FillBurn producing odd results on DEM

[p-schaefer]

I'm working in a region with pretty extensive mining. The provincial DEM is quite out of date, so I tried replacing an area of the DEM with some LIDAR data that I aggregated to the same spatial resolution (~15m x 15m). The figure below shows in colour the difference between the two DEMs against the original DEM in the background:

The differences in DEM range from -330m to + 615m where pits have been formed and waste rock piled up. Because of the extensive stream alterations in the region, its necessary to burn a stream line into the DEM to properly snap my sampling points to. However, when I run the fill burn tool in whitebox, I get this (I've changed DEM differences into contours here):

Its burned the stream layer to an elevation of -12,000m, and created odd artifacts. If I try to to use the breach depressions (least-cost, or regular), or fill depressions first, I get a fairly standard looking DEM, i.e.:

but trying to burn the stream into these layers still results in the weird artifacts and super deep stream channels.

For reference, I'm using the R version of whitebox on an Ubuntu 20.04 system, if that makes a difference. Burning the stream network into the original provincial DEM does not result in these artifacts, so it must have to do with replacing a section of the DEM.

Any assistance would be greatly apricated.

[p-schaefer]

I've resolved the issue by manually recreating the fillburn algorithm (or as close to it as I can figure out based on a brief descriptions I was able to find online) with 10,000m height difference instead of the 1,000m height difference for the initial fill offset calculation. I think the issue may have somehow related to the large elevational differences combined with the heavily modified nature of the stream network and landscape.

I wonder if the height that gets added during the fillburn algorithm might be a parameter you can feed the function.