SDSS Cosmic Void Catalogs

This page contains information about the cosmic void catalogs constructed from SDSS III data. The catalogs were constructed by Qingqing Mao at Vanderbilt University, using the ZOBOV void finding algorithm. The current void catalogs use the Baryon Oscillation Spectroscopic Survey (BOSS) DR12. All the catalog files can be downloaded here. Read below for a general description of the algorithm and the file formats for the catalogs. The catalogs are also described in detail in Mao et al. (2017). If you have any questions, please contact the authors.

Void Finding Algorithm

We construct our void catalog using the ZOBOV void finding algorithm developed by Mark Neyrinck. The detailed description of the algorithm itself can be found in Neyrinck (2008), as well as on the ZOBOV website.
In short, ZOBOV uses Voronoi tessellation to estimate densities for all galaxies, and then uses the 'watershed' concept to connect low density regions in order to find voids and sub-voids.

How this catalog is made


We assume a flat ΛCDM universe with Ωm=0.3 to convert redshifts into comoving distances for all galaxies. We then convert (RA, Dec, R) into Cartesian coordinates (x, y, z) and run ZOBOV on that.

Buffer Particles

To deal with the survey boundaries, we first make a buffer of randomly distributed points to fill all the "edges" and "holes" in the survey data. Using a buffer allows the Voronoi tessellation to work properly. After the tessellation process, the densities of the buffer are set to be artificially high, so that the buffer points are never part of the voids. ZOBOV can distinguish between the buffer and the real data properly. Using a buffer also helps to distinguish whether a void is fully embedded in the survey volume or is truncated by the survey geometry.


After the tessellation and before the void finding process, all densities (i.e. Voronoi volumes) are weighted using galaxy weights. Weights account for close pair fiber collisions, redshift failures, systematic weights, as well as sector completeness.

Density Threshold

Though ZOBOV can be parameter free, there is an optional density threshold parameter, which can limit the growth of voids into high-density regions. (See the ZOBOV help page for more information.) We provide catalogs with the density threshold parameter set to 0.5. If other parameters are preferred, please contact the authors.

Voids catalog

We parse the ZOBOV output and construct the catalog. For each void, we define the center of the void as the Voronoi volume-weighted average position of galaxies that belong to the void.
ZOBOV identifies all the local minima, which means that the result includes "false voids" in high density regions. A "probability" cut is necessary to ensure the significant detection of voids (refer to the ZOBOV paper for details). We then only keep voids that have higher than 2-sigma significance, and also have a core density lower than 0.5 times the mean density, where the core density is defined as the density of the lowest density galaxy in the void. Note that the choice of (0.5*mean density) is arbitrary, and stricter cuts can be applied easily. We also provide an uncut version of the catalogs that does not include these two quality cuts.

Raw outputs

Raw ZOBOV outputs can be provided upon request.

Description of Available Files

We construct our void catalog in four survey regions separately: CMASS North/South and LOWZ North/South. For each region there are several result files. Files from different regions can easily be identified by the file names. For each survey region, there is a set of *.dat files that contain the quality cuts described above, as well as a set of *_uncut.dat files that contain no quality cuts. Here we describe the format of the available files.


These files contain a list of voids, where each row includes the properties of one void. The columns are:

1) Void ID
2) Center RA
In degrees
3) Center Dec
In degrees
4) Center Redshift
5) Number of Galaxies
6) Total Voronoi volume
Volume is in (Mpc/h)^3. The volume is weighted by the galaxy weights.
7) Effective Radius
In Mpc/h, defined as (3 * Total Volume / 4 / pi)1/3 )
8) Core Number Density
In units of (h/Mpc)^3. This is the density of the lowest density galaxy in the void, defined as 1 / weighted Voronoi volume of that galaxy.
9) Core Density Contrast
Defined as (Core density / mean density - 1.0), where the Core Density is defined as above. The mean density here is redshift dependent. We bin in redshift and calculate n(z), then for each void its core density contrast is calculated by comparing its core density to the mean density at same redshift.
10) Void Density contrast
Defined in ZOBOV, this quantity compares the density at the edge of a void to the density at the center. This is then used to calculate the significance of a void.
11) Probability
Defined in ZOBOV, this shows the significance of a void.
12) Distance from edge
In units of Mpc/h. This is the distance of the void center from the survey boundary.


A file listing the galaxies that belong to each void. Each row contains the Void ID, Number of galaxies in the void, and a list of galaxy IDs. These IDs can be used to match the galaxies in the volume files "*.volumes.dat", which are all in orders starting from 0. Notice that each line has a different number of columns. You can first read the number of members in column (2) to know how many to expect. Columns are:

Void ID, Number of member galaxies, galaxy 1 ID, galaxy 2 ID, .....


A file listing the Voronoi volumes for individual galaxies, as well as galaxy positions. The Voronoi volumes have been weighted by the galaxy weights. Galaxy IDs are the IDs in the original galaxy catalog fits files. If the Voronoi volume is zero, it means that this galaxy is next to the edge of the survey geometry and its Voronoi volume is unreliable. Columns are:

Galaxy ID, RA, Dec, Distance, Voronoi volume

Note of Caution

As shown in the release paper Mao et al. (2017), void sizes are correlated with distance to the survey boundary, which indicates that voids near the boundaries are systematically being truncated and thus have their properties affected. In addition, there may be a problem with spurious voids near the survey boundary (S. Nadathur, private communication). The severity of these issues will depend on the science goal in question and we thus leave it to the user to address them. We provide the distance to the survey edge in our catalogs to make it easy for users to filter out potentially problematic voids.

Authors and Collaborators

This cosmic void catalog is created by Qingqing Mao at Vanderbilt University. The collaborators include Mark Neyrinck (JHU), Andreas Berlind (Vanderbilt), Robert Scherrer (Vanderbilt), Jeremy Tinker (NYU), Roman Scoccimarro (NYU), Cameron McBride (CfA). Send all questions and comments here.