Reminiscences of MODE
17
hothouse of the scientific program gained substantially from the relaxed physical
environment of Woods Hole in summer and from the assistance provided by Mary
Thayer to families such as mine who were visiting from overseas.
During the 1960s computers were gradually introduced into routine weather
prediction, an experience that has many parallels in present-day oceanography. At
first, the models were not very sophisticated, with one or two layers in the vertical,
generally assuming quasi-geostrophic balance, and for every forecast they had to be
reinitialized from a traditional, hand-drawn analysis. I recall visiting the Royal Meteorological Office in Bracknell, England (probably in 1964). By chance it was the
first day that the new automated data system and analysis procedure was operational.
Everyone was very proud of this achievement, so I was surprised to find the Chief
Duty Forecaster at the bench redrawing the 12Z weather map by hand. He explained
that the latest satellite cloud image showed that the cold front in the mid-Atlantic
had been placed too far to the west. However, to communicate this information to
the computer, he had to generate bogus ship reports in the required international data
format, and feed them into the automated data system for the next analysis cycle.
There were also concerns about the data from a ship that had reported its position to
be in the middle of Greenland. For me, these were sobering reminders that the practical benefits of using computers to solve the equations of motion would be realized
only after painstaking attention to the realities of observation systems, particularly
the characteristics and incompatibilities of the information they do or do not provide.
Forty years later that wheel has come full circle. Computing capabilities have become enormously more powerful. Following sustained developments efforts by the
meteorological services of the world, a systematic framework has been developed for
addressing issues such as quality control on the available data streams, the adequacy
of sampling, and the blending of measurements of different types within a numerical
model into dynamically coherent representation of the atmosphere that evolves continuously on a daily basis. Indeed, these automated operational systems have become
so pervasive and complex that they are impenetrable by outsiders, and their representations are often treated as fact, with only lip service paid to the shortcomings of the
underlying model and to the assumptions governing the data screening and quality
control.
In 1969 I moved to the Johns Hopkins University in Baltimore, Maryland, in
the Department of Earth and Planetary Sciences.
THE ROOTS OF MODE
Soon after my arrival in Baltimore, Henry Stommel described his ideas for a project
to pursue the intriguing discovery made a decade earlier by John Swallow and Hal
Worthington (1957).
At that time our only knowledge of the circulation in the deep ocean came
from analyses of hydrographic soundings. These were water samples taken from
17
hothouse of the scientific program gained substantially from the relaxed physical
environment of Woods Hole in summer and from the assistance provided by Mary
Thayer to families such as mine who were visiting from overseas.
During the 1960s computers were gradually introduced into routine weather
prediction, an experience that has many parallels in present-day oceanography. At
first, the models were not very sophisticated, with one or two layers in the vertical,
generally assuming quasi-geostrophic balance, and for every forecast they had to be
reinitialized from a traditional, hand-drawn analysis. I recall visiting the Royal Meteorological Office in Bracknell, England (probably in 1964). By chance it was the
first day that the new automated data system and analysis procedure was operational.
Everyone was very proud of this achievement, so I was surprised to find the Chief
Duty Forecaster at the bench redrawing the 12Z weather map by hand. He explained
that the latest satellite cloud image showed that the cold front in the mid-Atlantic
had been placed too far to the west. However, to communicate this information to
the computer, he had to generate bogus ship reports in the required international data
format, and feed them into the automated data system for the next analysis cycle.
There were also concerns about the data from a ship that had reported its position to
be in the middle of Greenland. For me, these were sobering reminders that the practical benefits of using computers to solve the equations of motion would be realized
only after painstaking attention to the realities of observation systems, particularly
the characteristics and incompatibilities of the information they do or do not provide.
Forty years later that wheel has come full circle. Computing capabilities have become enormously more powerful. Following sustained developments efforts by the
meteorological services of the world, a systematic framework has been developed for
addressing issues such as quality control on the available data streams, the adequacy
of sampling, and the blending of measurements of different types within a numerical
model into dynamically coherent representation of the atmosphere that evolves continuously on a daily basis. Indeed, these automated operational systems have become
so pervasive and complex that they are impenetrable by outsiders, and their representations are often treated as fact, with only lip service paid to the shortcomings of the
underlying model and to the assumptions governing the data screening and quality
control.
In 1969 I moved to the Johns Hopkins University in Baltimore, Maryland, in
the Department of Earth and Planetary Sciences.
THE ROOTS OF MODE
Soon after my arrival in Baltimore, Henry Stommel described his ideas for a project
to pursue the intriguing discovery made a decade earlier by John Swallow and Hal
Worthington (1957).
At that time our only knowledge of the circulation in the deep ocean came
from analyses of hydrographic soundings. These were water samples taken from