Mosaic DEM Problem

[StevePlatt]

Hi,

I am wanting to mosaic many DEM Grids into one large DEM Mosaic. They do NOT overlap and they appear to be perfectly adjascent to one another. When I mosaic them, everything appears to turn out well. However, when I run a hillshade, you can notice a distinct seam/line where the tiles met. This is a problem because it is affecting further processing on the mosaic DEM (ex/ Hydrology, Flow Direction etc...).

Is there a way to create a trully seamless mosaic? Perhaps there is a way to interpolate the seam area? Perhaps I should be using another tool (merge)?

Any help would be appreciated.

Steve

[EricRice]

Hi Steve,

Is it possible for you to upload a screenshot of your hillshade result so I could have a look.  Also, can you describe a little about the pixel values around the seam/problem area?  For example, are the values identical on either side of the seam?  Do they have similar slopes? 

I typically use Hillshade just for cartographic reasons.  If your use of the hillshade is for this reason I see no harm in running a focal mean on the DEM to smooth out this seam.  It may result in a better visualization of the surface.  Since you mention further analysis though you probably won't want to use the result of the focal mean for anything other than creating the hillshade effect.  Can you explain how your existing DEM is impacting your hydrologic analysis?

[DanaRobinson]

I am having the identical problem.  I am using 1 degree by 3 degree 10 m resolution, 32 bit float DEMs from the USGS. I need to mosaic 6 total dems from two areas.  The seam appears between the two Areas, in particular between only two DEMs.  The seam that appears between the two mosaiced images is approximately 1 pixel wide.  The pixel elevation values on either side do not match exactly, and vary from less then 1 m difference to less then 10 m difference.   I have tried a variety of workflows to solve this problem from changing the Mosaic method (last or blend), changing the loading sequence for the rasters, clipping the rasters before or after the mosaic process, and reprojecting before and after the mosaic.  The problem never resolves itself.  I have attached a screenshot for you to view. Can you suggest a solution even if it involves some type of resampling to close the gap and smooth the jump in elevation values?

[JeffreyEvans]

There are very few options, other than smoothing the data, for dealing with actual seams in the DEM. However, the missing value issue is easy. You can fill the missing data using isnull nested in a conditional statement.

In raster calculator:
CON(ISNULL[dem], FOCALMEDIAN([dem], RECTANGLE, 3, 3]), [dem])

[DanaRobinson]

I have very limited experience with raster calculator.  Do I simply copy and paste the calculation line that you wrote out?