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.

BMRC and Australian NWP: Institutional Context (1968–1985)

Research notes for the spectral methods post anchored on William Bourke. The puzzle: why did the dominant operational NWP technique of 1983–2010 get invented at a peripheral Southern Hemisphere centre rather than at NCAR, GFDL, ECMWF, or the UK Met Office? This document assembles the institutional, computational, and personal context.

1. The chronology of Australian atmospheric research

Three parent institutions, not one

The Bourke story sits at the seam between two Commonwealth science organisations and through three successive research units:

  • Commonwealth Bureau of Meteorology (the operational agency, founded 1 January 1908 under the Meteorology Act 1906; restructured under the Meteorology Act 1955).¹
  • CSIRO Division of Meteorological Physics at Aspendale, founded 1946 by C.H.B. (Bill) Priestley brought from the UK Met Office; renamed Division of Atmospheric Physics in 1971.²
  • Three successive joint research units in Melbourne: Commonwealth Meteorology Research Centre (CMRC, 1969–1974) → Australian Numerical Meteorology Research Centre (ANMRC, 1974–1984) → Bureau of Meteorology Research Centre (BMRC, 1985 onward).³

CMRC was established in 1967 on paper and became operational in 1969 as a joint Bureau/CSIRO unit. It was created specifically to give Australian NWP a stable home that bridged operational and academic-style research — the operational forecasters at the Bureau and the basic-research atmospheric physicists at CSIRO Aspendale had different cultures, and CMRC was the negotiated compromise.³

Brian Tucker — the indispensable founding figure

The man who built the institution Bourke worked in was Gilbert Brian “Brian” Tucker (1930–2010). Tucker, born in Ogmore Vale, South Wales, graduated from Aberystwyth in 1950 and earned a PhD from Imperial College London in 1954. After RAF forecasting service and nine years (1956–1965) as a research scientist at the UK Meteorological Office, he was recruited to Melbourne in May 1965 to become Assistant Director (Research and Development) of the Bureau.⁴

When CMRC was set up in 1967–1969, Tucker was transferred from the Bureau to CSIRO and became its founding Officer-in-Charge (1969–1973). Under him “the centre blossomed as an independent research organisation bridging the research and operational responsibilities of the bureau.”⁴ In 1973 Tucker moved upstairs to become Chief of the CSIRO Division of Atmospheric Physics (1973–1988) and then Chief of the renamed CSIRO Division of Atmospheric Research (1988–1992).⁴

Tucker’s significance for Bourke’s career is structural rather than scientific: he was a UK-Met-Office-trained dynamicist who recognised that basic research on numerical methods was not a luxury but the only way a small country could matter in NWP, and he protected that space against operational forecasters’ impatience.⁴

Directors of the Bureau itself

For context: William James “Bill” Gibbs was Director of the Bureau of Meteorology 1962–1978 (and first Vice-President of WMO 1967–1975).⁵ He was succeeded by John Zillman, Director 1978–2003.⁶ Gibbs’s directorship covers the entire spectral-method development period, and the AMOS “Gibbs Medal” for operational meteorology is named after him to honour exactly that 1960s–70s era when “Bill Gibbs shaped and transformed operational meteorology in the Bureau.”⁵

2. The CMRC → ANMRC → BMRC sequence

CMRC (1969–1974)

CMRC operated under Tucker’s leadership from 1969 to 1973 (Tucker held the formal title Officer-in-Charge of CMRC 1969–1973).⁴ It was reviewed in 1973 — internal politics around Bureau/CSIRO governance — and renamed ANMRC in 1974.³

This is the institutional environment in which Bourke wrote his 1972 single-level paper. In other words: Bourke’s “Efficient, One-Level, Primitive-Equation Spectral Model” (Monthly Weather Review 100:683–689, 1972)⁷ was published from CMRC, an institution barely three years old, into the American Meteorological Society’s flagship journal.

