OceanDepths is a new global dataset that combines surface and subsurface ocean observations. It pairs satellite data with in situ measurements from the EN4 database, covering sea surface temperature, salinity, and height, as well as subsurface temperature and salinity profiles. The dataset spans 2000-2024 at a resolution of 0.1 degrees by 0.1 degrees and weekly temporal resolution. It has over 9.5 million paired profiles, making it a challenging testbed for AI methods. --- Why it matters: This dataset matters to researchers in AI because it provides a unique opportunity to develop novel methods for subsurface state reconstruction and observation-based forecasting. Its high resolution and long temporal extent make it an ideal testbed for testing the limits of current AI techniques. Source: https://arxiv.org/abs/2608.16373