#1088: r.fillnulls: support other interpolation methods
-------------------------+--------------------------------------------------
Reporter: kyngchaos | Owner: grass-dev@…
Type: enhancement | Status: new
Priority: normal | Milestone: 6.5.0
Component: Raster | Version: svn-develbranch6
Keywords: fillnulls | Platform: All
Cpu: All |
-------------------------+--------------------------------------------------
Comment(by mmetz):
In grass 7, it's already implemented in r.resamp.bspline. r.resamp.bspline
interpolates a raster to the current resolution, filling NULL cells on the
fly, or optionally only interpolates NULL
cells. Available methods are bilinear and bicubic, tested with SRTM data,
filling large gaps in the European Alps.
For grass 6.x, I would use a different approach to r.fillnulls to follow
the design of v.surf.bspline, because only interpolating NULL cells is
already possible with v.surf.bspline:
Create a new raster where NULL cells in the original surface raster are
set to something else than NULL, others to NULL. Convert both raster maps
to vector points. Use the vector points representing NULL cells as sparse
points input for v.surf.bspline. Recommended settings for v.surf.bspline
in this case are sie = 2 * ewres, sin = 2 * nsres, lambda = 0.005 (best in
my tests, should be somewhere between 0.001 and 0.01). The output raster
holds the interpolated NULL cells and can be patched with the original
surface.
Although all lidar tools work within the current region, none respects a
MASK. I'm not sure how this could be implemented properly and efficiently,
currently there will be IMHO harmless warnings about no points in the
current subregion.
Markus M
--
Ticket URL: <https://trac.osgeo.org/grass/ticket/1088#comment:7>
GRASS GIS <http://grass.osgeo.org>