astro-legacy-archive/dirbe-zsma

COBE/DIRBE Zodi-Subtracted Mission Average maps

[image]

This dataset contains the ten COBE/DIRBE Zodi-Subtracted Mission Average maps served by LAMBDA at 1.25, 2.2, 3.5, 4.9, 12, 25, 60, 100, 140, and 240 microns. Each configuration is one source FITS table with 393,216 rows in the COBE resolution-9 quadrilateralized spherical-cube pixelisation.

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

from datasets import load_dataset
dataset = load_dataset("astro-legacy-archive/dirbe-zsma", "DIRBE_BAND1A_ZSMA", split="train[:1]", columns=["Pixel_no", "Resid"])
row = dataset[0]
print(row["Pixel_no"], row["Resid"])

0 0.4949333071708679

import json
from huggingface_hub import hf_hub_download
import pyarrow.parquet as pq
path = hf_hub_download("astro-legacy-archive/dirbe-zsma", "DIRBE_BAND1A_ZSMA.parquet", repo_type="dataset")
print(json.loads(pq.read_metadata(path).metadata[b"explicit_pixel_table"])["pixelisation"]["scheme"]["resolution"]["value"])

9

Each table retains the source columns Pixel_no, Resid, NumObs, NumWks, and StdDev in source order. Resid and StdDev use MJy/sr; the other fields have blank source TUNIT values. Pixel_no runs from 0 through 393,215. No source-documented replacement code occurs in the pinned files.

Product and table structure

LAMBDA describes ZSMA construction as subtracting the DIRBE Sky and Zodi Atlas zodiacal-light estimate from each weekly map before averaging the residual intensities. The primary headers identify Pass 3B processing, the full sky, equinox 2000, observations from 1989-11-12 through 1990-09-21, solar elongations from 68 to 120 degrees, combined approach vectors, and IPD model FD6. These statements remain source metadata rather than new table columns.

Every configuration has the same five scalar field shapes and row width of 17 bytes. Resid is the unweighted mean residual intensity after IPD-model removal. NumObs is the number of contributing 1/8-second observations; NumWks is the number of weekly residuals used in that mean; and StdDev is the standard deviation of the mean computed from the weekly residuals. The source states that its quoted intensity assumes nu I(nu) is constant and that other spectra require colour correction.

The source projection has six 256 by 256 faces. Its approximately equal-area pixels are equivalent to 0.32-degree squares, while the instantaneous DIRBE beam is 0.7 by 0.7 degrees. Pixel identities are the source Pixel_no values; coordinate tables published separately by COBE can relate their centres to ecliptic, Galactic, and equatorial coordinates.

The ten bands have 3,932,160 rows in total. Every numeric value is finite. The residual ranges extend from 0.0256427769 to 215.8180542 MJy/sr at 1.25 microns and from -6.1458148956 to 6781.1699219 MJy/sr at 240 microns. The widest measured range is in the 140-micron map, from -10.9704360962 to 12269.9726562 MJy/sr. StdDev includes a small number of genuine zero values in every band. None is assigned a special-code meaning.

Each source file is 6,701,760 bytes. Band-specific wavelength keywords and product identifiers remain in the individual embedded headers; measured ranges and observation-count distributions are recorded in facts.yaml. Loading one configuration reads one released band and does not materialise or join the other nine.

Provenance

The embedded explicit_pixel_table document contains both FITS headers, the source pin, column definitions, pixelisation statements and their bases, credit, and regeneration command.

Source product: LAMBDA COBE/DIRBE ZSMA. The ten FITS sources total 67,017,600 bytes. Their individual SHA-256 pins are stored in each Parquet file and in facts.yaml. Rebuild with python tools/make_dirbe_zsma.py data/dirbe-zsma.

Configuration names are source filename stems, including the source's different 1A, 2A, 3A, and zero-padded 04–09 spellings. A rebuild checks every source byte count and SHA-256 before conversion, validates the self-description, and compares the Parquet values with the FITS binary-table bytes. The comparison covers TTYPE and TUNIT identity, TFORM numeric width, scalar shape, value bits, column order, and row order. It does not claim that the reconstructed FITS container has identical cards, padding, or checksum bytes.

Considerations for use

ZSMA is a zodiacal-light-subtracted mission average, not the individual weekly observations or the DIRBE Sky and Zodi Atlas. Its residuals still contain Galactic and extragalactic sky emission and any residual foreground or instrumental effects. The beam is broader than a resolution-9 pixel, and neighbouring table rows should not be interpreted as independent beam-resolution measurements. Counts and standard deviations vary with sky position because the weekly and 1/8-second observation coverage varies.

No explicit data licence was found on the checked LAMBDA pages. LAMBDA's COBE credit is: “The COBE datasets were developed by the NASA Goddard Space Flight Center under the guidance of the COBE Science Working Group.” Cite Hauser et al. (1998), doi:10.1086/306379, and Kelsall et al. (1998), doi:10.1086/306380.

Hauser et al. describe the DIRBE search for the cosmic infrared background; Kelsall et al. describe the interplanetary-dust model subtracted in this product. Analyses should retain the source band convention and apply the appropriate published colour correction when the assumed spectrum is not nu I(nu) constant.

Licence

HEASARC states verbatim:

HEASARC materials are all available freely for your use.

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

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.