Installation#

pip

pip install easysnowdata

conda / mamba

conda install -c conda-forge easysnowdata
# or
mamba install -c conda-forge easysnowdata

pixi

pixi add easysnowdata

uv

uv pip install easysnowdata

One install, no extras. Everything in the catalog is reachable from a plain pip install easysnowdata — there is no easysnowdata[sar] to remember, and no import that fails because an optional dependency is missing. The cost is a larger install; the benefit is that a snippet from these docs runs in anyone’s environment.

Python 3.12 or newer is required. The floor comes from the stack, not from us: current rasterio (1.5, GDAL ≥ 3.8), rioxarray, zarr, numpy, and earthaccess releases have all dropped 3.11.

conda-forge or PyPI?#

Both work. Prefer conda-forge when you can, for two reasons specific to this package:

GDAL drivers. Reading MODIS granules from NSIDC (all six products: MOD10A1, MYD10A1, MOD10A2, MYD10A2, MOD10A1F, and MYD10A1F) needs GDAL’s HDF4 driver, which conda-forge splits into its own libgdal-hdf4 package and which the PyPI rasterio wheels do not carry. Without it the NSIDC route raises a driver error; source="planetary-computer" and every other product are unaffected.
PROJ data. The conda-forge build ships a complete PROJ grid set, which matters for accurate vertical and datum transforms.

Development install#

The repository is a pixi workspace, and pixi.lock pins every environment down to the GDAL build, so CI and your laptop resolve to the same stack.

git clone https://github.com/egagli/easysnowdata.git
cd easysnowdata
pixi install                 # the default environment
pixi run -e test-py313 test-unit    # offline tests: no network, no credentials

The environments and what they are for:

environment	contents	used by
`default`	the package and its runtime dependencies	`pixi run python`
`dev`	plus pytest (socket, recording, cov, and xdist plugins), ruff, pre-commit, ipykernel, and the release tools (build, twine, bump-my-version, git-changelog)	the test tiers
`docs`	plus Sphinx, sphinx-gallery, myst-nb, pydata-sphinx-theme	building this site
`lint`	ruff only	the fast CI lint job
`test-py312`, `test-py313`, `test-py314`	`dev` on one interpreter	the CI test matrix

Contributing lists the tasks each environment offers.

Checking the install#

import easysnowdata as esd

print(esd.__version__)
esd.auth.status()  # which providers are configured; network-free
esd.catalog.list()  # every product, by theme

On an interactive session import easysnowdata prints a single line naming the version, which credentials it found, and whether it is running somewhere with direct S3 access. It makes no network request to do that. Set EASYSNOWDATA_QUIET=1 to silence it.

Credentials#

Most of the catalog is open. The products that are not say so on their catalog page, and Credentials has the setup for each provider. Nothing needs to be configured before installing.