plot global variable with nans

[raybellwaves]

The default .plot() function seems to struggle with global variables that have nans.

Here's is a file with u10 (u component of wind at 10 m) and swh (significant wave height)

out.nc.zip

import xarray as xr

ds = xr.open_dataset("out.nc")

ds["u10"].plot()

ds["swh"].plot()

AFAICT only difference is swh has NaNs

Same behavior with pcolomesh

ds["swh"].plot.pcolormesh("longitude", "latitude")

[keewis]

there seems to be something wrong with your data:

ds.swh.where(ds.swh.notnull(), drop=True).plot()

results in the expected plot, but the latitude and longitude coordinates cover only 0 to 180 degrees on the northern hemisphere (despite the plot covering the whole globe).

Maybe there's something wrong with the processing steps producing this dataset?

[raybellwaves]

thanks. I'll take a look.

FYI source is similar to...

import cdsapi

c = cdsapi.Client()

c.retrieve(
    "reanalysis-era5-single-levels",
    {
        "product_type": "reanalysis",
        "variable": [
            "10m_u_component_of_wind",
            "significant_height_of_combined_wind_waves_and_swell",
        ],
        "year": "2020",
        "month": "01",
        "day": "01",
        "time": "00:00",
        "format": "netcdf",
    },
    "/tmp/download.nc",
)

[raybellwaves]

Think I got it. Wind and waves are at different resolution in ERA5 so think it does something funny when you request them together.