ANMRC (1974–1984)

The 1974 review made the centre formally independent of both parent agencies, with an Officer-in-Charge “responsible to the CSIRO and the Secretary, Department of Science (and successor agencies).”³ This is politically important: it gave Bourke and colleagues a degree of insulation from operational forecasting demands that they would not have had inside the Bureau proper. ANMRC was the home for Bourke’s 1974 multi-level paper, the 1977 GASP-precursor model (Bourke, McAvaney, Puri & Thurling, Methods in Computational Physics) and the 1978 McAvaney, Bourke & Puri global spectral GCM (Journal of the Atmospheric Sciences 35:1557).⁸ ⁹

ANMRC was wound up in 1984. It produced a 160-page valedictory report covering 1969–1984.³

BMRC (1985 onward)

In 1985 the Bureau folded the research function back inside itself, creating BMRC as “an essentially self-contained research division of the Bureau of Meteorology”; “many ANMRC staff accepted positions with the BMRC.”³ This is the political reverse of 1973: the Bureau wanted closer control of research, presumably because by 1985 NWP was clearly operational and the basic-research justification for an independent unit had weakened. (The BMRC was eventually merged with CSIRO’s atmospheric research arm in 2007 to form CAWCR — the Centre for Australian Weather and Climate Research — but that is well outside our period.)

The sources I consulted do not name BMRC’s first director on the open web; the EOAS record describes BMRC as a Bureau division rather than a freestanding body with a sequence of named heads. For the post it is enough to say: by 1985, when BMRC opened, Bourke’s spectral method had already conquered the world. The post should treat 1968–1984 (CMRC → ANMRC) as the period that matters; “BMRC” is the convenient label posterity uses for the Australian spectral lineage as a whole.

3. Australian operational NWP — the slow climb

1969: Australia’s first real-time numerical forecasts, single-layer barotropic model, covering portions of the Southern Hemisphere.¹⁰

1972: D.J. Gauntlett, R.S. Seaman, W.R. Kininmonth and J.C. Langford publish “An operational evaluation of a numerical analysis-prognosis system for the southern hemisphere” in the Australian Meteorological Magazine.¹¹ This is the operational side, in parallel with Bourke’s research-side spectral work.

1976: D.J. Gauntlett publishes “A semi-implicit forecast model using the flux form of the primitive equations” in the Quarterly Journal of the Royal Meteorological Society¹² — the year before the operational primitive-equation rollout.

1977: Australia goes operational with primitive-equation models.¹⁰ This is finite-difference grid-point, not yet spectral.

Late 1980s / early 1990s: GASP (Global AnalysiS and Prediction system) becomes operational; this is the operational spectral model that finally inherits Bourke’s research lineage. GASP “uses spectral numerical analysis techniques.”¹³ The publicly searchable archive runs 1998 onward, but the system was operational well before that.

So the timeline is sharp: Bourke invents the operational world’s spectral method in 1972 at CMRC; Australia goes operational with primitive-equation gridpoint models five years later (1977); ECMWF goes operational with a spectral model in 1983; and Australia’s own operational spectral system (GASP) is a 1990s development. Australia was first to invent it and late to deploy it operationally. This is typical of small national centres: research lead, operational lag.

4. The cast

William Bourke

Beyond his three foundational papers (1972, 1974a, 1974b with Puri), Bourke’s 1988 book chapter “Spectral Methods in Global Climate and Weather Prediction Models” (in M.E. Schlesinger, ed., Physically-Based Modelling and Simulation of Climate and Climatic Change, vol. I, Kluwer, pp. 169–220)¹⁴ became the canonical review and pedagogical text. He remained at the Bureau for the rest of his career and was still co-authoring mesoscale-system verification papers in the 2008 Australian Meteorological Magazine.¹⁵

I could not locate (on the open web) a formal obituary, an AMOS Priestley Medal citation, or an AMOS Fellow citation that explicitly names Bourke during my research window. AMOS’s later “Zillman Medal” for late-career contribution would have been a natural fit, but I did not see Bourke in the recipient lists I could surface. For the post, do not claim he won the Priestley Medal unless you can verify it independently. What is clear is that the intellectual recognition is enormous: every modern spectral-transform NWP paper cites Bourke 1972 and/or 1974.

