There are a lot of NoDATA values (-9999, 32765) present in the data in Track 1. It is present in all bands of training and validation images. For example, in file "o.tif", band 3 doesn't even have one good value. The whole band is -9999. Also, band 4 (DEM) has a value of 0, for the whole image. Is anyone facing these issues? Is it the same for track 2?
Posted by: navintony @ Feb. 28, 2024, 3:45 p.m.Looking at the location of the patch ( which is not provided to participants ), we can say that it’s located in a coastal area, on sea side, 100% water.
In such a case, nodata is indeed expected for elevation, we keep it as provided in the original products that are : -9999 for MERIT DEM and 0 for Copernicus DEM.
Best regards