There appears to be artifacts on the edges of the CHM, which appears to be due to misregistration of the DTM and DSM. This misregistraiton is due to the fact that the extents of the output from points2grid cannot be specified exactly and only uses the closest interior point.
This issue is exacerbated by the fact that the CHM calculation doesn't appear to respect nodata values (if pixel is nodata in either DTM or DSM the CHM should be nodata).
There appears to be artifacts on the edges of the CHM, which appears to be due to misregistration of the DTM and DSM. This misregistraiton is due to the fact that the extents of the output from points2grid cannot be specified exactly and only uses the closest interior point.
This issue is exacerbated by the fact that the CHM calculation doesn't appear to respect nodata values (if pixel is nodata in either DTM or DSM the CHM should be nodata).