v0.4.0

What's Changed

• Python 3.12 Support!
• chore: Remove numpy pin, pin datashader, drop Python 3.7 by @jthetzel in #789
• Bunch of Fixes by @brendancol in #796
• Move docs to readthedocs [WIP] by @tcmetzger in #797
• Fix typo in readthedocs config by @tcmetzger in #799
• Update readthedocs configuration [WIP] by @tcmetzger in #798
• Add links to docs to readme by @tcmetzger in #800
• Prep PR for 0.4.0 Release by @brendancol in #801

Full Changelog: v0.3.7...v0.4.0

Version 0.3.7 - 2023-06-09

The 0.3.7 release is a hot fix for 0.3.6, which has problem with publishing to PyPi as the package exceeds the limit of 100MB. In this new release, example notebooks are cleaned up to reduce the package size.

What's Changed

• Clear example notebook outputs by @thuydotm in #786
• Change log for release v0.3.7 by @thuydotm in #788

Version 0.3.6 - 2023-06-02

With the 0.3.6 release, xarray-spatial now supports python 3.11. This release focuses on demonstrating the reliability of the library by adding more tests against GDAL/QGIS.

What's Changed

• Fixed feature-proposal.md config by @nodell111 in #718
• WIP Add viewshed gpu notebook by @thuydotm in #720
• Update README.md by @RichardScottOZ in #719
• update pharmacy desserts notebook by @thuydotm in #723
• Update contributor badge by @thuydotm in #740
• Use setuptools by @giancastro in #749
• More examples to user guide by @thuydotm in #742
• WIP Housing price feature engineering notebook by @thuydotm in #725
• Zonal_crosstab 3D: Ensure content of input param values is preserved by @thuydotm in #754
• Classification tools: use binary search in numpy case by @thuydotm in #760
• Zonal crosstab user guide notebook enhancement by @thuydotm in #759
• Test against QGIS, GDAL by @thuydotm in #744
• classify.binary: handle NaNs and infinite values by @thuydotm in #763
• Multispectral tools: convert data to float32 dtype before doing calculations by @thuydotm in #755
• Updated citation for 2023 by @brendancol in #769
• Test all functions to ensure inputs unmodified by @thuydotm in #764
• Pin numpy version by @thuydotm in #780
• correct docs for proximity by @thuydotm in #778
• Test on Python 3.11 by @thuydotm in #750
• Change log for release v0.3.6 by @thuydotm in #785

Version 0.3.5 - 2022-06-05

The 0.3.5 release mainly addresses the scaling issue in GPU viewshed to gain better accurate triangulation.
The GPU raytraced viewshed should now give comparable results to the CPU version.
However, the 2 versions use 2 different approaches, there can be slightly differences at some points
where a version returns visible while the other considers them as invisible.
Many thanks to @nodell111, @a7az0th, and the maintainers @thuydotm for contributing to this release.

Enhancements

• command to get change log (#716)
• Added Feature Proposal Template (#714)

Bug Fixes

• Improved viewshed rtx. Now result should match the CPU version (#715)

Version 0.3.4 - 2022-06-01

Version 0.3.4 - 2022-06-01

The 0.3.4 release primarily a bug fix release but also includes a number of enhancements with a focus on GPU supports.
Many thanks to @mkeenan195, @a7az0th, and the maintainers @ianthomas23 and @thuydotm for contributing to this release.

Enhancements

• NumPy zonal stats: return a data array of calculated stats (#685)
• set unit for hotspots output (#686)
• More robust cuda and cupy identification (#657)
• Remove deprecated tiles module (#698)
• Test on python 3.10, remove 3.6 (#694)
• moved all tests to github actions (#689)
• Add isort to pytest (#700)
• Add flake8 to pytest (#697)
• Remove unnecessary executable flags (#696)
• updated test hotspots gpu (#692)
• 3D numpy zonal_crosstab to support more agg methods (#687)

Bug Fixes

• Fix rtx viewshed rendering blank image (#711)
• Convolve_2d gpu fixes (#702)
• focal.mean(): only do data type conversion once (#699)
• Update to remote sensing notebook (#688)
• focal_stats(): gpu case (#709)
• focal apply: drop gpu support (#706)
• drop gpu support (#705)
• enabled numba.cuda.jit in hotspots cupy (#691)

Documentation

• Correct examples in docstrings (#703)
• Fix doc build dependencies in CI (#683)
• Fix link to Austin road network notebook (#695)

Version 0.3.3 - 2022-03-21

• fixed ubuntu version (#681)
• Don't calculate angle when not needed (#677)
• codecov: ignore all tests at once (#674)
• add more tests to focal module (#676)
• classify: more tests (#675)
• Codecov: disable Numba; ignore tests, experimental, and gpu_rtx (#673)
• Improve linter: add isort (#672)
• removed stale test files from project root (#670)
• User guide fixes (#665)
• license year in README to include 2022 (#668)
• install dependencies specified in test config (#666)
• Pytests for CuPy zonal stats (#658)
• add Codecov badge to README
• codecov with github action (#663)
• Modernise build system (#654)
• classify tools: classify infinite values as nans, natural_breaks: classify all data points when using sub sample (#653)
• Add more benchmarks (#648)
• Stubbed out function for Analytics module (#621)
• Fix doc build failure due to Jinja2 version (#651)

Version 0.3.2 - 2022-02-04

• Remove numpy version pin (#637)
• aspect: added benchmarks (#640)
• Clean gitignore and manifest files (#642)
• Benchmark results (#643)
• handle CLI errors #442 (#644)
• Cupy zonal (#639)
• Tests improvements (#636)

Version 0.3.1 - 2022-01-10

• Add benchmarking framework using asv (#595)
• Fix classify bug with dask array (#599)
• polygonize function on cpu for numpy-backed xarray DataArrays (#585)
• Test python 3.9 on CI (#602)
• crosstab: speedup dask case (#596)
• Add benchmark for CPU polygonize (#605)
• Change copyright year to include 2021 (#610)
• Docs enhancement (#604, #628)
• code refactor: use array function mapper, add messages param to not_implemented_func() (#612)
• condense tests (#613)
• Multispectral fixes (#617)
• Change copyright year to 2022 (#622)
• Aspect: convert to float if int dtype input raster (#619)
• direction(), allocation(): set all NaNs at initalization (#618)
• Add rtx gpu hillshade with shadows (#608)
• Add hillshade benchmarking, for numpy, cupy and rtxpy (#625)
• Focal mean: handle nans inside kernel (#623)
• Convert to float32 if input raster is in int dtype (#629)

Version 0.3.0 - 2021-12-01

• Added a pure numba hillshade that is 10x faster compared to numpy (#542)
• dask case proximity: process whole raster at once if max_distance exceed max possible distance (#558)
• pathfinding: start and goal in (y, x) format (#550)
• generate_terrain: cupy case, dask numpy case (#555)
• Optimize natural_break on large inputs (#562)
• Fixes in CPU version of natural_breaks. (#562) (#563)
• zonal stats, speed up numpy case (#568)
• Ensure that cupy is not None (#570)
• Use explicit cupy to numpy conversion in tests (#573)
• zonal stats: speed up dask case (#572)
• zonal_stats: ensure chunksizes of zones and values are matching (#574)
• validate_arrays: ensure chunksizes of arrays are matching (#577)
• set default value for num_sample (#580)
• Add rtx gpu viewshed and improve cpu viewshed (#588)

Version 0.2.9 - 9/01/2021

• Refactored proximity module to avoid rechunking (#549)