Repository navigation
Grid and ancillary files
This page includes important information on the ALFALFA grids and the accompanying ancillary files; the authoritative source is Haynes et al. 2018. For directions on accessing the grids via the NRAO archive, go here.
There is example code to work out which grid your source(s) might be in the notebooks folder of this repository.
The ALFALFA grids are split into four spectral ranges. These spectral ranges overlap slightly, to ensure that galaxies always fall fully within a single grid.
| Grid | Velocity range (km/s) |
|---|---|
| heliocentric optical = cz | |
| a | -2000 to 3300 |
| b | 2500 to 7950 |
| c | 7200 to 12,800 |
| d | 12,100 to 17,912 |
The channels are spaced 24.4 kHz apart, which corresponds to 5.1 km/s at 1420.4058 MHz. The data are Hanning smoothed, so the effective spectral resolution is 10 km/s.
Note that the data at velocities from -120 km/s to 120 km/s contain the emission from the Milky Way, which is not reliably traced by ALFALFA because of the baselining scheme adopted (see Haynes et al. 2018 for details). The data from -120 km/s to 120 km/s should not be used for science. Consider using the GALFA survey data in this velocity range instead.
The ALFALFA grids have a spatial extent of 2.4°x2.4°. The grids are spaced regulary, with grid centers separated by 8 minutes in right acsension and 2° in declination. The pixel size is 1', and the grid shape is 144x144 pixels. The grids overlap spatially so that (unresolved) sources always fall fully within a single grid spatially.
The grids have two Stokes axes: these are the LL and RR polarizations. The two Stokes parameters can be averaged together to produce Stokes I. Note that which axis is LL and which RR is unclear but that as HI emission is unpolarized, averaging to Stokes I is the appropriate approach to take. The two separate polarizations are included because RFI can be strongly polarized, or a single polarization can be flagged. Thus examining the two polarizations can be used to help verify the validity of a detection.
Notebooks for working with the grids can be found in the notebooks folder of this repository. Briefly, these are:
- example_DECaLS_overlay.ipynb: How to create a total intensity HI map and overlay on a DECaLS optical image
- GridFind.ipynb: How to plot grids, and find which grid a sky position is located in
- PlotGridCatalogue.ipynb: Make plots of the ALFALFA 100 catalogue with ALFALFA grids overlaid
There is one grid that exists outside of the standard ALFALFA grid. This is 1234+06, which was created to capture the long tail of NGC 4532 (Koopmann+ 2008).
Additionally, there are a few grids that were not baselined or flatfielded. This is clearly represented in the name of these grids when downloaded from the NRAO archive, and these grids should be used with caution. These grids are:
- 0724+09
- 0732+01
- 1628+17
- 1636+09
- 1636+29
- 2148+19
Grids contain four axes:
- two spatial axes of 144 pixels.
- a frequency axis in heliocentric frequency units (the grid is defined in frequency -- the spacing is uniform in frequency)
- a polarization axis, with data in two polarizations. For most applications users should average the two polarizations together.
SIMPLE = T / conforms to FITS standard BITPIX = -32 / array data type NAXIS = 4 / number of array dimensions NAXIS1 = 144 NAXIS2 = 144 NAXIS3 = 1024 NAXIS4 = 2
The ALFALFA beam full-width at half maximum (FWHM) is approximately 3.5 arcminutes, and is described in detail in section 3 of Giovanelli et al. 2005.
BMAJ = 0.06333333333333332 / Major axis ALFALFA beam degrees BMIN = 0.055 / Minor axis ALFALFA beam degrees
All observations were conducted over 7 years by members of the ALFALFA survey team. Grids are named by their center position in HHMM+DD. Flux is in mJy/beam.
OBSERVER= 'ALFALFA Survey Team' / Observers OBJECT = '1044+13 ' / ALFALFA data cube and grid center TELESCOP= 'Arecibo ' / Telescope INSTRUME= 'ALFA ' / Receiver instrument BUNIT = 'mJY/BEAM' / Units of flux
Header contains WCS information. However, note that the projection across the grids may not be fully consistent - there may be minor offsets. Also note that ALFALFA pixels were measured from the “center” of the pixel rather than the corners, which may result in 0.5 pixel offsets in some use cases.
EPOCH = 2000.0 / Epoch of RA DEC VELREF = 2 / HELIO RESTFREQ= 1420405751.0 / Rest frequency CTYPE1 = 'RA---CAR' / X-axis type CRVAL1 = 162.2315687471258 / Reference pixel value RA Degrees CDELT1 = -0.01674359374301118 / RA Degrees/pixel CRPIX1 = 0.0 / Reference pixel CROTA1 = 0.0 CTYPE2 = 'DEC--CAR' / Y-axis type CRVAL2 = 11.8 / Reference pixel value Dec Degrees CDELT2 = 0.01666666666666667 / Dec Degrees/pixel CRPIX2 = 0.0 / Reference pixel CROTA2 = 0.0 CTYPE3 = 'FREQ-HEL' / Z-axis type CRVAL3 = 1404.97068847848 / Reference pixel value frequency (MHz) CDELT3 = 0.02441406317211658 / Channel width (MHz) CRPIX3 = 0.0 / Reference pixel CROTA3 = 0.0 CTYPE4 = 'STOKES ' / Pol-axis type CRVAL4 = 1.0 CDELT4 = 1.0 CRPIX4 = 1.0 CROTA4 = 0.0 EQUINOX = 2000.0 / Equinox of coordinates ORIGIN = 'Cornell University' / File creation location SURVEY = 'ALFALFA ' / Survey Name
Weights maps are not statistical weights, but should be thought of as quality flags, or rough estimates of the probability that the data is bad. The value roughly represents the fraction of data flagged, removed, or missing at each pixel/channel, with lower values representing pixels with large amounts of flagged or missing from many different causes. Please note that values are not normalized to 1, nor are they normalized across cubes, and any attempt to use them in this way should be approached with extreme caution.
Continuum files measure the continuum emission at 21cm, and provide a way to measure and normalize, e.g., flux cubes across scales.