I was wondering about the possibility of performing queries based on vector Geometries, but I didn’t see that mentioned anywhere in the geospatial documentation, nor does it seem that ogrinfo recognises the TileDB format.
I can understand why TileDB would be more geared towards raster operations, but I just wanted to confirm whether it was or was not possible.
We do have TileDB sparse array support in the PDAL project for read/write point operations.
However we currently don’t have OGR support, with the unified GDAL/OGR driver API it is something that we are thinking of adding.
Can you describe your use case and we will look into how we can help you.
We have weather data (values on a point grid over time), satellite rasters and vector polygons, and use the polygons to query portions of the former. For example, we might need to subset a raster image based on a polygon.
It’s not a problem for the raster images, because gdal can crop raster data based on external vector data, and we don’t need to store those vectors in TileDB.
We’ve been using ogr for filtering the vector points of the weather data, but I’ll look into PDAL, since that data is strictly limited to points. It looks like that can achieve something similar.
@michael.raymond we have been working with weather data a fair bit, there is a blog post about the use of TileDB here - https://medium.com/informatics-lab/storing-cloud-ready-geoscience-data-with-tiledb-34d454c33055
In addition to PDAL we do have multiple language integrations with TileDB including Python and
TileDB-Py (https://github.com/TileDB-Inc/TileDB-Py) combined with Fiona (https://fiona.readthedocs.io/en/latest/manual.html) would give you the OGR functionality you are familiar with.
Please let us know if you have any more questions as you explore the use of TileDB.