Meteorologist, oceanographer, Python solvem probler, experimenter, Raspberry Pi tinkerer, geek dad, manager, space enthusiast, ham radio lover, science columnist, ADHDer, synthetic intellect curious, linux admin cosplayer.

Image Research Round 2: Spectral Era Supercomputers (1980-2019)

Research compiled 2026-05-13 for the Bourke / spectral-methods NWP post. This round complements IMAGES_BOURKE_SPECTRAL.md (2026-05-12) which already located Cray-1, Cray X-MP, and spherical-harmonic / Gaussian-grid concept images.

The repo already carries (from earlier posts) several Cray photos that all show the same one or two machines: Cray-1_Computer_Museum_of_America.jpg, Cray_1_EPFL.jpg, Cray_1_museum.jpg, Cray_X-MP_Linkoping.jpeg, Cray_prototype_1957.jpg, header-cray.jpg, header-cray2.jpg. Cray-1 and Cray X-MP variety is already covered. Round 2 focuses on different machines that the spectral GFS / Met Office spectral / ECMWF spectral models ran on between 1980 and 2019, plus institutional shots.

All licences verified by direct fetch of the Wikimedia source page on 2026-05-13.

Summary of negative findings (re-confirmed)

  • No CDC Cyber 205 photograph exists on Wikimedia Commons. Broader queries (Cyber 205 with quotes, Cyber 200, Met Office Bracknell + Cyber, FSU + Cyber) all return nothing. The only Cyber 205 images on the open web are at Google Arts & Culture (via the Met Office archive) and at the Science Museum Group; neither is CC-licensed. The Science Museum Group’s CC0 declaration covers metadata only, not images. Do not use.
  • No CC-licensed image of Joseph Sela exists. NCEP / NOAA / EMC sites carry text obituaries and a memorial seminar announcement but no free-licensed portrait.
  • No clean CC portrait of Brian Hoskins exists. A 2013 Indian government ceremony photograph under GODL-India shows him as a side figure in a group shot at the Indian Ministry of Earth Sciences. Not usable as a portrait.
  • No Adrian Simmons free portrait on Wikimedia Commons.
  • NOAA Digital Library (formerly NOAA Photo Library) is currently 403-locked to WebFetch and the search results within Wikimedia Commons cover only a subset of NOAA institutional photos.

Candidate A: Cray Y-190A at NASA Ames (1990) — replaces Cyber 205 gap

  • Source page: https://commons.wikimedia.org/wiki/File:Cray_Y_190A_Supercomputer_-_GPN-2000-001635.jpg
  • Direct file URL: https://upload.wikimedia.org/wikipedia/commons/9/92/Cray_Y_190A_Supercomputer_-_GPN-2000-001635.jpg
  • License: Public domain (PD-USGov-NASA - NASA Ames Research Center, photo ID GPN-2000-001635, originally AC90-0121-8)
  • Author: NASA (uncredited photographer)
  • Date: 1 March 1990
  • Dimensions: 2,884 x 2,268 px, 7.18 MB, landscape orientation
  • Depicts: Cray Y-190A vector supercomputer at NASA Ames, Mountain View CA. The Y-190A is the Y-MP/8 air-cooled variant; functionally identical to the vector machines that ECMWF and NCEP ran their spectral models on through the early 1990s.
  • Suggested filename: Cray_Y-MP_NASA_Ames_1990.jpg
  • Suggested caption (Polish): “Cray Y-MP w NASA Ames, marzec 1990. Wektorowa maszyna nastepnej generacji po Crayu X-MP - na takich superkomputerach operacyjne osrodki prognozy pogody (NCEP, ECMWF) liczyly swoje modele spektralne w pierwszej polowie lat 90. NASA, public domain, via Wikimedia Commons.”
  • Where in the post: Mid-section on 1990s operational spectral models, after the Cray X-MP discussion. A second machine in the same lineage shows that spectral GFS migrated through a succession of vector Crays, not one.

Candidate B: Cray C90 8-CPU at Computer Museum of America (CC BY-SA)

  • Source page: https://commons.wikimedia.org/wiki/File:Computer_Museum_of_America_(27).jpg
  • Direct file URL: https://upload.wikimedia.org/wikipedia/commons/7/76/Computer_Museum_of_America_%2827%29.jpg
  • License: CC BY-SA 3.0
  • Author: Jud McCranie (Wikimedia: Bubba73)
  • Date: 11 August 2019
  • Dimensions: 4,016 x 6,016 px, 18.22 MB (portrait orientation)
  • Depicts: Cray C90 8-CPU machine that ran as rain in the Cray Research data centre; now at the Computer Museum of America, Roswell GA.
  • Suggested filename: Cray_C90_Computer_Museum_of_America.jpg
  • Suggested caption (Polish): “Cray C90 z osmioma procesorami - w sluzbie ‘rain’ w centrum danych Cray Research, dzis w Computer Museum of America. To na takich maszynach ECMWF i Met Office liczyly modele spektralne w pierwszej polowie lat 90. Jud McCranie, CC BY-SA 3.0, via Wikimedia Commons.”
  • Where in the post: Pairs with Candidate A or replaces it. C90 was the follow-on to Y-MP and the actual ECMWF operational machine 1994-1996.
  • Note: Portrait orientation works well for a sidebar / inline-text layout.

