bekaiser/TensorOcean

Ocean turbulence prediction using Tensorflow for training Mixture Density Networks with the UCSD microstructure database.

Deep learning for the prediction of the turbulent dissipation rate of kinetic energy in the ocean

Tensorflow is used to train mixture density networks for predicting the dissipation rate of turbulent kinetic energy from conservative temperature, absolution salinity, the square of the buoyancy frequency, and the location in the water column. The UCSD microstructure database (https://microstructure.ucsd.edu) is used for training.