The main use case in Sedona would be to write raster files in other formats than Geotiff.
The binary file data source in Spark makes it easy to support reading different raster formats in Sedona. We only need to implement a function to convert the binary content to a raster.
Spark doesn't offer a binary data source that supports writing. That could be implemented in Sedona.
The benefits to Sedona would be:
- Sedona would be able to support writing different raster formats by implementing different RS_As[ImageFormat] functions instead of creating a custom data source for each format. Like https://postgis.net/docs/RT_reference.html#Raster_Outputs
- Users would be able to transport rasters in common formats to other data stores like RDBMS.