Bryant McAvaney

Co-author with Bourke and Puri of the 1978 JAS paper.⁸ McAvaney was a joint recipient of the AMOS Priestley Medal in 1983 (with N.E. Davidson, both “from the Bureau of Meteorology”).¹⁶ This is the closest the spectral group came to formal AMOS mid-career recognition in our period, and it is worth quoting in the post: the Priestley Medal is named after C.H.B. Priestley of CSIRO Aspendale, so the citation implicitly links the McAvaney–Bourke–Puri spectral work back to the Priestley tradition of fundamental atmospheric physics at Aspendale.

Kamal Puri

PhD in physics from Manchester. With Bourke on the 1974 MWR paper on horizontal resolution in spectral integrations (Puri & Bourke, MWR 102:333–347). Co-author on the 1977 and 1978 Australian spectral GCM papers. Crucially, Puri took the Australian spectral model to NCAR during an extended visit. The NCAR Community Climate Model CCM0A was based directly on the Bourke–McAvaney–Puri–Thurling spectral model: “That model was adapted to the NCAR computers by K. Puri (Australian Numerical Meteorological Research Centre) during an extended visit and subsequently modified by E. Pitcher (University of Miami) and R. Malone (Los Alamos/DOE).”¹⁷ This is the single most important transmission event for the post: the Australian code physically becomes the seed of NCAR’s CCM lineage, which itself becomes CCSM and CAM.

Puri spent the rest of his career at the Bureau, eventually becoming Research Programme Leader of the Earth System Modelling Programme and leading the development of ACCESS, the Australian Community Climate and Earth System Simulator.¹⁸

Michael (Mike) Manton

M.J. Manton was at BMRC in the late 1980s and 1990s as a researcher.¹⁹ He is now at Monash. Do not describe him as BMRC director in the post without further verification — the sources I could surface describe him as a researcher and group leader, not as the unit head.

John McGregor

CSIRO (not Bureau/ANMRC) — known for the Conformal-Cubic Atmospheric Model (CCAM), an alternative-grid approach developed at CSIRO Aspendale.²⁰ McGregor’s whole research programme is, in a sense, a counterpoint to Bourke: spectral methods finesse the pole problem on the sphere by changing basis functions; conformal-cubic methods finesse the same problem by changing the grid. The post could mention this as the Australian grid-side lineage running in parallel.

Tony Hirst

Was at CSIRO atmospheric research. The sources I retrieved didn’t yield material specific to the spectral-NWP story in the 1970s. Don’t lean on him unless you have another source.

Reg Clarke

R.H. Clarke (1914–1990) — long career at Bureau, CSIRO, and University of Melbourne; Chairman of the Royal Meteorological Society Australian Branch (the pre-AMOS body) 1976–80.²¹ The AMOS “R.H. Clarke Lecture” is named in his memory. Useful as background colour about the senior generation Bourke worked under.

5. Why a peripheral centre got there first

This is the post’s central question. Several institutional answers, none sufficient on its own:

The Southern Hemisphere drove a different research agenda

The Southern Hemisphere is data-sparse — vast oceans, almost no radiosonde coverage south of New Zealand and South Africa, an Antarctic continent that closes off a polar cap rather than opening it (as the Arctic Ocean does). For data assimilation, that matters: every observation is precious, and methods that can ingest sparse asynoptic data have outsize value. The 1982 paper Le Marshall, Leslie & Spinoso, “ANMRC Data Assimilation for the Southern Hemisphere” (MWR 110:1749)²² is the data-assimilation companion to the spectral-model work.

For dynamical-core design, the pole problem matters more in the Southern Hemisphere only in a soft sense — both poles are equally polar to a gridpoint model. But the South Pole is a continent, which means realistic model topography crashes hard against a converging grid; the Antarctic ice sheet is right where the gridpoint geometry gets worst. Spectral methods finesse this entirely: spherical harmonics are uniform on the sphere; there is no pole problem.²³ For a centre with operational responsibility for the whole Southern Hemisphere down to and including Antarctica, that geometric uniformity was not aesthetic but useful.

The “small centre” advantage

CMRC/ANMRC employed dozens, not hundreds, and most of those people were focused on a single global model. That meant: short feedback loops, no committee-driven model-development politics, and freedom to bet everything on one numerical approach. GFDL, NCAR, NMC and the UK Met Office all had to support multiple model lineages and broader research programmes. A small centre could pivot harder.

