Hi, yes I should have provided more information for my post to be useful.
We do species identification at the individual tree level using ALS. We use raster CHM in our segementation software to delineate individual crowns. In order to minimize edge effects (crowns being chopped in two at the edge and being assigned two different ID's) we use a 20m buffer (this is customizable in our software) on each raster tile.
For small areas (like Petawawa Research Forest in Ontario, Canada ((100 sq. km.)) I use unbuffered las tiles (made with Tile LAS tool), create LAS dataset->las dataset to raster (25cm resolution DTM, DSM)->raster calculator for CHM (DTM-DSM), split raster with 20m overlap. This works very well.
We have other areas that are several thousand square kilometers in size (LAS dataset has 1800 tiles of 1 sq.km each for example) where ArcGIS Pro has no problem displaying the LAS data but creating a 25cm raster and then tiling it as above with the overlap takes a very long time.
Whereas tiling LAS with an overlap and then creating the raster from the tiles is much easier and efficient for us.
