The coastal heights data combines two datasets: the land-sea mask of the gridded EMODnet bathymetry and the height of high-water (HAT) relative to chart datum (LAT) as computed with the GTSM tide model. Near the coast a diffusion scheme is applied to make the heights consistent with the land-sea mask. Finally, only grid cells that are classified as sea, but neighbor a land cell are selected. The main assumption in this approach is that the height of the topography near the coastline is close to the high-water mark. The main purpose of this dataset is to improve extrapolation towards the coast. In many areas around Europe there is a (small) gap between the point closest to the coast for which a survey is available and the coastline. In this way interpolation from coarser background data is avoided.