Training and Test Datasets, Pretrained Weights and Predictions for HIDRA-D

HIDRA-D is a novel, state-of-the-art deep neural network model designed for dense sea surface height (SSH) prediction. The model leverages historical sea-level observations and tidal and geophysical forecasts. Published data contains HIDRA-D pre-trained weights, dense predictions for the test period, calculated as the ensemble mean of the forecasts generated from 50 atmospheric ensemble members, SLA measurements, normalization parameters, the geophysical training and testing datasets, and SSH measurements from Koper, Sobra, Stari Grad and Vela Luka.