I think William's right, the shape factor is important. A way to combine the two so that the shapes of the large null areas is the same as the gaps in the original raster would be to make polygons out of the original gap areas, then use William's method with r.grow, then making polygons that match the unfilled areas. The polygons that have the area of the original gaps that also overlap polygons generated using the r.grow method would then be the polygons to use as a mask. Thus the thin gaps would be filled in.
The only problem there is that large gaps with a bottleneck somewhere would be shut out of the analysis, although the bottleneck could theoretically be interpolated quite well. But that would have happened with my original suggestion too. Or you could use r.grow and just accept the fact that even the large interpolated areas will be interpolated on the edges. Then the bottlenecking problem would be gone. -- View this message in context: http://osgeo-org.1803224.n2.nabble.com/Limit-r-fillnulls-to-interpolate-only-small-gaps-tp6040101p6040361.html Sent from the Grass - Users mailing list archive at Nabble.com. _______________________________________________ grass-user mailing list grass-user@lists.osgeo.org http://lists.osgeo.org/mailman/listinfo/grass-user