Fix hillshade gradient to use Horn's method (#1175)#1178
Merged
Conversation
hillshade was using np.gradient (2-point central difference) for gradients while slope and aspect already used Horn's 8-neighbor Sobel kernel. This caused measurable divergence from GDAL on noisy terrain (up to 0.1 shading units on rough DEMs). Replace np.gradient in _run_numpy with an @ngjit Horn kernel matching slope.py. Update the CUDA kernel (_gpu_calc_numba) with the same 8-neighbor stencil. Both dask backends inherit the fix automatically through their chunk functions. Update test_hillshade_gdal_equivalence to validate against Horn's method reference instead of np.gradient. Add test_hillshade_horn_vs_central_diff to confirm the two methods produce meaningfully different results on noisy data.
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
_run_numpyand_gpu_calc_numbafromnp.gradient(2-point central difference) to Horn's 8-neighbor Sobel kernel, matching whatslopeandaspectalready usehillshade.py:40that claimed GDAL paritytest_hillshade_gdal_equivalenceto validate against a Horn's method reference instead ofnp.gradienttest_hillshade_horn_vs_central_diffto verify the two gradient methods actually produce different results on noisy terrain (max diff > 0.01)Context
hillshadewas computing gradients withnp.gradientwhileslopeandaspectused Horn's method. On noisy DEMs this caused up to 0.1 shading-unit divergence from GDAL. Smooth data was unaffected (diff < 0.001) since the two kernels converge when neighbors are similar.The dask backends (
_run_dask_numpy,_run_dask_cupy) call into the numpy/cupy chunk functions, so they pick up the fix automatically.Test plan
pytest xrspatial/tests/test_hillshade.py-- all 56 tests pass (numpy, cupy, dask+numpy, dask+cupy, boundary modes, RTX shadows)test_hillshade_gdal_equivalenceconfirms Horn reference match (atol=1e-6)test_hillshade_horn_vs_central_diffconfirms old method no longer matches