Adding SURFRAD Capabilities

[AdamTheisen]

• Tests added
• Documentation reflects changes
• PEP8 Standards or use of linter
• Xarray Dataset or DataArray variable naming follows 'ds' or 'da' naming

[AdamTheisen]

Looks like there are some errors with the PySP2 tests

[mgrover1]

@AdamTheisen it appears that there is some 0 shaped array being multiplied within the try/except block...

if np.isfinite(base):
            try:
                height = data_fit.max() - base
                pos = np.argmax(data)
            except ValueError:
                height = np.nan
                pos = np.nan
    
            try:
                start = np.where((data - base) < 50)[0]
>               if start == []:
E               ValueError: operands could not be broadcast together with shapes (14,) (0,)

[zssherman]

Nothing changed within the pysp2 repo itself, hmm so curious why it's failing now

[mgrover1]

Nothing changed within the pysp2 repo itself, hmm so curious why it's failing now

pysp2 does not currently have a nightly build running/testing with dependencies, so I suspect is has to do with some upstream library cutting a new release

[mgrover1]

@zssherman I suspect numpy is the culprit here... https://github.com/numpy/numpy/releases/tag/v1.25.0

They cut a new release two days ago, and there are mentions about deprecated masking functionality/changes with array shapes.

[zssherman]

@mgrover1 @AdamTheisen Looks like we need a change upstream. So when it checks the start for empty list, check if its a numpy array

try:
    start = np.where((data - base) < 50)[0]
    if start == []:
        start = np.nan

[mgrover1]

@mgrover1 @AdamTheisen Looks like we need a change upstream. So when it checks the start for empty list, check if its a numpy array

try:
    start = np.where((data - base) < 50)[0]
    if start == []:
        start = np.nan

Upstream as in pysp2?

[zssherman]

@mgrover1 I believe so

[mgrover1]

Ahh the good old method=nearest argument - good catch @AdamTheisen !

[AdamTheisen]

Ahh the good old method=nearest argument - good catch @AdamTheisen !

Found another issue with the plot and the text showing up as expected while looking so it was a good error to help improve the example! There was also a deprecation warning with the fix I had put in so had to redo that.

[mgrover1]

@AdamTheisen - does their documentation include anything about the standard_name for the variables? I think that would be helpful include here if that is the case.

[AdamTheisen]

@AdamTheisen - does their documentation include anything about the standard_name for the variables? I think that would be helpful include here if that is the case.

The only documentation they really have is what's in the README file. They don't have anything about standard_names from what I can see.

[mgrover1]

@AdamTheisen - does their documentation include anything about the standard_name for the variables? I think that would be helpful include here if that is the case.

The only documentation they really have is what's in the README file. They don't have anything about standard_names from what I can see.

Gotcha - any chance you could take a look at this table
https://cfconventions.org/Data/cf-standard-names/current/build/cf-standard-name-table.html

And see which ones would map? One benefit of using a cf-standard name for as many variables as possible is the ability to use tools such as cf-xarray to generically compare different datastreams using these community standards...

[zssherman]

Looks good to me, tests are now passing!

[AdamTheisen]

Just as a note, I did not get a chance to add in standard names yet. Was hoping to look into that tomorrow

[mgrover1]

Sounds good - I think that makes sense for separate PR 😄