Tucker’s protection

Tucker, as Officer-in-Charge and then CSIRO Division Chief, was a UK-Met-Office-trained dynamicist; he understood that the operational side of meteorology pulls toward immediate forecast improvement, and that you have to fence off basic-research time deliberately or it evaporates.⁴ The ANMRC’s 1974 reorganisation as an independent unit “responsible to CSIRO and the Secretary, Department of Science”³ was precisely such a fence. Bourke did not have to justify spectral methods in terms of next-week’s operational forecast skill.

Tucker came from the UK Met Office. C.H.B. Priestley founded CSIRO Aspendale after coming from the UK. The Australian community read Monthly Weather Review and Journal of the Atmospheric Sciences as their core journals, not Australian outlets. Bourke’s 1972 paper landed in MWR because that was simply where one published serious NWP work. The “Pacific crossing” the post asks about is mostly an illusion — there was never a Pacific gap to cross, because the Australian community treated the AMS journals as their natural home.

The Eliasen–Machenhauer–Rasmussen and Orszag context

The transform method itself was simultaneously invented by Eliasen, Machenhauer and Rasmussen (Copenhagen) and by Orszag (MIT/Princeton), both 1970.²⁴ Bourke’s 1972 contribution was specifically to make this work as a practical, efficient, global, primitive-equation NWP formulation — not to invent the transform method. The genuine novelty in Bourke 1972 was the concise vorticity-divergence formulation plus the grid-transform evaluation of nonlinear terms, in a form that demolished the cost of traditional interaction-coefficient methods.⁷ He was building on Eliasen et al. and Orszag, but he was the first to show the operational meteorology community that the method could actually fly.

6. Computers

The literature I could surface does not give a clean Bureau/CMRC machine list for 1968–1985, but the general picture:

  • The CSIRO side ran a CDC 3600 in Canberra from 1964 with CDC 3200 satellites in state capitals.²⁵ Aspendale would have had access via these.
  • The Bureau’s operational and research computing was run separately. Cyber and IBM machines were used over the 1970s–80s; precise dates remain to be confirmed.
  • “The use of supercomputers in Australia can be traced back to the late 1960s, led by two national agencies, the Bureau of Meteorology (BoM) and the national science agency, CSIRO — driven by the need for numerical weather prediction and computational support for research. Since then, these two agencies have operated separate and shared facilities, which have included IBM, CDC, Cray, Fujitsu and NEC systems.”²⁶
  • The Bureau awarded its supercomputing contract to NEC for an SX-4 in April 1997 (in a joint BoM/CSIRO High Performance Computing and Communications Centre), beating Cray and Fujitsu; the SX-4 was installed September 1997 at 150 Lonsdale Street, Melbourne.²⁶
  • Australia’s first Cray for meteorology proper was therefore not until the post-2010 ACCESS era; the much later Cray XC40 “Australis” came online 30 June 2016 at 1.6 petaflops, and Cray XC50 “Australis II” in April 2020 at 4.0 petaflops.²⁷

Key implication for the post: Bourke did the 1972 work on machines that were modest even by 1972 standards. The spectral method’s computational efficiency — its small constant factor for global integrations on limited-memory machines — wasn’t just an aesthetic advantage but a necessity for Australian computing.

7. International connections

  • NCAR: Puri’s extended visit transferred the Australian spectral model into the NCAR CCM lineage.¹⁷ This is the single best-attested cross-Pacific connection.
  • AMS journals: Bourke 1972, 1974; McAvaney–Bourke–Puri 1978; all in MWR or JAS. The Australian community was fully integrated into the American journal system.
  • ECMWF: Founded 1975, operational 1979, spectral from 1983. Australia has long-standing data and model exchanges with ECMWF; the formal “strategic relationship agreement” between BoM and ECMWF was signed only recently.²⁸ The 1970s and 80s relationship was through WMO/CAS channels rather than via formal bilateral agreements. Lennart Bengtsson, ECMWF founding director, knew the Australian work — he authored “ECMWF — Ten Years of European Meteorological Co-operation” documenting the centre’s founding choices, including the choice to go with grid-point initially and switch to spectral in 1983.²⁹
  • GFDL and the GARP programme (Global Atmospheric Research Programme, 1967–1980) provided the broader frame within which Bourke’s work was visible. Australia was a participant in the FGGE (First GARP Global Experiment, 1978–79), which generated the global observational dataset that everyone tested their global models against.

