astro-legacy-archive/dwingeloo-820mhz

Dwingeloo 820 MHz Northern-Sky Survey

[image]

Berkhuijsen's 820 MHz continuum survey covers the northern sky with the Dwingeloo 25 m telescope. CADE rendered the original WCS map onto HEALPix by the drizzling method of Paradis et al. 2012, Appendix A; NASA LAMBDA mirrors that file. The observations were made in October-November 1965, July 1966, and March-June 1967.

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/dwingeloo-820mhz", "Dwingeloo_Kelvins_1_256", split="train")
row = dataset.select([0]).select_columns(["UNKNOWN1"])[0]
print(row["UNKNOWN1"][0])

34.73250198364258

from huggingface_hub import hf_hub_download
import json
import pyarrow.parquet as pq
path = hf_hub_download(repo_id="astro-legacy-archive/dwingeloo-820mhz", repo_type="dataset", filename="dwingeloo-820mhz.parquet")
print(json.loads(pq.read_metadata(path).metadata[b"healpix_map"])["healpix"]["ordering"])

NESTED

The data example selects the first element of the first source row. The healpix_map document embedded in the Parquet schema self-describes the NSIDE, ordering, row/element pixel rule, field unit, coverage, provenance, and source headers. To form a validity mask, compare flattened UNKNOWN1 elements with the measured value in coverage.codes; 449,343 pixels pass and their measured mean is 8.1308975 K. Resolve requested coordinates using the embedded geometry and the documented row/element rule; at Galactic longitude 180.0 and latitude 0.0 degrees the value is 9.268175 K.

The map is about 2.1 MB and has 768 source rows. Each UNKNOWN1 value is a fixed-size list of 1,024 consecutive pixels, giving 786,432 pixels.

UNKNOWN1 is the placeholder column name in the CADE-distributed FITS file, commented "unknown content" in the header; the column description carries its physical meaning. Valid brightness temperatures span 4.90 to 102.98 K.

The source card declares OBJECT = 'FULLSKY' / Sky coverage, either FULLSKY or PARTIAL; measured coverage is 57.14%, while 42.86% of pixels carry -32768.0. No meaning is assigned to that declaration beyond its own words. The survey nominally spans declinations -7 to +85 degrees. Measured against that band, 336,298 coded pixels lie south of declination -7, 791 lie north of +85, and none lies inside the band: every pixel whose centre is between -7 and +85 carries data. Gridding extends valid edge pixels to declination -9.06 in the south and +89.90 in the north. Of 1,502 pixels north of +85, 711 carry data. The code patch near Galactic (l, b) = (123, +27) coincides with the north celestial pole and the nominal +85-degree survey boundary.

The exact value -32768.0 occurs 337,089 times and remains in-band here. A naive mean includes those values; the example reads the measured code from coverage.codes and excludes it before calculating statistics. No archive documents a meaning for this value. It lies outside DATAMIN/DATAMAX, which bracket exactly the extrema of the other values, and every occurrence lies outside the nominal declination band; these are measured correlations only. Several header numbers are strings: BMAJ = '1.2', DATAMIN = '4.9016538', and DATAMAX = '102.98037'. RESTFREQ = '0.82' records the nominal observing frequency in a rest-frequency keyword.

Random per-pixel error is +/-0.2 K; systematic differences between map parts are below 0.3 K. The absolute zero-level standard error is +/-0.6 K, conditional on the theoretical zenith extinction, while the temperature scale is known to better than 6%.

The paper describes an absolute-brightness survey corrected for antenna sidelobe radiation. The 1.2-degree effective resolution belongs to the published, interpolated map; it must not be confused with the telescope's 1.0-degree HPBW. The FITS header and both archive pages correctly use 1.2 degrees for the product served here.

Considerations for use

• —Exclude -32768.0 rather than using OBJECT = 'FULLSKY' when selecting coverage. The map is a northern survey on a full-sky pixel grid.
• —Keep the additive random, inter-region systematic and zero-level terms separate from the multiplicative temperature-scale uncertainty.
• —The preview's north-pole hole is expected survey coverage, not a corrupt patch.
• —NSIDE-256 pixels are about 0.23 degrees across, much finer than the 1.2-degree map resolution, so neighbouring pixels are strongly correlated.
• —The survey's polarisation tables are not part of this product.
• —Values use the radio brightness-temperature convention.
• —The source sentinel was established by measurement; no archive documents -32768.0 for this product.

Provenance

• —LAMBDA product page
• —Pinned LAMBDA FITS, 3,156,480 bytes, SHA-256 6d0b1a8e438b6875e3d8d2142db006c27096bc7308fc8eb06d7a9aca8a77e96d.
• —CADE origin, SHA-256 a35f33ded848e0ccfe171c2eae1a6a8c42702a6c10ae44c8170e17b08abca462 when checked 2026-08-29.

Their data arrays are bit-identical. LAMBDA is pinned because CADE rewrites headers at its fixed URL. The files have the same 3,156,480-byte length and identical bintable arrays; their 1,280 differing bytes are confined to offsets 5600-7362 in the second header block, before the data at byte 8640. Relative to LAMBDA, CADE added AUTHOR = 'D. Paradis' and two HISTORY lines documenting that metadata edit. The pinned data match the corrected map served by CADE on 2026-08-29; CADE publishes no version identifier or changelog for its notice, "A bug has been corrected !! Please download this new map :". The original WCS data remain at the MPIfR Survey Sampler.

The source identifies its ordering as NESTED. The embedded pixel_rule specifies how row and list position identify a pixel. Orderings and conventions differ between datasets; read the embedded ordering and pixel_rule. Provenance (the complete headers of the pinned LAMBDA file, source URLs and hashes, retrieval date, and the regeneration command) is embedded in the Parquet file. The differing CADE-origin header is recorded by hash, not embedded. Rebuild with python tools/make_dwingeloo.py.

Licensing and citations

HEASARC states verbatim:

HEASARC materials are all available freely for your use.

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

No explicit licence is published by the authors, CADE, or LAMBDA. CADE requests:

We acknowledge the use of data provided by the Centre d'Analyse de Données Etendues (CADE), a service of IRAP-UPS/CNRS (https://cade.irap.omp.eu/, Paradis et al., 2012, A&A, 543, 103).

The correct article identifier is A&A 543, A103. Cite Berkhuijsen 1972, A&AS 5, 263, for the data; Berkhuijsen 1971, A&A 14, 359, for the discussion; and Paradis et al. 2012, A&A 543, A103, for the rendering. The 1972 survey paper predates arXiv, and ADS records no DOI for it; no eprint identifier is asserted for it.

LAMBDA requests:

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.

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.