v0.7 upgrade

[evetion]

[0.7.0]

Added

• Improved documentation by using VitePress and showcasing all methods on a DEM.
• Package extensions on GeoArrays and Rasters that automatically set the cellsize for all methdos using that kwarg, taking into account geographical data (lat/lon).
• Added hydrology operators: filldepressions, flowaccumulation, with different algorithms (D8, DInf, FD8) available via the method kwarg.
• Added multiscale options to relative terrain operations using a window kwarg for a Stencil from Stencils.jl package.
• Added BPI as a relative terrain operation.
• Overhauled curvature methods by introducing plan_curvature, profile_curvature and contour_curvature, while deprecating curvature for laplacian.
• Added direction kwarg to slope and curvature methods.
• Added radius kwarg to curvature methods.
• Added this CHANGELOG.md

Changed

• Using Stencils instead of LocalFilters.
• Refactored spread to choose from multiple algorithms with the method kwarg.

Fixed

• Relaxed Array input to AbstractArray for opening

[evetion]

@alex-s-gardner For which curvatures (plan, profile, contour) do you want to use the direction argument? Do you have something working currently?

[evetion]

@rafaqz @tiemvanderdeure Ideally we implement crstrait on Rasters itself. And there is a cellarea method, but ideally we have the underlying cellsize (resolution?) as well. If we get a unit with that, even better.

[alex-s-gardner]

@alex-s-gardner For which curvatures (plan, profile, contour) do you want to use the direction argument? Do you have something working currently?

I missed your earlier ping: I currently use directional profile curvature... instead of calculation curvature along maximum slope I calculate it along a direction.

[evetion]

@alex-s-gardner For which curvatures (plan, profile, contour) do you want to use the direction argument? Do you have something working currently?

I missed your earlier ping: I currently use directional profile curvature... instead of calculation curvature along maximum slope I calculate it along a direction.

Do you have a code example/snippet? My code is here: https://github.com/Deltares/Geomorphometry.jl/pull/41/files#diff-5e200c4ade1e6d95f1c218e8fa25406ee2b424c81f5dff53ecc59c1c641adc4fR295-R334, can be a bit hard to follow. A direction of 0 degrees is north. Might be best to just plot and compare, and see whether they're similar.

[alex-s-gardner]

Do you have a code example/snippet?

So in my work I was only calculating curvature along the x and y directions only so my stencil looked like this:

## planer curvature
    # https://pro.arcgis.com/en/pro-app/latest/tool-reference/spatial-analyst/how-curvature-works.htm
    ddx(v) = ((v[2] + v[8]) / 2 - v[5])  / lx^2
    ddy(v) = ((v[4] + v[6]) / 2 - v[5])  / ly^2

where v is the stencil extracted from the DEM, lx and ly are the pixel size (looks like I was using planer not profile

[evetion]

That's only the dz/dx and dz/dy calculation (so slope, not curvature?). Docs are now live at https://deltares.github.io/Geomorphometry.jl/dev/usage#curvature

[alex-s-gardner]

It's curvature as it subtracts the central value v[5] and divides by l[x/y] .^2 so its ddz/dx and ddy/dy

It's D and E in the first example found here: https://pro.arcgis.com/en/pro-app/latest/tool-reference/spatial-analyst/how-curvature-works.htm

But not worth spending any time on ... i use it to identify correlated biases in DEMs so I only use it as a relative metric of curvature... not something you should replicate... sorry for the confusion

[alex-s-gardner]

Docs are now live

These docs are so impressive... what a great package!

[evetion]

It's curvature as it subtracts the central value v[5] and divides by l[x/y] .^2 so its ddz/dx and ddy/dy

It's D and E in the first example found here: https://pro.arcgis.com/en/pro-app/latest/tool-reference/spatial-analyst/how-curvature-works.htm

But not worth spending any time on ... i use it to identify correlated biases in DEMs so I only use it as a relative metric of curvature... not something you should replicate... sorry for the confusion

Well, it's already in. It was named curvature, but based on the Minár paper now laplacian, and it takes a direction argument. See https://github.com/Deltares/Geomorphometry.jl/blob/main/src/terrain.jl#L244-L293 (I used the ArcGIS reference as well). The only thing that's different from directly using D and E is the -2 multiplication. I will add it to the examples as well.