The Atlas · Observatory Data Output · v1 · 2026-06-30

The Data Deluge, observatory by observatory

Every major observatory placed by when it ran (horizontal axis) and how much data it pours out per day (vertical, logarithmic). Each bar spans the facility's operational lifetime; its height is the average daily science-data output. Colour = waveband. Solid bars are operating now, faded bars are retired (the card gives their lifetime total), dashed bars are funded-but-future. The floor-to-ceiling climb — from a space telescope's megabytes-a-day to Rubin's 20 TB a night to the SKA's ~2 PB a day — is nine orders of magnitude in fifty years.

"Storage is cheap. Attention is not. This atlas is the supply curve."

Status

Operating now
Retired (card → lifetime total)
Future / under construction

Waveband

Click any bar above to see its full data record — daily output, operational span, lifetime total, waveband, and the archive that holds it.

How to read this atlas

The horizontal axis is calendar time, 1975 → 2045. The vertical axis is the average science-data output per day, in log10 bytes — each gridline is a factor of one thousand, running from ~10 kB/day at the floor to ~10 PB/day at the ceiling. A facility is drawn as a bar from the year it began returning science data to the year it stopped (or, for current instruments, to today; for future ones, to its planned start, dashed).

The point of plotting it this way is the slope. The earliest entries — IUE, IRAS, Hipparcos — sit near the floor, returning a few megabytes a day. Hubble lifts the line to ~10 GB/day in 1990. The wide-field surveys of the 2010s — Pan-STARRS, ZTF, LOFAR — punch into the terabytes-per-day band. Then Rubin (20 TB/night), Roman (~1.4 TB/day), and the SKA (~2 PB/day of archived science data) carry it into the petascale. The vertical gap between the 1980s and the 2030s is roughly nine orders of magnitude.

What "data output" means here

The metric is average science-data volume per day — the rate that lands in the public archive, not the raw rate at the detector. Many radio and gamma facilities digitise vastly more than they keep: the SKA correlator handles ~1 TB/s and CTAO's cameras generate hundreds of PB/year, but both reduce on-site to a small fraction before archiving. The bar height is the archived rate, because that is the data astronomers actually receive.
  • Daily rate = (downlink or archive growth per year) ÷ 365, or the published per-night / per-contact figure averaged over the duty cycle.
  • Lifetime total (shown for retired facilities in the detail card) = the full archive size at end of mission, where published; otherwise rate × operational days.
  • Confidence is flagged per entry. "Measured" = a published downlink/archive figure. "Estimate" = an order-of-magnitude reconstruction from instrument specs and duty cycle — treat the bar's vertical position as good to a factor of a few, not a decimal.

Raw correlator and camera rates, where dramatically higher than the archived rate, are noted in each facility's record so the reduction factor is visible.

References & data sources

Daily rates are compiled from mission documentation, observatory data-management papers, and archive manuals. Each bar's record links to its primary source where available.

  • Rubin / LSST — SLAC & Rubin Observatory: ~20 TB/night raw, 60 PB raw / 500 PB processed over the survey.
  • SKAO — skao.int "Big Data": ~710 PB/yr archived science data; ~8 Tbit/s; ~0.5–1 TB/s at the correlator.
  • JWST — STScI JWST User Documentation: ~57 GB/day downlink (two 4-h contacts).
  • Roman — STScI / NASA SVS: ~11 Tbit/day (~1.4 TB/day), ~20 PB over the primary mission.
  • Gaia — ESA / Gaia mission: ~40 GB/day compressed downlink; ~60 TB compressed / ~200 TB uncompressed lifetime.
  • HST — NASA / ESA Hubble: ~10–15 GB/day; >153 TB archive (2018).
  • ALMA — NRAO / JAO archive manuals & NAASC memos: ~300–500 TB/yr archived.
  • LOFAR — ASTRON LTA: ~5–7 PB/yr ingest; >40 PB archive.
  • Euclid — ESA / ESOC: ~100 GB/day compressed over the 6-yr mission.
  • CTAO — cta-observatory.org: hundreds of PB/yr raw → ~few PB/yr archived → few TB/yr science-ready.
  • Pan-STARRS / ZTF — IPAC & STScI: ~4 TB/night each; PS1 archive 1.6 PB.
  • Fermi / Chandra / TESS / Kepler / Euclid / eROSITA / SPHEREx — respective mission operations documentation.
  • MeerKAT / VLA / ngVLA — NRAO & SARAO: correlator output 2.2 Tbit/s (MeerKAT), 6.4 TB/s (ngVLA); archived rates are far lower.
Atlas · Observatory Data Output · v1 · 2026-06-30 · self-contained, no external data fetches · rates curated, not live-refreshed