HydroMT support for Delft-Dashboard data from openDAP server

[DirkEilander]

see https://github.com/Deltares/CoastalHazardsToolkit/blob/1290147cb77d18446a6dc4bf37d132ddc564a56a/src/cht/bathymetry/bathymetry_database.py#L1

[DirkEilander]

test for gebco22

compare nc vs COG formats tile overview in netcdf (DD style) and vrt xml formats. Check overview of xyztiles

TODO

• [ ] download gebco 22 @B-Dalmijn
• [ ] share script for tiling @maartenvanormondt
• [ ] build overview netcdf @maartenvanormondt
• [ ] test if tiles can easily be converted to GDAL compliant nc files @B-Dalmijn
• [ ] build vrt overview

[DirkEilander]

It seems all static tiling methods reproject data to webmercator and save data to image formats such as png/jpeg etc. Using cloud optimized geotiffs (COGs) you can dynamically serve map tiles. These files have an internal organization that is optimized for http get requests and are ideally (but not necessarily) in webmercator and with internal overviews (zoomlevels).

The gdal_retile.py routine can create tiles using z/y/x scheme in any support gdal raster format with different resampling methods and optionally write tileindex files (shape) and csv files per zoomlevel. The script does not support average resampling (gdal does!) and does not write toplevel metadata.

conclusion:

• we need to agree on a tiling scheme (see options below, the bing quadtree seems interesting!)
• we need to define a toplevel metadata file
• we need a script to create the tiles and save in required format (gdal compliant nc is fine). This can be adapted from maarten's script or this SO examples
• we can either create a vrt per zoomlevel for out of the box reading or create a custom script (or both)

references

• data-driven maps https://carbonplan.org/blog/maps-library-release
• different tiling schemes: https://documentation.maptiler.com/hc/en-us/articles/4405435316369-Tiling-scheme-in-Folder-output