I am attempting to recreate a data management model decribed in the "What is a mosaic dataset?" help topic.
[INDENT]For example, you could create a mosaic dataset to manage all your DEM data, then create a referenced mosaic dataset to produce a hillshade or a slope product from the source mosaic dataset. [/INDENT]
My problem is that I am only able to render the referenced mosaic datasets using 'Stretched' and 'Discrete Color' and I would like to, but am not able to use 'Classifed' renderers even after calculating statistics on the DEM mosaic datasets and/or derived reference mosaic sets.
THis limitation occurs whether or not I am attempting to render the original DEM mosic without raster functions OR the refernced mosaic datasets with 'Slope' raster function or 'Aspect' raster function applied.
What is odd is that if I use the Image Analysis window to apply a Slope or Aspect raster function to a DEM raster dataset or DEM mosaic dataset, the temporary 'on-the fly' layer that is automatically added by the Image Analysis window to the map table of contents will allow me to apply a 'Classified renderer'
Ideally I would like to be able to apply a 'Classified' renderer that references the mosaic or referenced mosaic dataset with raster functions for slope and aspect.
Is this limitation a bug or is there a valid reason for this restriction on mosaic datasets?
Or is there there just some step that I am missing and it should be possible to render a single-band mosaic dataset using a 'Classified' renderer?
Finally, Is there any effective difference between saving a layer file of the Image Analysis window's on-the-fly depiction of a raster function ( e.g. 'slope' applied to a DEM) and a raster function added via ArcCatalog to a mosaic dataset stored in the geodatabase?
If there is no significant disadvantage to the layer file from the on-the-fly Image analysis window raster function approach , I am not sure why one would bother to create referenced mosaic datasets with raster functions applied since both methods share the advantage of not having to duplicate the source data and there seems to be more control over how the data is rendered using the layer file approach
Thanks to anyone who has answers and insights to these questions
Mark