|
When opening different grib files using UCAR NetCDF Java libs you may get the same variable using different time dimensions across different files.
As an instance: file1.grib2 dump ============= air_temperature(time1, lon, lat) air_temperature_hourly_maximum(time2, lon, lat)
file2.grib2 dump ============= air_temperature(time2, lon, lat) air_temperature_hourly_maximum(time1, lon, lat)
(see the swapped time1, time2 dimension names). The name depends on the position of a specific GRIB record within the dataset itself. Since GRIB records are "sparse" records within a binary format the order the parameters are found can't be known in advance, resulting in this "timeX" dimensions "random" naming.
This results in issues populating a granules catalog since the dimension<->schema attribute mapping isn't respected anymore. Adding a flag to map all "timeX" dimensions to simply "time" attribute in DB will solve the issue.
|