Get E(B-V) values from SFD dust map data
Switch branches/tags
Nothing to show
Clone or download
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Failed to load latest commit information.
testdata add scaling (with default 0.86), test cutout Oct 29, 2016
.travis.yml allow import with no fits reader installed May 13, 2017


Build Status Coverage Status PyPI

A minimal, fast, MIT-licensed Python module for getting E(B-V) values from Schlegel, Finkbeiner & Davis (1998) dust map FITS files.

>>> import sfdmap

>>> m = sfdmap.SFDMap()

>>> m.ebv(100., 40.)  # Get E(B-V) value at RA=100 degrees, Dec=40 degrees


Requirements: numpy and a FITS reader (either fitsio or astropy).

pip install sfdmap

The FITS files comprising the map must be downloaded separately. Among other locations, they are available from On UNIX systems, run the following to download the maps (93 MB download size):

tar xzf master.tar.gz

A directory "sfddata-master" will be created. Move or rename as you like.

Detailed Usage

Initialize map:

>>> import sfdmap

>>> m = sfdmap.SFDMap('/path/to/dustmap/files')
>>> m = sfdmap.SFDMap()  # get directory from SFD_DIR environment variable

By default, a scaling of 0.86 is applied to the map values to reflect the recalibration by Schlafly & Finkbeiner (2011). To get the original values, use scaling=1.0 when constructing the map:

>>> m = sfdmap.SFDMap(scaling=1.0)

Get E(B-V) value at RA, Dec = 0., 0. (ICRS frame)

>>> m.ebv(0., 0.)

Get E(B-V) at three locations (first argument is RA, second is Dec):

>>> m.ebv([0., 5., 10.], [0., 1.5, 2.1])
array([ 0.03181879,  0.02374864,  0.01746732])

By default the coordinates are assumed to be in degrees in the ICRS coordinate system (e.g., "J2000"). This can be changed with the frame and unit keywords:

>>> m.ebv(1.68140, -1.0504884, frame='galactic', unit='radian')

The dust map values are linearly interpolated by default. Change this with the interpolate keyword:

>>> m.ebv(1.68140, -1.0504884, frame='galactic', unit='radian', interpolate=False)

You can pass an astropy SkyCoord instance:

>>> from astropy.coordinates import SkyCoord

>>> coords = SkyCoord([0., 5., 10.], [0., 1.5, 2.1], frame='icrs', unit='degree')

>>> m.ebv(coords)
array([ 0.03181879,  0.02374864,  0.01746732])

Finally, there is a convenience function in the module so that you don't have to construct a SFDMap instance if you just want to query the map once:

>>> sfdmap.ebv(0., 0.)  # get map directory from SFD_DIR environment variable

>>> sfdmap.ebv(0., 0., mapdir='/path/to/dust/files')

How do I get extinction at a specific wavelength or in a specific filter?

The E(B-V) values from the map give information about the amplitude of dust extinction in a given direction. To get the extinction at a given wavelength or through a given filter, one needs information about the relative extinction between different wavelengths: an "extinction law". One can use the extinction package for this. For example, the following code gets the extinction in magnitudes at RA, Dec = (0., 0.) and wavelengths (4000, 5000):

>>> import extinction

>>> ebv = m.ebv(0., 0.)

>>> wave = np.array([4000., 5000.])

>>> extinction.fitzpatrick99(wave, 3.1 * ebv)
array([ 0.12074424,  0.09513746])

To get the extinction in a given bandpass, one needs to know the source spectrum, as different wavelengths in the bandpass will have different extinction. With an assumed source spectrum, one would integrate the source spectrum with extinction applied through the bandpass and then compare to the unextincted spectrum integrated through the same bandpass. Some papers provide extinction values in a number of common bandpasses; these assume some source spectrum.

Performance Note

Note that while passing an astropy SkyCoord object works, if you have coordinates in the ICRS or FK5 (epoch 2000) (e.g., J2000) systems, it is far faster to pass latitute and longitude directly. This is particularly true for small numbers of coordinates or scalar coordinates:

In [1]: from astropy.coordinates import SkyCoord

In [2]: import sfdmap

In [3]: m = sfdmap.SFDMap()

In [4]: m.ebv(0., 0.)  # evaluate once to trigger reading the FITS file
Out[4]: 0.03181878852100873

In [5]: coord = SkyCoord(0., 0., unit='degree')

In [6]: %timeit m.ebv(coord)  # time with SkyCoord object
100 loops, best of 3: 18.1 ms per loop

In [7]: %timeit m.ebv(0., 0., unit='degree')  # pass ra, dec directly
10000 loops, best of 3: 80 µs per loop


There are a couple other packages that support getting dust values from this map. Both these packages have a bigger scope than this one and include several other 2-d or 3-d galactic dust maps. Check them out if you want to compare between different maps or need 3-d maps. Below, I note a few relevant differences from this package.

mwdust [docs]

mwdust.SFD gives the extinction in a given band rather than E(B-V). The API is geared towards 3-d maps, so a distance must be given. Python 3 is not currently supported. The license is BSD.

dustmaps [docs]

dustmaps.sfd.SFDQuery uses astropy's SkyCoord for all coordinate conversions and therefore has suboptimal performance for small numbers of coordinates (see "Performance Note" above). The license is GPLv2.