Candidate C: Cray Y-MP at NASA Goddard NCCS (1993) — PD-USGov-NASA

  • Source page: https://commons.wikimedia.org/wiki/File:Cray_Y-MP_GSFC.jpg
  • Direct file URL: https://upload.wikimedia.org/wikipedia/commons/e/ec/Cray_Y-MP_GSFC.jpg
  • License: Public domain (NASA work, PD-USGov)
  • Author: Dave Pape (Davepape, NASA employee)
  • Date: January 1993
  • Depicts: The distinctive grey-and-red-trim Cray Y-MP installed at the NASA Goddard Space Flight Center, NCCS branch. This is the iconic 1993-vintage Y-MP form factor with the matching tape cabinets in the background.
  • Suggested filename: Cray_Y-MP_NASA_Goddard_1993.jpg
  • Suggested caption (Polish): “Cray Y-MP w NASA Goddard, styczen 1993. Charakterystyczna obudowa - szare panele z czerwona obwoluta - przedstawia maszyne tej samej rodziny, ktora NCEP uzywal do operacyjnego modelu spektralnego MRF/AVN przez pierwsza polowe lat 90. Dave Pape, NASA, public domain, via Wikimedia Commons.”
  • Where in the post: Alternative to Candidate A. Goddard NCCS is closer to the weather/climate use case than Ames; same machine generation, slightly more “weather-coded” institution.
  • Note: Choose one of A or C, not both. A is later (1990) and from a more iconic NASA archival series; C is closer to the climate-science world. The user’s audience may know NASA Goddard better as a weather/climate institution.

Candidate D: NOAA IBM supercomputer (2009) — actual NCEP spectral-GFS hardware

  • Source page: https://commons.wikimedia.org/wiki/File:NOAA_Weather_supercomputer.jpg
  • Direct file URL: https://upload.wikimedia.org/wikipedia/commons/5/5e/NOAA_Weather_supercomputer.jpg
  • License: CC BY 2.0
  • Author: NOAA Photo Library / NOAA’s National Weather Service (NWS) Collection, Image ID wea01799
  • Date: 8 September 2009 (uploaded to Flickr 4 September 2013)
  • Depicts: Rows of IBM Power-based supercomputer racks used by NCEP for weather and climate forecasts. This is the era when NCEP was running its spectral GFS on IBM Power clusters at Gaithersburg / Fairmont (West Virginia backup) - “Stratus” and “Cirrus” went live in August 2009, so the photo almost certainly shows one of those two machines.
  • Suggested filename: NOAA_IBM_supercomputer_2009.jpg
  • Suggested caption (Polish): “Superkomputery IBM Power w NCEP, wrzesien 2009 - ‘Stratus’ i ‘Cirrus’ weszly do sluzby operacyjnej w sierpniu 2009 roku. Wlasnie na takich klastrach spektralny GFS dzialal w pozniejszej fazie - rownolegle, na tysiacach procesorow, daleko od wektorowych Crayow z poczatkow. NOAA Photo Library, CC BY 2.0, via Wikimedia Commons.”
  • Where in the post: The “late spectral era” image. Bridges from vector Crays (1980s-1990s) to the parallel IBM clusters that ran spectral GFS until
    1. This is the only verified image showing the actual NCEP operational iron.

Candidate E: Cray XC40 “Hazel Hen” at HLRS Stuttgart (2015) — final spectral-era architecture

  • Source page: https://commons.wikimedia.org/wiki/File:High_Performance_Computing_Center_Stuttgart_HLRS_2015_09_Cray_XC40_Hazel_Hen.jpg
  • Direct file URL: https://upload.wikimedia.org/wikipedia/commons/0/09/High_Performance_Computing_Center_Stuttgart_HLRS_2015_09_Cray_XC40_Hazel_Hen.jpg
  • License: GFDL 1.2+ and CC BY 4.0 (dual)
  • Author: Julian Herzog
  • Date: 18 December 2015
  • Depicts: Full view of the Cray XC40 system “Hazel Hen” at the High Performance Computing Center Stuttgart (HLRS). The XC40 is the same Cray architecture (Aries interconnect, Xeon nodes) that NCEP ran spectral GFS on as Luna and Surge (the WCOSS Cray Phase 2 machines, 2015-2019).
  • Suggested filename: Cray_XC40_Hazel_Hen_HLRS.jpg
  • Suggested caption (Polish): “Cray XC40 ‘Hazel Hen’ w HLRS Stuttgart, 2015. Ta sama architektura - XC40 z interkonektem Aries - co maszyny NCEP ‘Luna’ i ‘Surge’, na ktorych spektralny GFS dzialal w ostatniej fazie, do czerwca 2019. Julian Herzog, CC BY 4.0, via Wikimedia Commons.”
  • Where in the post: Closing or near-closing supercomputer image. Shows what spectral GFS was running on at the very end, in dramatic contrast to the Cray X-MP at the start (1980).

