Sampling characteristics from isobaric floats in a convective eddy field

During the recent Labrador Sea Deep Convection Experiment, numerous isobaric floats were deployed. Interpretation of the quasi-Lagrangian measurements from these floats requires an understanding of any biases that may be introduced by the response of the floats to the flow in which they are embedded. To investigate the float measurement biases in convecting flow numerical simulations of isobaric floats in a domain containing several mesoscale eddies have been performed. When a surface heat loss is applied, spatially variable convective mixing and baroclinic instability result. The authors find that without surface cooling, probability density functions of Eulerian and isobaric float measurements of tracers and velocities are very similar, given an initial distribution of isobaric floats that is random with respect to the initial features of the tracer field. However, with cooling isobaric statistics are biased compared to the Eulerian statistics. In particular, in near-surface regions isobaric floats appear to oversample regions of dense downwelling fluid. Since in near-surface layers downwelling dense fluid is associated with convergence, a probable explanation of the isobaric float biases is a tendency for floats to concentrate in regions of horizontal convergence. Also, with cooling floats may be more easily exchanged between eddies and the ambient fluid. The escape of a float from an eddy can be identified from changes in the values of material tracers. The authors identify a positive skewness in the time derivative of the buoyancy measured by the individual near-surface floats as a indicator of convection in the presence of mesoscale eddies.