cdflib

1.3.10 verified Fri May 01 auth: no python

A Python library for reading and writing NASA Common Data Format (CDF) files. Current version 1.3.10, released 2025-05. Maintained by LASP, supports Python >=3.9. Includes integration with xarray and handles ISTP compliance checks.

pip install cdflib

Common errors

error ModuleNotFoundError: No module named 'cdflib.cdf' ↓

cause Incorrect import path; CDF is a class in the top-level cdflib package, not a submodule.

fix

Use from cdflib import CDF instead.

error AttributeError: module 'cdflib' has no attribute 'cdf_to_xarray' ↓

cause Attempting to import cdf_to_xarray from a wrong location or missing install (e.g., without xarray dependency).

fix

Install xarray (pip install xarray) and import using from cdflib import cdf_to_xarray.

error TypeError: 'numpy.float64' object cannot be interpreted as an integer ↓

cause Passing a numpy float as start/count index to varget(). varget() expects Python ints.

fix

Cast to int: start=int(start_val) or use varget('var', start=[int(s) for s in start]).

Warnings

breaking In version 1.3.3, the deprecated functions `from_unixtime`, `from_datetime`, `unixtime_to_cdftt2000`, `datetime_to_cdftt2000`, and `datetime64_to_cdftt2000` were fully removed. Use `CDFtime` class or `astropy.time.Time` conversion instead. ↓

fix Replace `from cdflib.util import from_unixtime` with `CDFtime(utc).tt2000` or use astropy.

gotcha When using `cdf_to_xarray`, `DELTA_VAR` variables are no longer promoted to coordinates (since 1.3.5). If you rely on them as coordinates, you must manually add them via xarray. ↓

fix After calling `cdf_to_xarray`, do `ds = ds.assign_coords(delta_var=ds['delta_var'])` if needed.

gotcha The CDF class is not thread-safe. Do not share a CDF object across threads; open separate instances per thread. ↓

fix Use a new CDF instance per thread (or use read locks).

deprecated As of version 1.3.10, `CDF.varget()` with `start` and `count` parameters uses 0-based indexing. Some users mistakenly use 1-based indexing from IDL/old CDF standards. ↓

fix Always use 0-based indices: `cdf.varget('var', start=[0], count=[10])`.

Imports

CDF
wrong
```
from cdflib.cdf import CDF
```
correct
```
from cdflib import CDF
```
CDF is a top-level class in cdflib, not a submodule.
cdf_to_xarray
wrong
```
from cdflib.xarray import cdf_to_xarray
```
correct
```
from cdflib import cdf_to_xarray
```
cdf_to_xarray is a top-level function, not nested in an xarray submodule.

xarray_to_cdf

wrong

from cdflib.xarray import xarray_to_cdf

correct

from cdflib import xarray_to_cdf

Same as cdf_to_xarray; both are top-level functions.

Quickstart

Open a CDF file, read global attributes, list variables, and read data for a variable.

from cdflib import CDF

cdf_file = CDF('example.cdf')
info = cdf_file.globalattsget()
print(info)
vars = cdf_file.cdf_info()
print(vars)
data = cdf_file.varget('variable_name')
print(data.shape)
cdf_file.close()