Candidate F: ECMWF Reading area — Met Office Bracknell HQ (1999, demolished 2007)

  • Source page: https://commons.wikimedia.org/wiki/File:The_Meteorological_Office,Bracknell-geograph.org.uk-_489065.jpg
  • Direct file URL: https://upload.wikimedia.org/wikipedia/commons/6/65/The_Meteorological_Office%2C_Bracknell_-geograph.org.uk-_489065.jpg
  • License: CC BY-SA 2.0
  • Author: Anthony Eden (via geograph.org.uk)
  • Date: 1999
  • Depicts: The Met Office HQ on London Road, Bracknell - the building that housed the CDC Cyber 205 (1981-1990, the first global spectral model for operational use in Britain), Cray Y-MPs and successor machines. The building was demolished circa 2007 when the Met Office moved to Exeter; this is the only free-licensed photo of the actual spectral-era institution.
  • Suggested filename: Met_Office_Bracknell_1999.jpg
  • Suggested caption (Polish): “Met Office na London Road, Bracknell - rok 1999, na osiem lat przed wyburzeniem. Wlasnie w tym budynku w 1982 r. uruchomiono CDC Cyber 205
    • pierwsza brytyjska maszyne, na ktorej spektralny globalny model meteo liczyl prognozy operacyjne. Met Office przeniosl sie do Exeter w 2003 r. Anthony Eden, CC BY-SA 2.0, via geograph.org.uk / Wikimedia Commons.”
  • Where in the post: When discussing the UK Met Office’s adoption of the Cyber 205 in 1981/1982 - the building is a stand-in for the missing Cyber 205 photo. Strong institutional anchor for the spectral story.

Optional secondary candidates (use only if extra slots remain)

Candidate G: ETA10 at Computer History Museum (1987-89 vintage, photo 2010)

  • Source page: https://commons.wikimedia.org/wiki/File:ETA_Systems_ETA10supercomputer(1987-1989)where_CPU_is_mounted_in_a_liquid_nitrogen_tank_for_liquid_cooling(End_of_an_ERA)_-_Computer_History_Museum,_2010-01-21_15.43.38_by_Jitze_Couperus.jpg
  • Direct file URL: https://upload.wikimedia.org/wikipedia/commons/6/60/ETA_Systems_ETA10supercomputer(1987-1989)where_CPU_is_mounted_in_a_liquid_nitrogen_tank_for_liquid_cooling(End_of_an_ERA)_-_Computer_History_Museum,_2010-01-21_15.43.38_by_Jitze_Couperus.jpg
  • License: CC BY 2.0
  • Author: Jitze Couperus
  • Date: 21 January 2010 (machine vintage 1987-1989)
  • Depicts: The ETA10 at CHM. ETA10 was the CDC successor to the Cyber 205, ran at FSU (Florida State University, home of the FSU spectral model) from 1987 to 1989. Cooled by an external liquid-nitrogen tank, visible in the shot.
  • Suggested filename: ETA10_CHM_2010.jpg
  • Note: Use this ONLY if you need a third late-1980s vector-machine image. It is adjacent rather than central to the spectral GFS story.

Candidate H: Cray C90 front panel (CC0 - cleanest licence)

  • Source page: https://commons.wikimedia.org/wiki/File:Cray_C90Frontpanel(137062958).jpg
  • Direct file URL: https://upload.wikimedia.org/wikipedia/commons/5/53/Cray_C90Frontpanel%28137062958%29.jpg
  • License: CC0 1.0 (public domain dedication)
  • Author: Wolfgang Stief, Tittmoning, Germany
  • Date: 29 April 2006
  • Depicts: Close-up of the C90 front control panel.
  • Note: A detail / close-up alternative if you want a “machine close-up” rather than a full-cabinet view. Smaller, lower-resolution than Candidate B.