8. Quotes, personalities, AMOS notes

  • Tucker is remembered as “a strong but hands-off leader and a much-admired mentor of younger staff, inspiring loyalty among professional colleagues but was not averse to vigorous debate on matters of science, policy or administration.” After CSIRO he was a Senior Fellow at the Institute of Public Affairs in Melbourne and became something of a climate-policy contrarian in retirement. He died at Mornington on 25 November 2010 aged 80.⁴
  • Tucker himself wrote a 1976 Weather article called “Research and Services: Differing Attitudes within the Science of Meteorology”³⁰ — exactly the operational/research tension that shaped CMRC’s governance.
  • McAvaney’s 1983 AMOS Priestley Medal is on the AMOS record.¹⁶

9. Verified vs unverified

Verified for the post:

  • Bourke 1972 single-level paper, MWR 100:683.⁷
  • Bourke 1974 multi-level paper, MWR 102:687.⁹
  • Puri & Bourke 1974 resolution paper, MWR 102:333. (referenced through context only)⁸
  • McAvaney, Bourke & Puri 1978 GCM paper, JAS 35:1557.⁸
  • Bourke 1988 book chapter in Schlesinger ed.¹⁴
  • CMRC founded 1969, joint Bureau/CSIRO.³
  • Brian Tucker founding OIC of CMRC 1969–1973.⁴
  • ANMRC name change 1974, independent governance.³
  • ANMRC wound up 1984, valedictory report covers 1969–1984.³
  • BMRC founded 1985 as Bureau division.³
  • Bill Gibbs Director of Bureau 1962–1978; John Zillman 1978–2003.⁵ ⁶
  • Australia’s first NWP forecasts 1969 (single-layer barotropic); primitive-equation operational 1977.¹⁰
  • Eliasen–Machenhauer–Rasmussen and Orszag both 1970 for the transform method.²⁴
  • NCAR CCM0A based directly on the Bourke–McAvaney–Puri–Thurling Australian spectral model; transferred by Puri during extended NCAR visit.¹⁷
  • Bryant McAvaney AMOS Priestley Medal 1983 (joint with N.E. Davidson).¹⁶
  • BoM/CSIRO NEC SX-4 supercomputer contract April 1997.²⁶

Not verified — do not assert in post:

  • Bourke as AMOS Priestley Medal recipient.
  • Bourke as AMOS Fellow.
  • BMRC’s first director’s name.
  • Mike Manton as BMRC director.
  • Specific year Australia went operational with a spectral model (probably early 1990s — likely 1990–1992 for GASP — but not pinned down in the sources I could reach).
  • Specific computer used at CMRC/ANMRC for Bourke 1972 and 1974 (likely Cyber-series, but not confirmed).

