps, each with approximately 20 viewing angles and a common set of 34 HST, JWST, and proposed WFIRST filters. These mock data products are also appropriate for modeling observations of future large-aperture telescopes.
Below we briefly summarize the basic properties and procedures for creating this dataset. A more comprehensive description can be found in papers such as those by Moody et al. (2013), Snyder et al. (2015), and Simons et al. (2019), as well as the references therein and below.
If you use these data products, you can refer to this collection using its DOI: https://doi.org/10.17909/t9-ge0b-jm58
OverviewThere are two basic data products stored in FITS format:
To simplify storage and access, the final dataset was organized on both a per-simulation and per-filter basis into two sets of compressed ".tar.gz" files. Each of these sets contains an identical copy of the data. Therefore, files can be obtained by first selecting a simulation then a filter, or by first selecting a filter to obtain the corresponding data for all simulations at once. Bulk download scripts are also available to automate this process in either case.
Hydrodynamical Simulation Inputs
Each simulation represents the time evolution of an individual galaxy forming, which are labeled VELA01 through VELA35. This project analyzed each simulation at approximately 10-50 timesteps over cosmic time. Cosmological hydrodynamic simulations were performed in a WMAP5 cosmology with the gasdynamics + N-body Adaptive Refinement Tree code (ART; Kravtsov, Klypin & Khokhlov 1997; Kravtsov 2003). In addition to gas dynamics and gravity, the simulations implement sub-grid physical models such as cooling, background photoionization, star formation, metal enrichment, thermal feedback from supernovae, and feedback from radiation pressure and heating from young stars (Ceverino et al. 2014).To initialize the galaxy formation simulations, dark matter haloes were selected randomly from those with virial masses 11 < log(Mhalo/Msun) < 12 at redshift 0.8 formed in a coarsely resolved N-body simulation. Their initial conditions were re-sampled and re-simulated in full hydrodynamics with much greater resolution, forming galaxies with stellar masses approximately 9 < log(Mstars/Msun) < 11. In total, the evolution of 35 haloes were simulated with several iterations of physical models, where the focus of this release is the "Generation 3" models described by Ceverino et al. 2014. Of the 35 Vela simulations, 34 satisfied the requirements for carrying out the systematic mock imaging project.
Radiative Transfer Calculations
The data products presented here were created using the dust radiative transfer code Sunrise (Jonsson 2006; Jonsson et al. 2010). The software pipeline to convert the ART simulations into Sunrise input files was written by Moody et al. (2013) and accelerated by Simons et al. (2019) to enable efficient re-creation of these mock datasets. This pipeline uses the YT package (Turk et al. 2011) and custom Python algorithms to read the ART data and construct the octree structure passed to Sunrise.
The team analyzed simulations outputs spaced by 0.01 in scale factor, generally from roughly 0.05 to 0.5, for up to 50 timesteps per simulation, or approximately 1200 timesteps in total. At each such timestep, the Sunrise radiative transfer simulations were performed based on stellar population and dust models described by Simons et al. (2019). In brief, they assume Starburst99 models augmented by HII region models based on MAPPINGS for very young stars (Groves et al. 2008). For dust grains, the calculations assumed a dust grain composition and size distribution that reproduces typical observations of the Milky Way (Weingartner & Draine 2001), with total dust mass proportional to the metal mass in each cell predicted by the hydro simulations.
With those assumptions, Sunrise performed three-dimensional polychromatic dust radiative transfer through this dust distribution, into a set of 19 "cameras" arranged around each source. A subset of these cameras targeted the "face-on", "edge-on" and "45-degree" inclinations with respect to the angular momentum of the galaxy gas, a subset targeted random directions fixed over cosmic time, and the final subset targeted randomly oriented directions that vary for each timestep.
The calculations resulted in multiwavelength high-spatial resolution predictions. For each of a set of 34 idealized HST, JWST, and WFIRST broadband filter throughput curves, Sunrise saved an 800x800 pixel image for each of the 19 cameras. These images constitute the main product provided this HLSP. In addition, Sunrise stores several ancillary images for enhancing subsequent analyses. A particularly useful set of 10 such images are known as the "auxiliary" quantities, which are projections of intrinsic simulation quantities into each of the same 800x800 pixels as the mock images. We provide these images for each source analyzed in this project. Another potentially useful set are images without the effect of diffuse dust attenuation in the ISM of each galaxy; this HLSP provides all such images for the VELA01 simulation, and to conserve disk space, a broad subset of 7 filters and 7 cameras for the remaining simulations.The Sunrise image pixels are stored as an idealized surface brightness of the sky at the camera position, normalized by the filter curve area. This is a distance invariant quantity that does not depend on the multiplicative normalization of the assumed filter curves but only on their shapes.
