Xarray tools for OPERA DISP

Documentation: https://mgovorcin.github.io/disp-xr/

pip install disp-xr

Analysis tools using xarray for OPERA Surface Displacement from Sentinel-1 (DISP-S1) and Surface Displacement from NISAR (DISP-NISAR).

Quick start

import disp_xr as dxr

df = dxr.find_products("path/to/granules/")   # or "s3://bucket/prefix/"
ds = dxr.open_disp(df, corrections=True)       # lazy (time, y, x), CRS attached
ds = dxr.chain(ds)                        # displacement from the first date

ds.displacement.isel(x=4000, y=3000).plot()    # one pixel's time series

open_disp returns displacement as stored in the granules, relative to each ministack's reference date; stack_disp chains the ministacks. See docs/specs.md for the plan and CHANGELOG.md for what has changed.

Documentation

The documentation (install, getting started, concepts, command line, a gallery of examples and the API reference) is published at https://mgovorcin.github.io/disp-xr/ once the repository is public. It builds locally with mkdocs:

pixi run -e docs mkdocs serve    # http://127.0.0.1:8000

Development setup

Prerequisite installs

1. Download source code:

git clone https://github.com/mgovorcin/disp-xr.git

2. Install with pixi (recommended). It reads pyproject.toml and installs disp_xr in editable mode:

cd disp-xr
pixi install            # core environment
pixi install -e dev     # plus tests, linters, aws, plot, mintpy and notebook extras
pixi run -e test test   # run the tests
pixi run -e test check  # run the pre-commit hooks

Environments: default, test, docs and dev.

Or install the dependencies with conda/mamba, either to a new environment:

mamba env create --name disp_xr --file disp-xr/environment.yml
conda activate disp_xr

or install within your existing env with mamba.

3. With conda/mamba only (pixi already did this), install disp_xr via pip in editable mode

python -m pip install --no-deps -e  disp-xr

Setup for contributing

We use pre-commit to automatically run linting, formatting, and mypy type checking. Additionally, we follow numpydoc conventions for docstrings. To install pre-commit locally, run:

pre-commit install

This adds a pre-commit hooks so that linting/formatting is done automatically. If code does not pass the checks, you will be prompted to fix it before committing. Remember to re-add any files you want to commit which have been altered by pre-commit. You can do this by re-running git add on the files.

Since we use black for formatting and flake8 for linting, it can be helpful to install these plugins into your editor so that code gets formatted and linted as you save.

Running the unit tests

Tests run on small synthetic DISP-S1 stacks written at run time (tests/fixtures/make_disp.py); no data download is needed. See tests/data/README.md.

pixi run -e test test   # pytest, including doctests in src/disp_xr
pixi run -e test cov    # with a coverage report

With conda/mamba, install the test extra and run pytest:

python -m pip install --no-deps -e .[test]
pytest

Tests marked xfail(strict=True) assert the correct behavior for known bugs listed in docs/code_review.md. When a fix lands, the test starts passing, pytest reports it as a failure (XPASS), and the marker is removed in the same change.