Possibility of adding longitude/latitude of grid point to NCSS output

[marceloandrioni]

The longitude and latitude returned by the NCSS (NetcdfSubset) are the ones informed by the user. I was wondering if is possible to add a flag to the url query (something like returnGridLatLon=true) so that the lat/long of the grid point is returned instead, or maybe to return both the user and grid point coordinates (two extra columns). For some applications like selecting winds near-shore it is important to know if the grid point selected is overland or water and I think this option would facilitate this process. I know it is possible to use the dods protocol to get the whole grid and find out the grid point nearest the user location, but to be able to do all this with NCSS would really be great.
Thank you.

[lesserwhirls]

Interesting, thank for bringing this to our attention! It looks like all of the return formats, even netCDF, use the requested lat/lon values, and not the actual lat/lon values from the nearest point on the grid. Rather than adding a keyword, I think both values should be returned by default.

[lesserwhirls]

@cwardgar - what do you think?

[cwardgar]

I think it's definitely worthwhile to include the lat/lon of the nearest centroid, as that's the location that the rest of the data in the response applies to.

In fact, is it desirable to include the user's requested lat/lon in the response at all? Can we just replace it with the centroid?

[lesserwhirls]

It might be nice to be able to see how far off the request was since the downloaded file may be used at a later date when the user cannot remember what request was actually made.

[marceloandrioni]

I am glad to know the possibility of including the grid point lat/lon will be considered in new versions of TDS. In my opinion the ideal case would be to offer the user both his and the grid point coordinates, so that, as said by @lesserwhirls, in the future the user can be sure what he requested and what the TDS replied. Second best case scenario would be only the grid point lat/lon.

[dopplershift]

I think including the requested grid lat/lon is a nice enhancement to NCSS; the fact that this is returning the requested values as if they were those for the data, though, is a flat out bug. It’s a huge semantic difference from what NCSS does for grids.

[marceloandrioni]

Hello, not sure if I should comment here or in the Unidata/tds, but testing TDS beta 7 in the thredds-test I saw that now a ncss point query like this one returns an extra column with station name like "GridPointAt[lat_lon]", e.g.:

time,station,latitude[unit="degrees_north"],longitude[unit="degrees_east"],Significant_height_of_combined_wind_waves_and_swell_surface[unit="m"]
2019-08-04T00:00:00Z,GridPointAt[24.123S_39.987W],-24.123,-39.987,2.6
2019-08-04T03:00:00Z,GridPointAt[24.123S_39.987W],-24.123,-39.987,2.38
2019-08-04T06:00:00Z,GridPointAt[24.123S_39.987W],-24.123,-39.987,2.33

Just wondering if this will be the new format for ncss point queries and if the possibility of adding a column (or attribute in the netcdf) with the actual used grid point location (and not just the user requested location) is still being considered. Maybe the station name ("GridPointAt[lat_lon]") could hold the actual values of the grid point.

Thank you.