Our PS1-based map of reddening over the entire sky north of declination -30 degrees is available as a FITS file.

The map is provided as a binary FITS table, ordered according to the HEALPix scheme. Each record in the array has the following fields

estimated E(B-V) at 4.5 kpc; see below for information on the overall scale
1 sigma formal uncertainty estimate, based on when chi^2 changes by 1
1 sigma formal uncertainty estimate, based on when chi^2 changes by 4
integer indicating the status of our analysis of this pixel. The following values are possible
Value Meaning
-999 no PS1 data in this pixel
-2 analysis failed to converge in this pixel
-1 interpolated value from adjacent pixels
1 analysis converged
number of PS1 stars used in the analysis (stars failing to fit the model are clipped)
number of PS1 stars with good photometry in the pixel
resolution (HEALPix nside) of the map in this pixel
Here is sample IDL code to read the map at positions l and b (without interpolating):
map = mrdfits('path/to/ps1-ebv-4.5kpc.fits', 1)
ang2pix_ring, 512, (90.-b)/!radeg, l/!radeg, pix
ebv = map[pix].ebv
or with healpy and pyfits in python (interpolating):
map = pyfits.getdata('path/to/ps1-ebv-4.5kpc.fits')
ebv = healpy.get_interp_val(map['ebv'], (90.-b)*numpy.pi/180., l*numpy.pi/180.)

The overall scaling of E(B-V) is somewhat subtle. We use measurements of the colors of stars in the PS1 bands, and infer E(B-V) in the context of a reddening vector that relates reddening E(B-V) to reddenings (and extinction) in the different bands. For this we adopt the coefficients of Schlafly & Finkbeiner (2011), who assume a Fitzpatrick (1999) reddening law, but additionally solve for an overall scaling which attempts to fix the overall reddening calibration of the Schlegel, Finkbeiner, and Davis (1998) map. By adopting the coefficients of Schlafly & Finkbeiner (2011), we are essentially putting our map on the same scale as Schlegel, Finkbeiner, and Davis (1998). This has the advantage that it means that the two maps can be directly compared, and that the coefficients of Schlafly & Finkbeiner (2011) can be directly used to convert from our E(B-V) estimates into other bands. It has the disadvantage that our E(B-V) values must be themselves converted to E(B-V), admittedly very confusingly.

More than 30 degrees from the Galactic plane, we use pixels with half the resolution of the lower Galactic latitude sky. To accomodate this in the FITS file, we fill in 4 identical records for each larger pixel, and mark those pixels as having a HEALPix nside of 256, as opposed to the usual 512.

We estimate that there is an additional 30 mmag systematic uncertainty in our data to due imperfections in our modeling of the Galaxy's stars; see the paper for details.