h5netcdf

What it is and what it does

h5netcdf is a Python interface to the netCDF4 file format that reads and writes HDF5 files directly via h5py, h5pyd, or the pure-Python Pyfive backend, without requiring the Unidata netCDF C library. It offers two APIs: a modern hierarchical interface modeled on h5py, and a legacy API designed for compatibility with netCDF4-python. The package supports both NETCDF4 and NETCDF4_CLASSIC formats, unlimited dimensions, groups, and hierarchical variable access.

The main appeal is reduced installation friction—if you already have h5py, reading netCDF4 files may be simpler than installing netCDF4-python and its C dependencies. It also provides a pure-Python reading path via Pyfive, useful in restricted or hard-to-build environments. The package has been in active development since 2015 and is marked production-stable, with support for Python 3.9 through 3.14.

Use it for:

• Read and write netCDF4 files in environments where the Unidata netCDF C library is difficult or impossible to install.
• Access HDF5 data via the netCDF4 interface when you already have h5py installed, avoiding duplicate binary dependencies.
• Use the pure-Python Pyfive backend to read netCDF files in restricted or containerized environments without C compilation.
• Migrate from netCDF4-python to h5netcdf using the legacy API for drop-in compatibility with minimal code changes.
• Work with hierarchical netCDF data using the modern API's h5py-like syntax for direct group and variable access.

Worth the install?

AI-flagged interpretation of the facts on this page — verify before relying

h5netcdf reads and writes netCDF4 files using h5py or h5pyd backends, bypassing the Unidata netCDF C library and offering an alternative interface to netCDF4-python with optional pure-Python support via Pyfive.

Yes. h5netcdf is production-stable, actively maintained, has low install friction, no known vulnerabilities, and a permissive license. Install it if you need netCDF4 support without the Unidata C library, or if you want a pure-Python reading option. The legacy API makes it a low-risk alternative to netCDF4-python for existing code.

Install

h5netcdf on PyPI

pip install h5netcdf

uv add h5netcdf

poetry add h5netcdf

Installing h5netcdf

Before you install

Low friction: pure Python wheel with only packaging and numpy as runtime dependencies. Actively maintained with recent commits and production-stable status. Optional extras (h5py, h5pyd, pyfive) can be installed separately to avoid forcing a specific HDF5 backend.

License in practice

BSD 3-Clause permissive license allows commercial and private use with minimal restrictions—retain copyright notice and disclaimer in distributions.

Quickstart

pip install h5netcdf[h5py]

import h5netcdf
import numpy as np

with h5netcdf.File("mydata.nc", "w") as f:
    f.dimensions = {"x": 5}
    v = f.create_variable("hello", ("x",), float)
    v[:] = np.ones(5)

Requires h5py (version 3.0+) or h5pyd installed separately; pure-Python Pyfive backend available but optional.

Verify before relying

• Whether the reported 4x performance improvement over netCDF4-python holds for typical workflows beyond the single cited case.
• Current state of unsupported netCDF operations (rename, delete) and whether patches have been accepted since publication.

Package facts

Evidence: h5netcdf-1.8.1-py3-none-any.whl

Tags

More Scientific/Engineering packages