Mock Image File Structure
Each image was converted into flux units by first assuming that the source has the same redshift as the underlying hydro simulation timestep and applying redshift-dependent cosmological surface-brightness dimming. The pixel units are then converted from internal Sunrise units (W/m/m*m/Sr) into nanoJanskies. This image is stored in FITS format as the primary HDU with extension name (EXTNAME) IMAGE_PRISTINE in all of the mock image files in this release. For the HST and JWST filters, an additional step was taken to resample the pixels and convolve with model point-spread functions (PSFs) supplied by TinyTim and WebbPSF, respectively. Pixel sizes were chosen to match either existing (for HST) or idealized (for JWST) surveys that achieve nearly the diffraction limit for the respective instruments, in other words assuming sub-pixel dither patterns that allow for Nyquist sampling of the PSF. These "mock observed" images, where they exist, are stored as the 2nd FITS extension named IMAGE_PSF. Where the no-diffuse-dust images were also included, they are saved with extension name IMAGE_PRISTINE_NONSCATTER.
All images in this HLSP are free of astronomical and instrumental noise. The intent is that users of these products will supply their own noise model appropriate for their particular application.
The directory and file naming conventions are as follows, assuming that any ".tar.gz" files have been unpacked already. Files are organized into one directory per simulation ("vela??/") with subdirectories for each camera ("cam??/") then mission ("hst/", "jwst/", or "wfirst/") and instrument (e.g. 'wfc3/'). Each mock image file (containing 1-3 image HDUs each) is then stored in the final level for each filter (e.g. "f160w/"), one for each simulation timestep. The naming convention is such that each file can be constructed as a string as follows:where:
Aux Image File Structure
Each set of mock images for a particular source has an associated set of auxiliary ("aux") images which present intrinsic properties of the simulation, such as gas mass and stellar mass, in each of the same pixels as the mock images. Therefore the mock data can be mapped directly to the intrinsic state of the galaxy simulation on a per-pixel basis. Since they are independent of observing mode, these images are stored at the same level of the directory structure as the "mission" element above. For the above example, the aux images are stored at:where the source ".tar.gz" file for this product is vela27/hlsp_vela_none_none_vela27_aux_v3_sim.tar.gz (14GB).
Each aux FITS file contains an HDU named CAMERA??-AUX, for this example "CAMERA12-AUX" which contains a 10x800x800 array, where each 800x800 pixel slice is an image array containing a given intrinsic property. The HDU header describes the index, name, and units of each slice. For example, the gas mass is slice 1 (index=0), and the stellar mass is slice 5 (index=4).
Catalogs File Structure
Each of these products has an associated set of ASCII catalog files summarizing basic properties of these products, such as magnitude and masses of the constituents of the source. Many of these properties are also stored as FITS keyword values in each of the underlying data files.
The Source Catalogs summarize very basic information about the mock images, such as the total flux and magnitude in the associated filter, as well as information from the aux images such as stellar mass, and the mass of dark matter within the virial radius calculated with YT. In addition, each line contains the path to the relevant file containing the image data. The Aux Catalogs are shorter and contain only the intrinsic information about the state of the simulation. Each type of catalog also has a concatenated version containing all entries for all 34 simulations.
The catalog files are in whitespace-delimited ASCII format with a single-line header containing the column names, which are:
Camera MeaningsThe 18 viewing directions were chosen as follows, where "face-on" and "edge-on" are defined with respect to the galaxy's baryons angular momentum vector. Quantitative definitions of these cameras are stored in the header of each FITS file.
Generation 3 Vela Simulation Mock Images
NOTE: Two copies of the same data are stored in the files below, so there is usually no need to obtain files using both the "By Filter" and "By Simulation" methods. Similarly, there is no need to apply both the "Bulk Download Script" and the "VELA?? Download Scripts". Users wishing to obtain a simple copy of all the available data are advised to use the "By Filter" method.