First support for ICESat GLAH06

[alex-s-gardner]

@evetion

Here is the first crack at a GLAH06 reader. Two things you might be able to help with:

1. What is your approach to making nice ascii tables for the function help. I found formatting by hand tricky and didn't manage to get it all

2. ICESat elevations need to be converted from the TOPEX/POSEIDON to WGS84 the ellipsoid. Do you know of a good way to do this in Julia?

[evetion]

That was quick! Nicely done. With regards to your points:

1. I use https://www.tablesgenerator.com/markdown_tables, you can paste in anything from Excel or similar in there and then export markdown. Any values in the markdown with _ (such as field definitions), need to be escaped with \.
2. This is a bit of a pain to get right, but I use PROJ to transform from ellipsoid to geoid. There's a function for that (egm2008!). PROJ can also be used to transform between ellipsoids, but this example should be tested and validated, it's very easy to make mistakes here.

Generate a pipeline with PROJ:
❯ projinfo -s "+proj=latlon +a=6378136.299999999813735 +b=6356751.600562936626375 +rf=298.257 +e=0.08181922146 +towgs84=0,0,0 +type=crs" -t EPSG:4979 --3d

Then use it in Julia:

julia> pipe = Proj.proj_create("+proj=pipeline +step +proj=unitconvert +xy_in=deg +z_in=m +xy_out=rad +z_out=m +step +inv +proj=longlat +a=6378136.3 +rf=298.257 +e=0.08181922146 +step +proj=cart +a=6378136.3 +rf=298.257 +step +inv +proj=cart +ellps=WGS84 +step +proj=unitconvert +xy_in=rad +z_in=m +xy_out=deg +z_out=m +step +proj=axisswap +order=2,1")
Ptr{Nothing} @0x00000001254061d0

julia> Proj.proj_trans(pipe, Proj.PJ_FWD, (0,0,0))
4-element Proj.Coord:
  0.0
  0.0
 -0.7000000001862645  # <-- so 70cm lower it seems
 Inf

[alex-s-gardner]

@evetion

Once you look this over it should be good to merge

[evetion]

Can you also write an integration test? See test/runtest.jl for the current suite. Would be good to test that the points function works and returns some actual rows, you can copy the GLAH14 test.

To make the reference file available, add a line like

download_artifact(v"0.1", "GLAH06_634_2131_002_0084_4_01_0001.H5")

[evetion]

This doesn't work on arrays right? You might need a broadcast and a zip here like: Proj.proj_trans.(pipe, Proj.PJ_FWD, zip(latitude,longitude, height_ref))

[alex-s-gardner]

Can you also write an integration test? See test/runtest.jl for the current suite. Would be good to test that the points function works and returns some actual rows, you can copy the GLAH14 test.

Now implemented

[alex-s-gardner]

all requested fixes and changes implemented

[evetion]

I've fixed some small things and removed the following lines

latitude = [x[1] for x in pts]::Vector{Float64}
latitude = [x[1] for x in pts]::Vector{Float64}

as they should be unchanged from their input (only vertical transformation) and allocate extra memory. For the testfile, it seems all shots have quality=false. We might be just unlucky with the granule, but it's something you might check with other granules.

Otherwise I'm happy to merge this.

[alex-s-gardner]

Thanks for the fixes...

I think there are 1.0e-06 differences in latitude because of the slight differences in ellipsoid so it might be best to keep, I'll add back in.

latitude = [x[1] for x in pts]::Vector{Float64}

Let be check on the quality masking

[alex-s-gardner]

Broadcast was missing in quality flag:
(i_numPk .== 1) .&

[evetion]

Good catch in the quality broadcast, missed that completely.

Do you expect differences in the latitude for an ellipsoid transformation? What we pass in are float32s, so they're only accurate up to 7 decimals which would be errors of centimeters or less.

[alex-s-gardner]

🤞 we're good to merge

[alex-s-gardner]

Yes, I expect very small differences in the latitude for an ellipsoid transformation (at least I think I have that right).. longitude is unchanged.

But I don't think it matters at all for ICESat data so I'll remove that again

[alex-s-gardner]

latitude = [x[1] for x in pts]::Vector{Float64} is now commented out

[alex-s-gardner]

@evetion you should be good to merge.. then I can start on other items