astro-legacy-archive/wmap-masks-9yr

WMAP Nine-Year Analysis Masks

[image]

This dataset contains six analysis and point-source masks from the WMAP DR5 nine-year release. Each config is a full-sky HEALPix mask whose value is 0 for an excluded pixel and 1 for an admitted pixel. The temperature and source masks support temperature analysis; the polarisation masks are intended for Q/U analysis.

Intended uses

Apply the appropriate mask to the WMAP frequency-band maps before computing any quantitative statistic. KQ85 is the standard temperature analysis cut; KQ75 is the extended, more conservative cut. Use the polarisation masks for analyses of the Q and U maps, and choose the source mask according to whether the analysis also requires the Galactic plane and Magellanic cloud regions to be excluded.

How to use

python -m venv .venv && .venv/bin/pip install datasets huggingface_hub pyarrow

from datasets import load_dataset
dataset = load_dataset("astro-legacy-archive/wmap-masks-9yr", "wmap_temperature_kq85_analysis_mask_r9_9yr_v5", split="train")
row = dataset.select([0]).select_columns(["TEMPERATURE"])[0]
print(row["TEMPERATURE"])

0.0

from huggingface_hub import hf_hub_download
import json
import pyarrow.parquet as pq
path = hf_hub_download(repo_id="astro-legacy-archive/wmap-masks-9yr", repo_type="dataset", filename="temperature-kq85-r9.parquet")
print(json.loads(pq.read_metadata(path).metadata[b"healpix_map"])["healpix"]["ordering"])

NESTED

The data example selects one column from one row of the representative KQ85 config; its value is the source flag 0, meaning excluded. Select another named config for its corresponding mask. The healpix_map document embedded in each Parquet schema self-describes its NSIDE, ordering, field units, coverage, provenance, and source headers. Its coverage.codes preserves the source rule for excluded pixels, allowing consumers to interpret flags without loading a complete mask; the separate NSIDE-16 config must be resolved using its own metadata rather than index-aligned with an NSIDE-512 map.

Configs

Each NSIDE-512 config is about 67-143 kB; the NSIDE-16 config is about 21 kB.

Dataset structure

Every config has the same two source columns:

The 0/1 mask flags are stored as float32 with the source's mK unit label; their values are exactly {0, 1}, not temperatures. N_OBS is bit-identical to TEMPERATURE in all six masks, carries no independent information, and can be ignored. In the polarisation masks too, the mask value lives in the column named TEMPERATURE, with the inherited mK unit label.

These configs contain no nulls. Source value 0 remains in-band. coverage.codes records the source FITS header text, "Excluded pixels have the field NOBS set to zero." `TEMPERATURE` is bit-identical to `NOBS; neither column is a null mask. Row i is NESTED pixel i`.

All masks use NSIDE 512 NESTED ordering except polarization-r4, the NSIDE 16 low-resolution companion to polarization-r9. It is at a different resolution and must not be index-aligned against NSIDE 512 maps.

LAMBDA gives the degradation rule: "For this mask, an r4 pixel is included (set to one) if more than 50% of its daughter r9 pixels are set to one."

The maps are in Galactic coordinates. The frame comes from the WMAP Explanatory Supplement.

Choosing a mask

LAMBDA's mask documentation describes the temperature cuts as follows: "For temperature analysis, we provide two cut levels: the standard cut ("temperaturekq85") corresponds roughly to the "Kp2" cut in the 3-year data release; the extended cut ("temperaturekq75") corresponds roughly to the "Kp0" cut in the 3-year data release."

It also states: "Diffuse emission in Stokes I (temperature) is masked using a combination of cuts in the K band and Q band maps, as described in Bennett, et.al., 2013."

For polarisation, "Diffuse emission in Stokes Q and U is masked using cuts in K band polarized intensity combined with a model of thermal dust emission." The source masking differs from the intensity-mask radii below: "An exclusion radius of 1.0 degrees is used for all but one source in the polarization masks. The exception is Cen A, which is masked to a radius of 3 degrees in polarization."