Candidate I: NCWCP building at College Park (NCEP’s current home)

  • Source page: https://commons.wikimedia.org/wiki/File:NCWCPcollegepark.jpg
  • Direct file URL: https://upload.wikimedia.org/wikipedia/commons/2/22/NCWCPcollegepark.jpg
  • License: PD-USGov (NOAA work)
  • Author: John T. Consoli / University of Maryland; published by NOAA
  • Date: 27 August 2012
  • Depicts: The National Center for Weather and Climate Prediction building at M Square Research Park, College Park MD. NCEP moved here in 2012; in the final years of the spectral GFS (2012-2019) this is the operational home of the model (the actual hardware was at Reston VA / Fairmont WV WCOSS data centres, but NCEP’s modellers and the EMC team work here).
  • Suggested filename: NCEP_NCWCP_College_Park.jpg
  • Note: Institutional photo - useful for “where the spectral GFS lived in its final years” but the building is generic-modern, not historically evocative. Use only if there’s a clear narrative slot for it.

Candidate J: 1982 Cold Sunday surface weather map (PD-NOAA)

  • Source page: https://commons.wikimedia.org/wiki/File:Cold_Sunday_1982-01-17_weather_map.png
  • Direct file URL: https://upload.wikimedia.org/wikipedia/commons/1/15/Cold_Sunday_1982-01-17_weather_map.png
  • License: PD-USGov (NOAA Central Library, Silver Spring MD)
  • Date: 17 January 1982
  • Depicts: Surface weather analysis map of the US documenting the “Cold Sunday” weather event - dated less than 18 months after the spectral GFS went operational at NMC (August 1980). One of the few free-licensed vintage NWS surface analyses from the right era.
  • Note: This is a surface analysis chart, not a spectral-model upper-air output. Useful for “what NCEP was publishing in the early spectral era” but not directly tied to spectral methods.

Candidate K: NWS 500 mb upper-air map, 1998 (PD-NOAA)

  • Source page: https://commons.wikimedia.org/wiki/File:NWS_Upper_air_map_500mb_1998-01-05.gif
  • Direct file URL: https://upload.wikimedia.org/wikipedia/commons/a/af/NWS_Upper_air_map_500mb_1998-01-05.gif
  • License: PD-USGov (NWS Burlington VT)
  • Date: 5 January 1998
  • Depicts: 500 mb upper-air analysis for the 1998 North American ice storm
    • trough over the Rockies, ridge along the East Coast. The 500 mb chart is the canonical synoptic field that spectral models actually predict.
  • Note: Better fit than Candidate J because (a) it is a 500 mb chart, the field spectral models predict, and (b) it is dated 1998 - mid-spectral-GFS era. Use this in preference to J if a vintage NWS chart is wanted.

Given that Round 1 already covered Cray-1, Cray X-MP, and the spherical-harmonic concept images, the Round-2 additions that actually diversify the post are:

  1. Candidate A or C (Cray Y-MP at NASA, 1990 or 1993) - the post-X-MP vector generation. Pick ONE - I lean toward C (Goddard, 1993) because the distinctive grey-and-red Y-MP cabinet is more visually different from the X-MP at Linkoping that Round 1 already includes.

  2. Candidate B (Cray C90 at Computer Museum of America) - the C90 was the ECMWF operational machine 1994-1996 and the immediate successor to the Y-MP. Different shape, different colour, very different from the Linkoping X-MP and the Y-MP at Goddard.

  3. Candidate D (NOAA IBM supercomputer, 2009) - this is the only verified image showing the actual NCEP operational hardware from the late spectral era. Bridges from vector Crays to the parallel IBM clusters.

  4. Candidate E (Cray XC40 at HLRS Stuttgart, 2015) - final spectral-era architecture. Dramatic visual contrast with the 1980 Cray X-MP at the start of the spectral GFS story.

  5. Candidate F (Met Office Bracknell HQ, 1999) - institutional anchor for the UK spectral story, the one institution that never got a free-licensed photo of its actual Cyber 205. The building itself is the next best thing.

If only 3 slots remain in the post, drop Candidate E or F (the XC40 is the most disposable - the IBM in Candidate D already covers “late spectral era hardware”; the Met Office building is the most disposable if the post is US-focused on NCEP).

If the post is generous on images, add Candidate K (1998 500 mb chart) as a “what spectral GFS actually produces” visual.

  • Candidate G (ETA10) - successor to Cyber 205, but ran at FSU (which had its own FSU-spectral-model, separate lineage from operational spectral GFS). Adjacent at best.
  • Brian Hoskins New Delhi event photo - shows Hoskins as a side figure in a group ceremony. Not a portrait, would be misleading as one.
  • NCWCP College Park building (Candidate I) - generic modern office building, no historical narrative value.
  • Cold Sunday 1982 map (Candidate J) - surface analysis, not the field that spectral methods predict; less narratively useful than the 500 mb chart (Candidate K).
  • Any Science Museum Group Cyber 205 imagery - CC0 covers metadata only; the images themselves are under museum copyright. Do not use.
  • Google Arts & Culture / Met Office archive Cyber 205 photos - not Wikimedia, not CC-licensed. Do not use.