You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Explorer has access to both the original .yaml location (in an s3 bucket alongside the raster files) and the API endpoint symlink http://explorer/dataset/<uuid>. If the STAC odc_yaml field provided the former, instead of the latter, it would facilitate easier indexing.
Trying to index the above "ODC Dataset YAML" (e.g. using datacube dataset add) leads to failures (not on index but later upon API data access), because the yaml only contains a relative reference to the .tif imagery location, and that relative reference is only valid for the original yaml location not the explorer API endpoint location that is referred to here.
The "ODC Dataset Overview" above is a html page containing (in its dataset-location div) a hyperlink to the original yaml path:
It looks as if indexing this yaml path should work fine (the relative path to the raster data will be valid).
If the "ODC Dataset YAML" (odc_yaml) stac field above instead expressed this yaml path then I think it would be quite straight-forward to index from stac searches:
(The stac json does contain an absolute reference to the image asset, so this would also be solved if the datacube CLI supported indexing stac json. Alternatively, may also be solvable by modifying the yaml files to include absolute location fields, e.g. upgrading to EO3 metadata format.)
The text was updated successfully, but these errors were encountered:
(@jeremyh notes that this proposal may not have worked in general, as the yaml sourced by explorer could be embedded in a netcdf rather than a simple location, hence the solution to inject absolute references into the yaml - as was already being done for EO3 datasets.)
Explorer has access to both the original .yaml location (in an s3 bucket alongside the raster files) and the API endpoint symlink
http://explorer/dataset/<uuid>
. If the STACodc_yaml
field provided the former, instead of the latter, it would facilitate easier indexing.Using the stac api to query for a EO dataset:
https://explorer-aws.dea.ga.gov.au/stac/search?collections=wofs_albers&datetime=2000-01-01T00:00:00Z/2001-01-01T00:00:00Z&bbox=137.01,-28.76,137.02,-28.75&limit=1
The result looks like:
Trying to index the above "ODC Dataset YAML" (e.g. using
datacube dataset add
) leads to failures (not on index but later upon API data access), because the yaml only contains a relative reference to the .tif imagery location, and that relative reference is only valid for the original yaml location not the explorer API endpoint location that is referred to here.The "ODC Dataset Overview" above is a html page containing (in its dataset-location div) a hyperlink to the original yaml path:
https://dea-public-data.s3.ap-southeast-2.amazonaws.com/WOfS/WOFLs/v2.1.5/combined/x_4/y_-32/2000/01/01/LS_WATER_3577_4_-32_20000101003706500000_v1526711988.yaml
It looks as if indexing this yaml path should work fine (the relative path to the raster data will be valid).
If the "ODC Dataset YAML" (
odc_yaml
) stac field above instead expressed this yaml path then I think it would be quite straight-forward to index from stac searches:(The stac json does contain an absolute reference to the image asset, so this would also be solved if the datacube CLI supported indexing stac json. Alternatively, may also be solvable by modifying the yaml files to include absolute location fields, e.g. upgrading to EO3 metadata format.)
The text was updated successfully, but these errors were encountered: