I am working on creating slope surfaces for the entire lower 48 using 1/3 arc-second data from USGS. Recently, I have begun experimenting with running the slope tool using the geodesic method without re-projecting the original rasters (which are in NAD83) to a local coordinate system.
For the most part, the geodesic method appears to work, but it introduces some "noise" in the resulting slope surface when compared slope derived from a planar method and local coordinate system. The major issue I see is that it produces illogical and clearly incorrect values along the 90th west meridian (see maps attached, I apologizes in advance for the ugly graphic). I thought this was user error, or an issue with the input DEM, however I have replicated this using various tiles and input data sources. I have not completed the lower 48, however, everywhere else in the lower 48 where I have completed this appears fine.
What is going on here?