Function to read ICGEM data files

[leouieda]

Description of the desired feature

The ICGEM Calculation Service is an amazing resource to get gravity data. Their grids are in an ASCII format that can be tricky to parse because some values are included in the header. It would be great to have a function that reads the ICGEM file and returns an xarray.Dataset.

This will really help with #9

[leouieda]

@santis19 would you mind porting fatiando/fatiando#421 to harmonica? There a few changes that will need to be done:

• Replace the asserts with proper if ... raise checks
• Use **kwargs to pass arguments to numpy.loadtxt instead of manually specifying usecols
• Return an xarray.Dataset instead of a dict. You can pass the entire header string as metadata.
• Probably no need to read in the latitude and longitude limits. But we do need to parse the grid shape from the header.

The function can go in a new harmonica/io.py module and then import it in harminca/__init__.py. This way we can use it as harmonica.load_icgem_gdf instead of having to load several modules.

[santisoler]

@leouieda I've just started working on this.
As soon I have a respectable code I'll PR.

I have one two questions: what about tests functions for this function?
Are you intending to leave tests functions for the future or to only merge PRs with 100% test coverage?

[santisoler]

We need xarray to be listed in environment.yml.
Do you want me to add it on the PR I'm currently working on?

[leouieda]

Are you intending to leave tests functions for the future or to only merge PRs with 100% test coverage?

Yep, test coverage is a must for merging. Github automatically bars merging if coverage of the diff is less than 80%. Leaving it for later is a recipe for never doing it, as I learned from fatiando.

what about tests functions for this function?

The fatiando version has some tests that you can use, if I'm not mistaken. But try to see if you can get 100% coverage. It's worth the exercise as you learn a lot about designing good code. For example, it might be best to try to allow a file name or a file object as input. This way we can use StringIO for testing.

Do you want me to add it on the PR I'm currently working on?

Please include it in the environment.yml, requirements.txt, setup.py, and doc/install.rst.