[THREDDS #UIE-535383]: Questions re THREDDS and GOES-R ABI L1B products

David,

Thanks for contacting us. See replies inline below:

> GOES-R ABI L1b/L2 full disk and CONUS products are on the ABI fixed grid
> projection. It is my understanding that Unidata has added support for
> this projection (i.e. a Java library), at least to convert a point on
> the fixed grid to a lat/lon and visa-versa. What I would like to find
> out is whether THREDDS can:
> 
> 1)  Subset/sectorize these products based on lat/lon corner points.

THREDDS has the NetCDF Subset Service (NCSS, see: 
https://www.unidata.ucar.edu/software/thredds/current/tds/reference/NetcdfSubsetServiceReference.html)
 which allows subsetting a dataset using lon/lat and time coordinates, as well
as only asking for specific variables.

> 2)  Change the projection of these products on the full disk projection
> to tropical mercator, lambert conformal, or polar stereographic.

This is not supported on the thredds server--it only provides services in the 
data's native coordinates, lon/lat, or index-based subset. Reprojection is 
currently a task left to the client.
We are investigating adding some generic server-side processing capabilities, 
of which this is a classic example, but this is very early investigations, and 
there is no timeline on when such features might appear in a release.

> 3)  Down-sample these products.

NCSS supports specifying a stride for both horizontal spatial dimensions and 
time. This means you can get, for example, every 3rd point.

> 4)  Change the bit depth of the products' pixels.

THREDDS does not do this, nor is it currently planned.

> 5)  The products contain data quality flag "images" that mirror the fixed
> grid data arrays. Will the functions for 1) and 2) be available for
> these data quality flag images?

Assuming the data quality flag images are NetCDF4 variables, then NCSS will 
work on these as well.

> 6)  Will the CF-compliant geolocation metadata be modified appropriately
> when functions 1) or 2) above are performed ?

Yes, NCSS writes out CF-compliant files in general.

> 7)  Will the product level metadata fields (i.e. scalar variables, such
> as number of good pixels in the image, be updated when functions for
> 1) an d 2) above are performed?

If I understand correctly, this metadata is not part of CF and is specific to 
the GOES-R ABI formatted files. As such, THREDDS would not currently handle 
this, nor is this currently on the roadmap.

> 8)  The standard names used for the coordinate variables in the native
> GOES-R ABI L1b/L2 products are "projection_x_coordinate" and
> "projection_y_coordinate". The units are supposed to be meters, but
> the GOES-R product designers "cheated" and is using these standard
> names with units of "radians".  Is this going to cause a problem?

I'm not sure. My gut says no, but that it will be straightforward to make it 
work. Such a problem is likely to be addressed once we start working with some 
real data.

> 9)  The other question I have is whether IDV has been revised to support
> displaying GOES-R ABI L1b/L2 on the ABI fixed grid projection.

For data accessed from THREDDS, the IDV will rely on the same netcdf-java code 
that power THREDDS.

> 10) Lastly, if THREDDS is not currently capable of any of the functionality
> described above, do you have plans to develop such functionality, and,
> if so, when?

All I can add is that we have plans to have a GOES-R receiver here. I'm sure 
once we have the data flowing we'll be smoothing the process of serving and 
processing the GOES-R data using THREDDS. This is especially true of problems 
that affect our internal clients (like IDV and THREDDS).
I can't promise, however, that all of the issues discussed above will be solved.

Ryan


Ticket Details
===================
Ticket ID: UIE-535383
Department: Support THREDDS
Priority: Normal
Status: Closed