Observations and Simulations of Upper-Ocean Response to Wind Events during the Ocean Storms Experiment

Abstract

The Ocean Storms dataset is used to compile observations of the oceanic response to midlatitude storms. Of particular interest are episodic mixed layer temperature cooling events whose characteristics are reviewed. The data include subsurface temperatures from drifting thermistor chains, mixed layer temperature and velocity from mixed layer drifters, conductivity-temperature-depth profiles, and radiation measurements from ships, and the surface meteorological parameters produced by the European Centre for Medium-Range Weather Forecasts. A method for processing irregular drifting buoy position fixes to yield estimates of the geostrophic, ageostrophic, and inertial mixed layer currents is developed and shown to yield residuals that can mostly be attributed to errors in the positioning. From these currents the ocean's dynamic responses, namely, the change in mixed layer inertial kinetic energy and the ageostrophic particle displacement, are computed. The process of removing horizontal and vertical advection and surface heating from potential energy and mixed layer temperature responses is described. Temperature change responses are shown to be related to inertial current generation. Large responses. including episodic cooling, are found to be forced not necessarily by large storms, but by storms whose wind stress vector rotates inertially. The observations suggest that the phase of preexisting inertial currents may modulate the responses. The spatial scale of response to one particular storm is found to be about 150 km. The compiled dataset is also used to provide the initial conditions and the surface forcing required to run three one-dimensional numerical models of ocean vertical mixing. All three models are shown to qualitatively exhibit the observed behavior, including episodic cooling. Quantitatively, all the models predict the dynamic responses well, considering the uncertainty in the wind stress forcing. However, one model, a nonlocal K-profile parameterization of the oceanic boundary layer, is found to be somewhat better in reproducing the observed vertical profile of temperature change. This model's success is due to its more realistic exchange of mixed layer water with water from much deeper in the thermocline. In particular, the deepest extent of this exchange is accurately observed and well simulated.