Woods Hole Oceanographic Institution
Meddy Trajectories in the Canary Basin
Measured during the Semaphore Experiment
P.L.Richardson & A.Tychensky
The Semaphore experiment which occurred in the Canary Basin during 1993-
1994 provided an opportunity to identify and seed Meddies (Mediterranean
Water Eddies) with floats with the goal of obtaining a few more case his-
tories.
The Meddies were tracked with freely drifting RAFOS floats purchased from
Seascan Corp. in Falmouth, Mass., and assembled, calibrated, and ballasted at
WHOI (Table 2). The floats recorded temperature, pressure and times of arrival
from moored sound sources. At the end of their 1.5 year mission each float
dropped a weight, rose to the surface and transmitted the data to the labor-
atory via the Argos satellite system. Three sound sources were deployed as
part of Semaphore near 30N 24W, 36N 27W, and 37N 21W. Although they trans-
mitted three times per day only two of these transmissions were recorded by
the floats due to memory constraints and the relatively long mission. The
third time of arrival was linearly interpolated from the recorded ones in
order to obtain an evenly spaced time series. Meddies 2-4 remained in the
vicinity of the Semaphore sound source array & the floats were tracked with-
out any difficulties. Meddy 1 drifted far to the west of the sound source
array and west of numerous seamounts. The recorded transmissions were inter-
mittently blocked by seamounts which caused gaps in the records. In order
to supplement the tracking of this Meddy, recorded TOAs were used from an
additional source launched near 43N 36W by H.T.Rossby. Tracking errors in
position within the Semaphore box were estimated to be around 4 km based on
a comparison of float launch locations and first tracked positions. West of
the seamounts the errors in tracking floats in Meddy 1 could be much larger
than this (~ 40km) due to poor tracking geometry, blocking by seamounts, and
interpolation.
In order to help resolve the looping trajectories of the floats in Meddy 1
(south of 33N) gaps in the time of arrival series were interpolated using an
objective analysis technique called krieging (Hansen and Herman, 1989). It
was necessary to specify the float looping frequency which was obtained from
the series before and after each gap or from another series. Problems with
the technique were encountered when the looping frequency varied over a gap
and when the data were noisy (i.e., low signal to noise ratio) which made it
difficult to eliminate erroneous data. Some bad data caused erroneous phase
shifts of the interpolated series. The floats in Meddy 1 were tracked using
the krieged times of arrival. Some portions of the trajectories looked in-
correct compared to good Meddy trajectories, and were discarded. These and
a few gaps where two or more TOA series were not available were interpolated
subjectively (~23% of the time series for float 177) using continuity and the
looping frequency and diameter as observed in the available TOA series. The
overall trajectory of float 177 shown for Meddy 1 is judged to be qualitative-
ly correct; the details of the loops south of 33N should be viewed with caution.
A cubic spline function was passed through the three daily float positions
to calculate velocity. The positions and velocities were smoothed to reduce
noise and high frequency tidal & inertial oscillations using a Gaussian shaped
filter (sigma = 1 day).
Ten floats were launched in the 4 Semaphore Meddies (Table II). Two floats
either did not surface or did not transmit. One of these (133) had been hand-
led very roughly during shipping as demonstrated by its broken glass pressure
hull. The glass hull was replaced and a clock timing problem adjusted before
launch. One float (168) recorded acoustic transmissions for only the first 100
days. This float sank around 200 m over its life which suggests that a slow
leak could have caused the acoustic receiver to fail. Another float (173) in
the same Meddy (Meddy 3) recorded poor quality data during the first 60 days.
Combining the two float trajectories gave a complete series for this Meddy.
We used hollow aluminum drop weights with the first eight floats and sealed
the glass tubes with an aluminum end plate and silicone sealant. These floats
tended to sink at a rate around 50 +/- 33 meters over the 1.5 years which implies
that either the hollow weights or float hulls leaked. Most of the mean sink rate
was caused by two floats which sank 170 m (float 171) and 200 m (float 168).
Based on early results in the AMUSE experiment (Bower et al., 1997) we switched
to solid drop weights and a better sealing technique for the last two floats. For
these the end plate was sealed to the glass hull with butyl tape and a shrink-
wrapped sleeve over the connection. A partial internal vacuum (~1/2 atmosphere)
was created to hold the end plate tightly against the hull during air shipment
and as a test for leaks.
Table 4. Float Coefficients
References:
Bower, A.S., L.Armi, and I.Ambar, 1997. Lagrangian observations of meddy form-
ation during A Mediterranean Undercurrent Seeding Experiment. J.Phys. Oceano.,
pp.2545-2575.
Hansen,D.V., and A.Herman, 1989. Temporal sampling requirements for surface
drifting buoys in the Tropical Atlantic. J.Atmos.Oceanic Tech., Vol.6, pp.599-
604.