[gdal-dev] best tiff tag names for start and end datetimes?

[Even Rouault]

Simon,

metadata is a tricky subject. To my knowledge, up to know most data 
producers of GeoTIFF files have used minimalist embedded metadata in the 
TIFF file itself (mostly using the ImageDescription, Copyright field), 
but rather using text or xml side car files to provide their own 
metadata model. See 
https://github.com/OSGeo/gdal/tree/master/gcore/mdreader for a number of 
metadata importers. Those metadata importers expose the "raw" metadata 
from the side car files, but also try to gather and expose a very 
minimum set of information in a IMAGERY metadata domain: 
https://gdal.org/user/raster_data_model.html#imagery-domain-remote-sensing

The TIFF tag mechanism makes it impractical to add "arbitary" metadata 
keys beyond what the baseline TIFF spec defines. So an option is to use 
a generic "container" like the GDAL_METADATA tag. Other options could be 
the GEO_METADATA one 
(https://www.awaresystems.be/imaging/tiff/tifftags/geo_metadata.html 
using a full ISO metadata model). Someone on the TIFF mailing list also 
mentioned the possibility of using XMP 
(https://www.awaresystems.be/imaging/tiff/tifftags/xmp.html). From a 
pure GDAL perspective, the GDAL_METADATA one is the most straightforward 
as this is what will be always reported by a gdalinfo dump for example.

I feel an initiative gathering a representative set of GeoTIFF 
geospatial products would make sense to define standardized metadata (to 
start with decide which existing TIFF tag should be the holder of the 
metadata, and then standardize metadata items) would make sense with COG 
growing adoption (possibly inspired by what STAC provides), rather than 
pushing quickly something for ad-hoc needs. Not sure which forum would 
be the best for that. Probably some dedicated github repository.

I'd see a first metadata item that would make sense, a METADATA_PROFILE 
/ METADATA_CONVENTION that would be a URL to a spec defining 
standardized metadata (similar to the Conventions item of the netCDF CF 
spec)

Regarding your need for start/end date times, I can already see an 
ambiguity: for some use cases, it could be understood as the start/end 
of acquisition of the scene, and for others (like weather forecast like 
you pointed in the initial TIFF discussion), or the start/end of 
validity of the information.

Even

Le 13/05/2022 à 01:35, Simon (Vsevolod) Ilyushchenko a écrit :
> (Reposting from the tiff list on Even's suggestion)
>
> Hi,
>
> I work on the Google Earth Engine team. Our users are asking about a 
> standard way to set start/end time on geotiffs exported from Earth 
> Engine. I know about TIFFTAG_DATETIME, but it has no timezone. In any 
> case, we'd like to use two different time tags for start and end.
>
> What would be the most standards-compliant way of setting datetimes?
>
> Thanks,
> Simon
>
> _______________________________________________
> gdal-dev mailing list
> gdal-dev at lists.osgeo.org
> https://lists.osgeo.org/mailman/listinfo/gdal-dev

-- 
http://www.spatialys.com
My software is free, but my time generally not.