Footnotes / sources

  1. Wikipedia, “Bureau of Meteorology”; Meteorology Act 1955.
  2. EOAS, “CSIRO Division of Meteorological Physics” (A000763) and “CSIRO Division of Atmospheric Physics” (A000644); CSIROpedia, “Priestley, Charles Henry Brian (Bill).”
  3. EOAS, “Australian Numerical Meteorology Research Centre (ANMRC)” (A000912) and “Bureau of Meteorology Research Centre (BMRC)” (A000913); EOAS “Commonwealth Bureau of Meteorology” (A000909).
  4. EOAS, “Tucker, Gilbert Brian (Brian)” (P003315); CSIROpedia, “Gilbert Brian Tucker [1930–2010].”
  5. EOAS, “Gibbs, William James (Bill)” (P003253); AMOS, “The Gibbs Medal.”
  6. EOAS, “Zillman, John William” (P003111); Wikipedia, “John Zillman.”
  7. Bourke, W., 1972: “An Efficient, One-Level, Primitive-Equation Spectral Model.” Monthly Weather Review 100:683–689.
  8. McAvaney, B.J., W. Bourke and K. Puri, 1978: “A Global Spectral Model for Simulation of the General Circulation.” Journal of the Atmospheric Sciences 35:1557; received 12 December 1977. Also Bourke, McAvaney, Puri & Thurling 1977, “Global modelling of atmospheric flow by spectral methods,” Methods in Computational Physics 17:267–324.
  9. Bourke, W., 1974: “A Multi-Level Spectral Model. I. Formulation and Hemispheric Integrations.” Monthly Weather Review 102:687–701.
  10. Wikipedia, “History of numerical weather prediction”; AMOS, “Statement on Weather Analysis and Prediction” (2018).
  11. Gauntlett, D.J., R.S. Seaman, W.R. Kininmonth, J.C. Langford, 1972: “An operational evaluation of a numerical analysis-prognosis system for the southern hemisphere,” Australian Meteorological Magazine.
  12. Gauntlett, D.J., 1976: “A semi-implicit forecast model using the flux form of the primitive equations.” Quarterly Journal of the Royal Meteorological Society 102:517.
  13. BoM GASP dataset documentation, UCAR; researchdata.edu.au “BOM (GASP) — Australian Bureau of Meteorology Global Analysis and Prediction global gridded dataset archive — 1998 to present.”
  14. Bourke, W., 1988: “Spectral Methods in Global Climate and Weather Prediction Models.” In M.E. Schlesinger, ed., Physically-Based Modelling and Simulation of Climate and Climatic Change, Vol. I, Kluwer, 169–220.
  15. Vincent, C.L., W. Bourke et al., 2008: “Verification of a high-resolution mesoscale NWP system,” Australian Meteorological Magazine 57(3).
  16. AMOS, “Priestley Medal,” recipient list, citing N.E. Davidson and B.J. McAvaney 1983 (Bureau of Meteorology).
  17. NCAR CCM0A documentation; Edwards, P.N., A Vast Machine, NCAR chapter (online appendix), explicit attribution of the CCM lineage to the Australian spectral model adapted by Puri during his extended NCAR visit.
  18. ResearchGate, “Kamal PURI”; The Conversation profile; CAWCR achievements page.
  19. ResearchGate / Monash, “Michael Manton.”
  20. CSIRO CCAM project pages; McGregor & Dix, “The CSIRO Conformal-Cubic Atmospheric GCM,” Springer chapter.
  21. EOAS, “Clarke, Reginald Henry” (P003236); AMOS “R.H. Clarke Lecture” citation.
  22. Le Marshall, J.F., L.M. Leslie and C. Spinoso (or similar co-authors), 1982: “ANMRC Data Assimilation for the Southern Hemisphere.” Monthly Weather Review 110:1749–1760.
  23. General NWP textbook context — pole problem in latitude–longitude grids; spherical harmonics are uniform on the sphere. See Krishnamurti et al., An Introduction to Global Spectral Modeling, 2006; and the 100-Years monograph by Randall et al., Meteorological Monographs 59(1), 2019.
  24. Eliasen, E., B. Machenhauer and E. Rasmussen, 1970: “On a numerical method for integration of the hydrodynamical equations with a spectral representation of the horizontal fields,” Copenhagen University Institute for Theoretical Meteorology Report No. 2. Orszag, S.A., 1970: “Transform Method for the Calculation of Vector-Coupled Sums: Application to the Spectral Form of the Vorticity Equation,” Journal of the Atmospheric Sciences 27:890.
  25. CSIROpedia, “CSIRO Computing History,” Chapters 5 and 7.
  26. CSIROpedia, “CSIRO Computing History” Chapter 7; cray-history.net 1997 retrospective.
  27. iTnews, “Cray to build Australia’s biggest supercomputer”; BoM Australis press materials.
  28. Meteorological Technology International, “Australian Bureau of Meteorology partners with ECMWF” (post-2020 announcement).
  29. Bengtsson, L., “ECMWF — Ten Years of European Meteorological Co-operation,” MPG repository.
  30. Tucker, G.B., 1976: “Research and Services: Differing Attitudes within the Science of Meteorology.” Weather 31:381.