如何将数组保存为 NetCDF 文件？

Question

Another user helped me to build the following code which works perfectly. With this code, I have an array with the sum_values (conditionally) of my initial data for each grid and year. I want to save the "summed_values_per_year" array as a NetCDF file. I tried multiple ways to do it, however, I am new to Python and I didn't make it.

Please share if you have any ideas and let me know what else I have to provide, like the attributes of an object. Thank you very much in advance.

file = netCDF4.Dataset ('/mnt/data/rcp45/tas/merged_rcp45/summed_values_per_year')

lat  = file.variables['lat'][:]
lon  = file.variables['lon'][:]
data = file.variables['tas'][:]

summed_values = np.zeros((14,16))

for grid_index, grid in enumerate(data):
    for lon_index, lon_column in enumerate(grid):
        for lat_index, value in enumerate(lon_column):
            if  0 < value and value < 25:
                summed_values[lon_index][lat_index] += value

n_years = data.shape[0] // 365
summed_values_per_year = np.zeros((n_years + 1, 14, 16))

for grid_index, grid in enumerate(data):
    for lon_index, lon_column in enumerate(grid):
        for lat_index, value in enumerate(lon_column):
            if  0 < value and value < 25:
                year = grid_index // 365
                **summed_values_per_year** [year][lon_index][lat_index] += value

Answer 1

I would take a look at xarray . Specifically, see the docs on creating a dataset , and then on reading and writing files .

You can directly create an xarray dataset from summed_values_per_year(year, lon, lat) like this:

ds = xr.Dataset({
    'summed_values_per_year': xr.DataArray(
        summed_values_per_year,
        dims=['year', 'lon', 'lat'],
        coords=[range(nyears), lon, lat],  # using the variables you defined
    ),
})

ds.to_netcdf("some/output/filepath.nc")

That said, it looks like what you're doing is to first mask the data to only include the values between 0 and 25, then group by year and sum. This can be done really easily in xarray:

ds = xr.open_dataset(
    '/mnt/data/rcp45/tas/merged_rcp45/summed_values_per_year'
)

summed_values_per_year = (
    ds.where((ds.tas > 0) & (ds.tas < 25))
    .groupby('time.year')
    .sum(dim='time')
)

summed_values_per_year.to_netcdf("some/output/filepath.nc")

See the xarray docs on computation for more info.