Short Fourier Transforms data files

This directory contains Short Fourier Transform (SFT) data files. SFTs are a standard input data source for continuous gravitational wave analysis pipelines. An SFT is stored in an SFT file in a binary format. Each SFT file may contain multiple SFTs from the same detector. The SFT file format and naming scheme are further described in the SFT specification.

The SFTs in this directory are organised as follows.

Top level

Top level directories (at the same level at this README file) contain an individual SFT dataset. They are named according to the following scheme:

 {top-level-dir} = {det}_{Tsft}SFT
                   _O{run}{kind}+R{rev}+C{chan}+W{win}
                   _{bandw}

(Note that the right hand side is broken into separate lines for readability; no whitespace appears in the actual directory names.)

Terms in braces denote the following components:

• det: Gravitational wave detector of the data. Common choices are:

  • H1: LIGO Hanford (4 km)
  • K1: KAGRA
  • L1: LIGO Livingston
  • V1: Virgo

• Tsft: Timespan of the data, in seconds, which is Fourier transformed to create each SFT. A typical choice is 1800 seconds, which is generally appropriate for searches for continuous gravitational waves from isolated neutron stars.

• run: Observing run number. For example, run = 1 corresponds to the first observing run (O1) of the LIGO Observatory, from September 12, 2015 through January 19, 2016.

• rev: Revision number of the SFT dataset. Typically rev = 1 the first time the dataset is generated. If, however, it is later found that a dataset was generated incorrectly, it may be regenerated with an incremented revision number, e.g. rev = 2. Please check carefully that you are using the latest revision of an SFT dataset!

• chan: Name of the data channel (without the detector prefix) from which the SFTs are generated. Due to limitations on the SFT file/directory naming scheme, the channel name is stripped of any non-alphanumeric characters, which may make it difficult to read. For example, chan = GDSCALIBSTRAINCLEANGATEDG01 corresponds to the data channel {det}:GDS-CALIB_STRAIN_CLEAN_GATED_G01.

• win: Windowing applied to the SFTs. Typical choices are:

  • TKEY{n}: Tukey windowing, with parameter beta = n / 5000. A standard choice for beta is 0.001, which corresponds to n = 5. Note that, with this choice of beta, the Tukey window is very close to a rectangular window. This windowing is chosen to mitigate any transients at the start and end of the SFT. The loss in signal power due to the windowing is on the order of 0.1%; see footnote 65 in this paper.

  • HANN: Hann windowing.

• bandw: Bandwidth of the SFTs. This may be one of two choices:

  • BROADBAND: SFTs which are broadband in frequency, typically covering the entire useful/sensitive band of the detectors. Conversely, these SFT files are typically limited in time; each SFT file contains only 1 SFT, and spans a time Tsft.

  • NARROWBAND: SFTs which are narrowband in frequency, typically covering only a few Hz. Conversely, these SFT files typically cover longer time ranges; each SFT file may contain more than one SFT, and each SFT file may therefore span a time much greater than Tsft.

Broadband SFT subdirectories

Broadband SFT files (where bandw = BROADBAND) are organised below the top level into subdirectories based on their starting GPS time, as follows:

 {broadband-sft-path} = {top-level-dir}/
                        GPS{start-1e6}M/
                        {broadband-sft-filename}

where start-1e6 is the starting GPS time of the file divided by 1e6, rounded down. For example, an SFT starting at GPS time 1,389,258,149 would be found in the GPS1389M subdirectory. Within each GPS... subdirectory, SFT files are named as follows:

 {broadband-sft-filename} = {site}-{num}
                            _{det}_{Tsft}SFT
                            _O{run}{kind}+R{rev}+C{chan}+W{win}
                            -{start}-{span}.sft

where, in addition to the top-level directory components:

• site: Gravitational wave observatory site. This is always the first letter of det, e.g. H, K, L, V.
• num: Number of SFTs contained in the file.
• start: Starting GPS time of the SFT data in the file.
• span: Total timespan of all the SFTs in the file. If num = 1, then span = Tsft, otherwise span > Tsft.

Narrowband SFT subdirectories

Narrowband SFT files (where bandw = NARROWBAND) are organised below the top level directory into subdirectories based on their frequency bands, as follows:

 {narrowband-sft-path} = {top-level-dir}/
                         NBF{freq-Hz}Hz{freq-bin}W{width-Hz}Hz{width-bin}/
                         {narrowband-sft-filename}

{freq-Hz} and {freq-Hz} are the result of the integer division of the starting bin of the SFT by Tsft, and the remainder of that division, respectively. For example, a Tsft = 1800 narrowband SFT starting at bin 180,900 would have

 freq-Hz = 180,900 / 1800 s = 100.5 Hz = 100 Hz, rounded down
 width-Hz = 180,900 - 100 Hz * 1800 s = 900 bins
 i.e.
 starting bin = 100 Hz * 1800 s + 900 bins = bin 180,900

Similarly, {width-Hz} and {width-Hz} are calculated from the bandwidth of the SFT. Within each NBF... subdirectory, SFT files are named as follows:

 narrowband-sft-filename = {site}-{num}
                           _{det}_{Tsft}SFT
                           _O{run}{kind}+R{rev}+C{chan}+W{win}
                           _NBF{freq-Hz}Hz{freq-bin}W{width-Hz}Hz{width-bin}
                           -{start}-{span}.sft

where the components are as described above.

For further information

Please consult the SFT specification.