astro-legacy-archive/planck-frequency-maps-pr3

Planck PR3 Legacy Frequency Maps

This dataset contains the nine Planck Public Release 3 (2018 Legacy) full-mission, full-channel frequency maps. It serves the source binary tables exactly as released: source column names, units, scalar shapes, row order, dtypes, and value bits are preserved. Self-description in the Parquet metadata resolves the otherwise opaque source identifiers; it does not rename or normalise them.

The LFI products are the R3.00 maps at 30, 44, and 70 GHz. The HFI products are the R3.01 maps at 100, 143, 217, 353, 545, and 857 GHz. Detector, survey, year, half-mission, ring-split, field-only, bandpass-corrected LFI, and 353-psb products are not included. The 353-psb map is a separate polarization-specific product. The HFI release paper reports that including SWBs in the full 353 GHz map raises high-multipole signal-to-noise but produces intensity-to-polarization leakage that dominates the systematic residuals at very low multipoles, especially the quadrupole; it therefore recommends the map built without SWBs for polarization studies.

Release	Planck PR3 / 2018 Legacy
Frequencies	30, 44, 70, 100, 143, 217, 353, 545, 857 GHz
Pixelisation	HEALPix; LFI NSIDE 1024, HFI NSIDE 2048; NESTED
Coordinate card	COORDSYS='GALACTIC'
Polarisation convention	POLCCONV='COSMO'
LFI processing	PROCVER='DX12'
HFI processing	PROCVER='RD12_RC4'

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/planck-frequency-maps-pr3", "HFI_SkyMap_545_2048_R3.01_full", split="train[:1]", columns=["I_STOKES"])
row = dataset[0]
print(row["I_STOKES"])

69.98165130615234

import json
from huggingface_hub import hf_hub_download
import pyarrow.parquet as pq
path = hf_hub_download("astro-legacy-archive/planck-frequency-maps-pr3", "545-ghz.parquet", repo_type="dataset")
print(json.loads(pq.read_schema(path).metadata[b"healpix_map"])["schema"])

astro-legacy-archive/healpix-map

The embedded document records the 545 GHz pixelisation, ordering, coordinate-card value, scalar pixel rule, unit, and coverage. Other configurations retain their source-defined field spellings, including mixed-case LFI names.

Dataset structure

Each PR3 source field is scalar, so each Parquet row is one source HEALPix pixel. LFI configs contain 12,582,912 rows; HFI configs contain 50,331,648. Row i is NESTED pixel i. No pixel index or derived column has been added.

The 30--353 GHz configs contain Stokes I, Q, and U, hit count, and all six elements of the per-pixel IQU covariance matrix. The 545 and 857 GHz configs contain I_STOKES, HITS, and II_COV only.

PR3's source spelling and units differ by instrument and are retained:

configs	signal unit	covariance unit	hit/covariance spelling
LFI 30--70 GHz	K_CMB	(K_CMB)^2	Hits, II_cov through UU_cov
HFI 100--353 GHz	Kcmb	Kcmb^2	HITS, uppercase covariance names
HFI 545/857 GHz	MJy/sr	(Mjy/sr)^2	HITS, II_COV

All nine table headers say COORDSYS='GALACTIC'; that keyword and its table-HDU basis are embedded. Every Arrow field carries its source TUNIT value. Structured physics records the nominal frequency and the release papers' effective-beam FWHM. The LFI configs also carry the source-header RESTFREQ values 28.456, 44.121, and 70.467 GHz with FITS-keyword bases.

There are no Parquet nulls. The PR3 100 GHz source has the BAD_DATA value at one common pixel in each Stokes field. The header declares BAD_DATA=-1.6375e30 with comment HEALPIX bad pixel value; three coverage records retain the source-column ordinals and report one common blank pixel. Its six covariance fields instead contain an exact zero at that pixel. The converter does not collapse this source distinction into nulls. The other eight PR3 headers also declare BAD_DATA, but it occurs zero times in their source fields.

The HFI paper cautions that map monopoles cannot be extracted from Planck data alone. The supplied zero levels include model-dependent CIB and zodiacal monopole contributions; analyses requiring different absolute zero levels must remove or replace those contributions for their particular problem. Their uncertainties are not Planck measurements and differ between components.

Provenance

The authoritative source files are the canonical full-channel maps in the IRSA Planck release-3 map directory. Source URLs, byte counts, SHA-256 hashes, original headers, retrieval date, exact config-to-file mapping, and regeneration command are embedded in every Parquet. Each selected binary table reconstructs in source column and row order with identical TFORM, shape, dtype, and value bits. FITS container serialisation is outside that promise.

The release is identified by ESA as "LFI & HFI PR3 Legacy Frequency Sky Maps", version PR3. Its requested dataset citation is:

European Space Agency, 2018, LFI & HFI PR3 Legacy Frequency Sky Maps, Version PR3, European Space Agency, https://doi.org/10.5270/esa-siyydh3

Licensing and credit

ESA states verbatim:

Data hosted in the ESA Space Science Archives are distributed under the CC BY-NC 3.0 IGO license.

The licence is the Creative Commons Attribution-NonCommercial 3.0 IGO licence. The embedded source credit is: "Planck Collaboration, Planck 2018 results II and III, Astronomy & Astrophysics 641, A2 and A3 (2020), doi:10.1051/0004-6361/201833293 and doi:10.1051/0004-6361/201832909. Data provided by the Planck Collaboration and hosted by IRSA."

Citations

Please cite the PR3 instrument papers:

• Planck Collaboration, "Planck 2018 results. II. Low Frequency Instrument data processing", Astronomy & Astrophysics 641, A2 (2020), doi:10.1051/0004-6361/201833293.
• Planck Collaboration, "Planck 2018 results. III. High Frequency Instrument data processing and frequency maps", Astronomy & Astrophysics 641, A3 (2020), doi:10.1051/0004-6361/201832909.

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 README space's Community tab or the Community tab of any dataset. Confirmed corrections are made in the source conversion tools and published by rebuilding the dataset.