y dimension (latitude) not found

[shuai-zhou]

I was processing NDVI data (one of the files can be downloaded here). The data looks like the following:

However, when I was writing the crs using ds.rio.write_crs('epsg:4326', inplace=True). I got the error MissingSpatialDimensionError: y dimension (latitude) not found. Data variable: lon_bnds. I googled and tried to fix this but did not make it.

[snowman2]

Did you use decode_coords="all" when opening the file? It looks like it already has a crs coordinate that may already have the necessary information.

https://corteva.github.io/rioxarray/stable/getting_started/getting_started.html

[snowman2]

Also, what version of rioxarray are you using?

[shuai-zhou]

Hi, I was using rioxarray version 0.7.1. While decode_coords="all" solved the problem above, another error occurred when I was clipping the gridded data using clipped = ds.rio.clip(poly4326.geometry.apply(mapping), poly4326.crs). The error is ValueError: Could not convert object to NumPy datetime. But the time is already in datetime64[ns] format...

[snowman2]

See #459. Might be worth a separate discussion.

[shuai-zhou]

Thanks for the help. Yes, it turns out that there are missing values in the ds; the missingness is not on the time variable but on TIMEOFDAY and QA. Dropping the two variables solve the problem.