The temperature source mask is a "Temperature-only Point Source Exclusion Mask" and admits 94.67% of the sky. The point source catalog mask additionally excludes the Galactic plane and Magellanic cloud regions, admits 82.79%, and "was used to exclude Galactic plane and Magellanic cloud regions in making the WMAP 9-year point source catalogs".

For the intensity masks, "The source exclusion radius is 0.6 degree for the majority of sources in the intensity masks, with the exception that a radius of 1.2 degrees is used for point sources brighter than 5 Jy in any of the WMAP bands."

Provenance

• —Product page: WMAP nine-year masks
• —Source files follow https://lambda.gsfc.nasa.gov/data/map/dr5/ancillary/masks/<name>.fits.

Other products on the product page are not included here: a Kp8 temperature cleaning mask (wmap_kp8_mask_r9_9yr_v5.fits), per-band processing masks for K/Ka/Q/V/W at r4/r9/r10 (including yr2/notyr2 variants for K), high-resolution r10 variants, and a tar.gz bundle of the analysis masks. The WMAP release at LAMBDA remains the authoritative source.

Provenance, including the original headers, source hashes and URLs, retrieval date, and regeneration command, is embedded in the Parquet metadata under key healpix_map, with schema identifier astro-legacy-archive/healpix-map version 2.0.0. The Parquet field metadata also carries per-column descriptions and units, visible through PyArrow schema inspection; datasets.load_dataset strips field metadata, so Dataset.features shows types only. Each NSIDE-512 file has three row groups; the NSIDE-16 file has one. All remain in source row order.

Each selected bintable reconstructs in source column and row order with identical TFORM, shape, dtype, and value bits. FITS container serialisation is outside that promise.

Considerations for use

• —Mask values are binary flags despite the inherited TEMPERATURE label and mK unit in the source format.
• —Apply a mask before calculating a quantitative statistic from the corresponding band map.
• —KQ75 excludes more of the sky than KQ85.
• —Use a polarisation mask for Q/U analysis.
• —polarization-r4 is the NSIDE 16 low-resolution companion; the other configs use NSIDE 512.

Licensing and credit

HEASARC states verbatim:

HEASARC materials are all available freely for your use.

This is a free-use statement rather than a named formal licence.

WMAP is a NASA mission, and these masks are products of the WMAP Science Team hosted by LAMBDA. No explicit licence text has been published. LAMBDA requests the following acknowledgment:

We acknowledge the use of the Legacy Archive for Microwave Background Data Analysis (LAMBDA), part of the High Energy Astrophysics Science Archive Center (HEASARC). HEASARC/LAMBDA is a service of the Astrophysics Science Division at the NASA Goddard Space Flight Center.

Citations

Please cite the WMAP nine-year final maps and results paper:

@article{Bennett2013,
  author  = {Bennett, C. L. and Larson, D. and Weiland, J. L. and
             Jarosik, N. and Hinshaw, G. and Odegard, N. and
             Smith, K. M. and Hill, R. S. and Gold, B. and
             Halpern, M. and Komatsu, E. and Nolta, M. R. and
             Page, L. and Spergel, D. N. and Wollack, E. and
             Dunkley, J. and Kogut, A. and Limon, M. and
             Meyer, S. S. and Tucker, G. S. and Wright, E. L.},
  title   = {Nine-year Wilkinson Microwave Anisotropy Probe (WMAP)
             Observations: Final Maps and Results},
  journal = {The Astrophysical Journal Supplement Series},
  volume  = {208},
  pages   = {20},
  year    = {2013},
  doi     = {10.1088/0067-0049/208/2/20},
  eprint  = {1212.5225}
}

Errata and requests

To report an erratum, propose a correction, or ask a question about this dataset, open a Discussion or Pull Request in its Community tab; posting requires a free Hugging Face account. Dataset requests and archive-wide questions may be posted in the Community tab of any dataset. Confirmed corrections are made in the source conversion tools and published by rebuilding the dataset.