The majority of the code in our examples is used to create input datasets. We need to find a better way to handle example/input data.
We could do something with pooch, like xarray, and a curated, cleaned subset of input data.
I'm not sure. Some of our data gets pretty gnarly. I think part of what makes an isku approach nice is that large data can be streamed from local storage or public cloud buckets. It would be nice to showcase that a bit more.
I think this will also be a problem when we start to populate examples/ (#4).
We can store moderatly size data with:
Using https://planetarycomputer.microsoft.com/dataset/cil-gdpcir-cc0 would be cool.
The majority of the code in our examples is used to create input datasets. We need to find a better way to handle example/input data.
We could do something with pooch, like xarray, and a curated, cleaned subset of input data.
I'm not sure. Some of our data gets pretty gnarly. I think part of what makes an isku approach nice is that large data can be streamed from local storage or public cloud buckets. It would be nice to showcase that a bit more.
I think this will also be a problem when we start to populate examples/ (#4).
We can store moderatly size data with:
under a DOI and read it with pooch.
Using https://planetarycomputer.microsoft.com/dataset/cil-gdpcir-cc0 would be cool.