7

Since upgrading to ArcGIS 10.1 (SP 1) from 10.0, I've been encountering an issue when running the arcpy.MosaicToNewRaster_management() tool with output to a file geodatabase. I'm principally working with python, but I've duplicated the issue in ArcMap.

Consider a sample dataset, available here, which includes two states (MA and CT) of gSSURGO soil rasters. Create a file geodatabase for outputs, then:

outGDB = <your file geodatabase>    
arcpy.env.outputCoordinateSystem = arcpy.SpatialReference("North America Albers Equal Area Conic")
arcpy.env.OverwriteOutput = True
arcpy.env.pyramid = "PYRAMIDS -1 NEAREST DEFAULT"
#arcpy.env.snapRaster = <your snap raster> # I snap to an NLCD layer, not sure its pertinent to the example, though
arcpy.env.cellSize = 30

inFolder = <the folder which contains the extracted zip file contents>
inList = []
arcpy.env.workspace= inFolder
for f in arcpy.ListFiles():
    inList.append(inFolder+f+"/"+f)

arcpy.MosaicToNewRaster_management(ify(inList,";"),outGDB,"soils",albers,"32_BIT_UNSIGNED","#","1",'FIRST')

The operation successfully produces a new raster, but for me it has some unusual properties. Namely, I find large sections of the file do not display when zoomed out: enter image description here

And as I zoom in incrementally, the gaps fill: enter image description here enter image description here

A few notes:

  • The "Identify" tool tip shows that there is data present even when not depicted. The same issue occurs in both ArcMap and ArcCatalog.
  • The issue does not occur when I perform arcpy.MosaicToNewRaster_management() with output to a regular system folder.
  • Scale range is set to "Show layers at all scales".
  • As I zoom out, the whole extent disappears eventually.
  • Have you managed to solve this issue? I'm having similar problem with large terrain raster dataset and desperately looking for solution. – radek Oct 27 '14 at 16:59
  • 1
    @radek, never got this fixed. I think that I just put up with writing to a system folder and then copying into the gdb. – metasequoia Oct 27 '14 at 17:47
  • Thank for information. Could you please tell me in what format you write to system file, and then in which - to gdb? – radek Oct 28 '14 at 14:47
  • 1
    @radek, I use tiff format for the image and file geodatabases, but I suspect other formats will work as well. – metasequoia Oct 28 '14 at 15:32
2

Check to see if build overviews or pyramids is on when you run this process. The missing display indicates that the overview or pyramid for the "white" area did not generate correctly.

  • I initially suspected this also, but I've rebuilt overviews and pyramids and also recalculated statistics - none of which have provided improvement. – metasequoia May 23 '13 at 3:24
  • Hmmm....I wonder if generating the pyramids is picking up white space cell values and with nearest neighbor they are being promoted to larger bands at the pyramid levels. Is your file geodatabase local on your computer, and the standard file outputs it local? Another maybe is network bandwidth. – user18424 May 23 '13 at 10:25
  • All files are local. I'm curious if you've been able to replicate the issue. – metasequoia May 23 '13 at 12:13
  • For the record I just downloaded the data and ran it through 10.2 and used the Mosaic To New Raster tool storing the final raster in a file geodatabase. I did not get the white lines as shown in your images. – Hornbydd Nov 21 '13 at 16:45
  • @Hornbydd I think you should add that as an answer. – PolyGeo Dec 23 '15 at 9:58
1

I'm having the same issue when mosaicking to new fgdb raster on version 10.2.2.

My solution is to output to a GRID format. The same mosaic operation output to GRID format does not have any of these issues.

  • Thanks for your reply. Unfortunately, I don't have access to ArcGIS anymore, and so I can't verify your workflow. Perhaps others can. – metasequoia Aug 10 '16 at 18:35
1

I just had this happen when mosiacking to a raster in a file geodatabase using Arc 10.3. I also think the problem is the pyramids.

I deleted the existing pyramids, and rebuilt them using the bilinear resampling technique (as opposed to the default technique of nearest). After the new pyramids were done, the raster looked much better.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.