Test tiled data set

[lisakaser]

Good sample data set: IRTIT3 (it has tiles and is in netcdf - CF but not NSIDC compliant)

Acceptance criteria:

Confirm correct spatial in UMM-G

[lisakaser]

From Amy: /disks/sidads_staging/SAMPLE_DATA/DUCk_project/IRTIT3_DUCk holds the copy of data for us to work with.

[juliacollins]

@afitzgerrell the geolocations for each test file are quite close to what I see in Earthdata Search, although I don't think they're precisely the same. The EDS versions are represented with a four-cornered polygon, whereas our current code produces output with about twenty polygon points. The other item of interest is that each metadata file in EDS has start and end time values of 10:00:00 and 18:00:00 (on each of their respective dates). I'm guessing those were specified manually (as part of SIPS MetGen?), but is it possible to confirm that guess? I see nothing in the netCDF file pertaining to time of data observation.

[afitzgerrell]

Hi @juliacollins. I can confirm that I can't find any source of the temporal range that's specified in EDSC for each of the granules. I think they were perhaps made up for the sake of ingesting them with something. I tend to assume that each glacier area was an individual flight during which the tomographic survey data was collected and considering that, there's no way all of the granules would in reality have the same start and end times!!

As for the tile footprints, they are not precisely the same (at least not for the Pine Island Glacier = PIG 🐷 ). I've attached a screen grab of what I'm seeing when I plot them all manually together. It's interesting because there's:

1. the actual [mostly] black grid extent when I view the data file and display no-data values in addition to the data collected along the flight line,
2. the pink dots which I plotted from the geojson lat/lon values you generated,
3. and the green outline is the polygon I plotted using the coordinate pairs from the Pine Island Glacier's (lovingly= PIG) EDSC metadata.

There's what looks to be a pretty uniform offset between the points you generated and the data grid itself.

Manually, I can query for the corner extents from the grid itself, and those, in PS EPSG:3031 are:
-1761112.5, -417912.5
-1761112.5, -47087.5
-1489212.5, -47087.5
-1489212.5, -417912.5
-1761112.5, -417912.5

...which, if I use gdaltranslate to spit out epsg:4326 dd coords are:
-103.349386172016 -73.4525412979783
-91.5315730213014 -73.8882534793254
-91.8110353378813 -76.3492762975119
-105.675527508388 -75.8340220150565
-103.349386172016 -73.4525412979783

I don't know if any of this will help in troubleshooting, but, these are my observations for what they're worth!

[afitzgerrell]

hmm, in hindsight, I could have thought to do this first, but why walk the straight line between two points!! It dawned on me if I export one of the PIG file's variables and make it a tiff, I can use gdalinfo to tell me everything it knows about the grid.
This is its report of the spatial details (I omitted the CRS details since those aren't what's in question in the netcdf):

Files: PIG_IT.tif
Size is 10876, 14833
Data axis to CRS axis mapping: 1,2
Origin = (-1761112.500000000000000,-47087.500000000000000)
Pixel Size = (25.000000000000000,-25.000000000000000)
Corner Coordinates:
Upper Left (-1761112.500, -47087.500) ( 91d31'53.66"W, 73d53'17.71"S)
Lower Left (-1761112.500, -417912.500) (103d20'57.79"W, 73d27' 9.15"S)
Upper Right (-1489212.500, -47087.500) ( 91d48'39.73"W, 76d20'57.39"S)
Lower Right (-1489212.500, -417912.500) (105d40'31.90"W, 75d50' 2.48"S)
Center (-1625162.500, -232500.000) ( 98d 8'29.90"W, 74d58'24.64"S)

again, maybe helpful? maybe interesting? mebbe not.

[juliacollins]

@afitzgerrell I take some small consolation in the fact that the pink dots seem to follow the black perimeter, except for the consistent and annoying shift. I'm going to ignore the green bounding box and its implications for our ability to trust the metadata in EDSC!