Flowaccumulation Ldd: NaN values

[AquaPore]

Great job to put Ldd in Julia language.

There could be issues outside the "catchement" were there is NaN values, but this may be normal behaviour.

Output_Ldd         = "Ldd.tiff"
Path_Output_Ldd    = joinpath(Path_Root, Path_Output, Output_Ldd)
println(Path_Output_Ldd)

println(Dem_Resample[1,1])

Acc, Ldd = Geomorphometry.flowaccumulation(Dem_Resample; method=D8())

Rasters.write(Path_Output_Ldd, Ldd; ext=".tiff", force=true, verbose=true, missingval= NaN)

CairoMakie.heatmap(log10.(Ldd); colormap=:rain)

[evetion]

It's a bit undefined what happens in non-finite regions of an image. As you can see, it's used in the actual calculation (the weird stripes). Note you can pass a second argument with a boolean mask (trues will be skipped). That said, the algorithm advances from the edges of the image, so I'm not 100% sure if you can mask out an island completely/correctly.

[rafaqz]

Can you use raster missing values as the mask?

[evetion]

Might be good to have a mask function, replacing the default falses mask, that can be overridden in the package extensions indeed.

[rafaqz]

Sounds good. I think it will be confusing for people that's it's not automatically handled for them.

[AquaPore]

Thanks for handling this very complex and non trivial boundary issues. I am wandering before we try and perform an artificial mask, do we understand why the flowaccumulation is performing calculation although there is NaN values?