52
Russ E. Davis
more than qualitative. These findings and a disappointing data return from MODE-0
had important consequences. Bill Schmitz took a stronger hand in the Buoy Group
and applied technical leadership in design, operations, and personnel; the improvements throughout MODE and after were significant and obvious. The Buoy Group
also stopped using surface moorings except when they were needed to observe meteorology or the near-surface ocean. In short order subsurface deployments of 2 years
became common.
An improved SOFAR float introduced by Rossby and Webb was a big step
toward the duration and tracking range needed for basin-scale observations. These
giant floats, 5 m long and weighing 425 kg, broadcasted at 270 Hz, could be tracked
from 2000 km or more, and were designed to operate for a year. High mechanical
stresses produced by the loud sound source caused initial mechanical failures in the
MODE floats. High-frequency pingers allowed them to be recovered and as Doug
Webb (personal communication) reports,
The early failure . . . was caused by unbonding of the urethane window on the
transducers. Fortunately R.V. Researcher was equipped with a large lathe. The
floats were recovered, the faces removed, machined, a polyurethane disk inserted
in a machined groove and an improvised 60 ton manually operated press deformed
the aluminum to seal the face in place, and the floats were relaunched.
After repair the floats functioned well throughout MODE-1 and two were tracked for 2
years. The data awakened great interest in analysis of Lagrangian data and gave us the
first comprehensive views of horizontal stirring and particle dispersion in the ocean
(Freeland, Rhines, and Rossby, 1975). Unfortunately, the policy that publication rights
belonged to data producers, led to many early and valuable studies encouraged for
planning being relegated to the gray literature—the oft-cited fundamental analysis of
Lagrangian statistics, “Particle dispersion in the western North Atlantic” by Jim Price
(1982) of WHOI comes to my mind. Many simply abandoned the program because
of this policy. Today, all participants in research programs typically can collaborate
in publishing data analysis; the Tropical Pacific TOGA Observing System and Argo
both distribute data immediately.
From the observational point of view, many of MODE’s scientific payoffs were
realized by statistical descriptions of scales, vertical modes, and lateral propagation
and by tests of geostrophy and horizontal nondivergence carried out with statistics. The
results most in keeping with the array design were the objective-mapping analyses by
McWilliams (1976a, b) in which float velocities were combined with density from the
STD/CTD surveys to produce maps of geostrophic streamfunction at all levels. Quasigeostrophic potential vorticity was computed from these maps and its conservation
examined as a test of quasi-geostrophy. McWilliams argued that (a) because MODE-1
was too short for statistical reliability, any dynamical test needed to be applied on a
synoptic basis and (b) that a test of potential vorticity conservation was preferable to a
direct test of geostrophy because it was less affected by inevitable observational error.
The first point was Hasselman’s argument for a statistical array put the other way.
Russ E. Davis
more than qualitative. These findings and a disappointing data return from MODE-0
had important consequences. Bill Schmitz took a stronger hand in the Buoy Group
and applied technical leadership in design, operations, and personnel; the improvements throughout MODE and after were significant and obvious. The Buoy Group
also stopped using surface moorings except when they were needed to observe meteorology or the near-surface ocean. In short order subsurface deployments of 2 years
became common.
An improved SOFAR float introduced by Rossby and Webb was a big step
toward the duration and tracking range needed for basin-scale observations. These
giant floats, 5 m long and weighing 425 kg, broadcasted at 270 Hz, could be tracked
from 2000 km or more, and were designed to operate for a year. High mechanical
stresses produced by the loud sound source caused initial mechanical failures in the
MODE floats. High-frequency pingers allowed them to be recovered and as Doug
Webb (personal communication) reports,
The early failure . . . was caused by unbonding of the urethane window on the
transducers. Fortunately R.V. Researcher was equipped with a large lathe. The
floats were recovered, the faces removed, machined, a polyurethane disk inserted
in a machined groove and an improvised 60 ton manually operated press deformed
the aluminum to seal the face in place, and the floats were relaunched.
After repair the floats functioned well throughout MODE-1 and two were tracked for 2
years. The data awakened great interest in analysis of Lagrangian data and gave us the
first comprehensive views of horizontal stirring and particle dispersion in the ocean
(Freeland, Rhines, and Rossby, 1975). Unfortunately, the policy that publication rights
belonged to data producers, led to many early and valuable studies encouraged for
planning being relegated to the gray literature—the oft-cited fundamental analysis of
Lagrangian statistics, “Particle dispersion in the western North Atlantic” by Jim Price
(1982) of WHOI comes to my mind. Many simply abandoned the program because
of this policy. Today, all participants in research programs typically can collaborate
in publishing data analysis; the Tropical Pacific TOGA Observing System and Argo
both distribute data immediately.
From the observational point of view, many of MODE’s scientific payoffs were
realized by statistical descriptions of scales, vertical modes, and lateral propagation
and by tests of geostrophy and horizontal nondivergence carried out with statistics. The
results most in keeping with the array design were the objective-mapping analyses by
McWilliams (1976a, b) in which float velocities were combined with density from the
STD/CTD surveys to produce maps of geostrophic streamfunction at all levels. Quasigeostrophic potential vorticity was computed from these maps and its conservation
examined as a test of quasi-geostrophy. McWilliams argued that (a) because MODE-1
was too short for statistical reliability, any dynamical test needed to be applied on a
synoptic basis and (b) that a test of potential vorticity conservation was preferable to a
direct test of geostrophy because it was less affected by inevitable observational error.
The first point was Hasselman’s argument for a statistical array put the other way.