Best Way to Merge Two Rasters that overlap

05-29-2020 08:52 AM
Occasional Contributor III

I am using Pro.  I have tried Mosaic to New Raster, Mosaic, and I have tried importing two rasters into a raster data set and a mosaic data set.

I cannot seem to "lose" the extent of the one of the images. When merged or mosaiced it consistently blocks out the extent as no values or does not port over the overlap.

What am I doing wrong?

What it looks like when I import or try to make it:

What I would like after (image shows with symbology with mask on):

3 Replies
Esri Frequent Contributor

I usually use a mosaic dataset and then build footprints then overviews.

Here is some documentation: Using Mosaic Datasets to Manage Imagery—Imagery Workflows | Documentation 

Build Footprints—Data Management toolbox | Documentation 

Build Overviews—Data Management toolbox | Documentation 

Here is one that I have made up of multiple different rasters:

Then you could export the raster as a single mosaic.

--- George T.
Esri Regular Contributor

George (and David)

Thanks, that looks pretty good but the black "nodata" at the edges hasn't been completely removed.  I am guessing you have compressed data (JPG, JP2000, MrSid etc.) and where NoData was originally 0 it has been changed to a value of 1 or 2 due to the compression, and these nonzero values confuse the Build Footprints tool.  Note that tool includes an option to "shrink" footprints, so presuming your data has plenty of overlap, adding a value (in map units like meters, not pixels) to reduce the footprint extent should eliminate the remaining NoData.  

You can also add Seamlines which can improve the visual appearance.  Further you can consider running color correction but note if you are merging 8 bit preprocessed orthos that have already been processed for color correction, matching datasets from different years/seasons (or if clouds/shadows are present) may be challenging.

Cody B. 

MVP Regular Contributor

Agreed with Build Footprints, you can draw your own, but the RADIOMETRY option may be handy in this instance.

0 Kudos