Robert,
Well, that gave me completely inexplicable results.
Original_DEM: 3251 rows x 4618 columns, 3m cellsize. Min elevation, 262.03m. Max elevation 305.78m, Mean elevation 280.69m (values rounded).
Corrected_DEM: 3251 rows x 4618 columns, 3m cellsize. Min elevation, 265.31m. Max elevation 299.39m, Mean elevation 283.40m (values rounded).
The Corrected_DEM was created this way: Line segments were digitized where culverts needed to be "burned in" to the elevation model. Through a number of moves the line segments became the Corrected_DEM, which is NODATA off the line segments and interpolated elevation values along the line segments (now @ 10 gridcells long, each). Also, I set the environment to be the same extent and cell size as, and had the line segments snap to, the Original_DEM.
When I did my attempts at the CON, I got ... exactly the same as my Corrected_DEM. When I did YOUR version, I got this oddball result. The properties indicate the same number of rows and columns, the same cell size, but the Min, Max, and Mean elevation values are all 264.57m (rounded) and all the values are null except for one cluster of gridcells in the far upper left of the DEM measuring about 383m x 383m. This is no where near any of the corrections.
I did this twice -- once just putting in your suggestions, one using your suggestions AND setting the environment to match the Original_DEM. Same results both times.
This should NOT be so difficult! Any further ideas? Thanks, Darlene