Posted by Arthur Mariano on May 11, 2001 at 07:15:48:
I enjoyed this virtual poster because it contained a very
well-planned comparison between in-situ near-surface drifter
velocities and Lagrangian velocities simulated by a coarse-
and high- resolution POP model. Seeding numerical floats at
the same time and location of the real data and letting them
"drift" for the same amount time reduced a number of possible
biases in the comparison and the authors should be commended
for this part of the study.
The primary result of McClean et al.'s study, that an O(10 km) horizontal resolution model is needed for realistic simulations
of ocean circulation, agrees with results seen by other modellers.
Nevertheless, their Lagrangian-based comparison is an excellent
and the best quantitative demonstration of how much better an
OGCM performs at eddy resolving rather than eddy permitting
resoltuion. The use of daily winds, instead of averaged wind products, is another factor for the good agreement in first- and second-order velocity statistics. It would be interesting to see if the statistics of the trajectories would change when other wind products and if 6 hr winds are used. Lagrangian statistics of near-surface drifters should be more sensitive to the wind forcing than Eulerian statistics because of the chaotic nature of Lagrangian flow.
My only complaint is that the authors should have shown difference
fields. It is time-consuming to toggle between figures to compare.
The following questions for the virtual community were motivated
by this poster:
Lagrangian integral time scale values of 2-4 days have now been
computed from many near-surface drifter data sets from around the globe. Given the different forcing, inertial periods, MKE/EKE, stratification, etc., Why is the range of integral time scales
so small?? In other words, when ocean drifter data is properly
detrended, Why does Taylor's hypothesis hold so well even with
all its assumptions?
Integral time scales of two days imply at least daily sampling
to avoid aliasing based on mesoscale phenomena and more
frequent sampling to avoid aliasing by inertial periods. This
is a costly fact of life for those needed real-time operational
velocity products. The sampling reuirements for data with
errors of order 1-10% is about three times more severe than
the Nyquist criterion.
Calculation of dynamical quantities will require data sampling
at rates of one per hr for near shore regimes (assuming semi-diurnal tides are important) and 8 hrs in the ocean mixed
layer (assuming integral time scales of 2 days).