diskatlas: all-sky interactive atlas for resolved circumstellar disks and imaged companions

diskatlas: 星周盘图像与直接成像伴体 · 交互式全天图

🌐 Browse it live: https://seawander.github.io/diskatlas/ · Source & data on GitHub — or clone and double-click index.html for the fully offline version.

An offline-capable, interactive all-sky atlas of every system with a resolved circumstellar disk (mm/submm/cm interferometry, high-contrast scattered light / thermal IR, and resolved far-IR single-dish imaging), a directly imaged planet or substellar companion, or a coronagraphically imaged quasar host. Click any object on the sky map to browse its images across instruments, wavelengths and epochs, with clickable arXiv / SciX citations for every image.

双击根目录的 index.html 即可离线使用。 Double-click index.html — no server, no internet needed.

What's inside

621 systems · 2854 image records, every one a panel cropped from a peer-reviewed figure with full citations; coordinates for every system; a companion catalogue with per-object status (confirmed / candidate / disputed / refuted). python3 backend-data/build.py prints the exact live counts — trust it over any number in prose. Session-by-session history lives in docs/HISTORY.md.

The viewer offers three views — Sky map, Coverage matrix, and Tonight (an observability planner with per-target airmass.org night-chart links and CSV export) — with faceted filters (wavelength band, facility, instrument, missing modality), a light/dark theme, and a 12-language UI (including right-to-left Arabic and translated constellation names). Everything runs client-side from file:// — no server, no build step, no CDN. See frontend/README.md for the full feature map.

Contributing

Contributions are very welcome — new systems, missing instruments/epochs, coordinate or citation fixes, translations, frontend improvements — whether you edit by hand or drive an AI agent on your own machine (most of the atlas was built that way). Fork, make your change, and open a pull request; to just suggest a paper or target, open an issue.

• Read CONTRIBUTING.md for the full recipe (human and agent-driven).
• Using an agent? AGENTS.md (and CLAUDE.md for Claude Code) auto-loads in most harnesses and orients it the moment you open the repo.

Two rules for any PR: python3 backend-data/validate.py must report 0 errors, and never hand-edit the generated frontend/data.js (rebuild with python3 backend-data/build.py).

Repository layout

index.html          ← THE app. Open this. (loads frontend/ + images/)
frontend/           ← app.js, style.css, data.js (data.js is GENERATED — do not hand-edit)
data/               ← the database: one JSON per system + docs. THIS is what you edit.
backend-data/       ← Python pipeline: seed → coordinates → crop → validate → build
images/             ← _sources/ (downloaded figures) + <system_id>/*.png (cropped panels)

Every folder has its own README.md. Agents: CLAUDE.md / AGENTS.md route you to the one doc your task needs; HANDOFF.md is the complete handbook.

Scope / 收录标准

• Disks: spatially resolved (≥ a few beams / PSFs) circumstellar disks — protoplanetary (incl. transition, edge-on, eruptive-star and Class 0/I embedded) and debris — imaged by mm/submm/cm interferometers (ALMA, SMA, PdBI/NOEMA, VLA, CARMA, OVRO, BIMA, ATCA), in scattered light / thermal IR by high-contrast instruments (HST, JWST, GPI, SPHERE, Subaru HiCIAO/SCExAO, NACO, MagAO(-X), LBTI, …), by VLTI interferometric reconstruction, and in the resolved far-IR (Herschel/Spitzer/JCMT/CSO). Orion proplyd silhouettes included.
• Companions: directly imaged planets and brown-dwarf companions (status confirmed/candidate/disputed/refuted tracked per companion; a few borderline or later-refuted objects are kept for the historical record and flagged).
• Quasar hosts: coronagraphic / PSF-subtracted host-galaxy imaging (redshift instead of dist_pc).
• Resolved far-IR/submm single-dish images (Herschel PACS/SPIRE, Spitzer MIPS, JCMT, CSO) are in scope; marginally resolved single-beam detections and unresolved excesses remain excluded.
• Press-release composites are not used; every image is cropped from a peer-reviewed figure and linked to its paper (arXiv + SciX).

License / 许可

Three parts, three answers:

• Code (index.html, frontend/ JS+CSS, backend-data/ Python): MIT — see LICENSE.
• Compiled database (data/systems/*.json, the metadata in generated frontend/data.js): CC BY 4.0 — see LICENSE-DATA.md. Cite this repository (and the accompanying paper, once published) when reusing.
• Images (images/**/*.png): not covered by either license. They are low-resolution crops of figures from peer-reviewed papers — cropped from the authors' own arXiv source files (preprint figures), not from the journal-typeset articles. Rights remain © the original authors (and the journals for the published versions — AAS, A&A, MNRAS, …); each crop is reproduced here with a per-image credit line and full citation for scholarly reference. Follow each record's links (arXiv / SciX / DOI) for the originals and for any reuse permissions.

Disclaimer & fair-use statement / 免责声明与合理使用

diskatlas is a non-commercial, not-for-profit resource created solely for scientific research and education. The image excerpts are small, reduced-resolution crops taken from the authors' own arXiv source files (preprint figures) — not from the journals' typeset versions of record — to help researchers locate and navigate the primary literature. Every excerpt carries a credit line and links back to the original paper (arXiv / SciX / DOI), and is not a substitute for the published article. We believe this constitutes fair use (17 U.S.C. § 107) / fair dealing for purposes of research and scholarship.

We have no intention to infringe any copyright. All rights in the original figures remain with their authors (and, for the versions of record, their publishers). If you are a rights holder and would prefer that a particular image not appear here, please open an issue and it will be removed promptly.