Several months ago I worked with my colleagues at Towson University’s Division of Innovation and Applied Research to produce a training video about ESRI’s mosaic dataset.  In the video I described the differences between the mosaic dataset and other ESRI raster storage formats and presented the benefits of using a mosaic dataset to store (and serve) raster data.   What I did not touch on was some of the difficulties one who is new to using mosaic datasets might face.  The intention of this blog post is to fill that gap by explaining some of the issues I recently faced when creating a mosaic dataset.

Over the last couple weeks, I have been responsible for creating a statewide tax map that displays property parcel boundaries throughout Maryland.  The task involved working with thousands of tax map TIFF images that covered all but four of Maryland’s counties plus Baltimore City.  With so many TIFF images to work with, limited processing and maintenance time, and a desire to serve the end product as an image service I decided to create a mosaic dataset.  Creating a mosaic dataset using the default settings produced a couple issues that I had to correct before publishing the mosaic dataset as an image service.

Overlapping Rasters

The figure below shows the problem that existed at the tax map boundaries.  One can see that the white borders of some tax maps obscure the property boundaries shown on neighboring tax maps.

Smalloverlapissue

Overlapping raster issue: some tax maps obscure the property boundaries shown on neighboring tax maps. Click the map for larger image.

The problem occurred because the default mosaic method of a mosaic dataset is set to “First”.  This means that the raster shown to the user in the case of overlapping rasters is the raster that was added first to the mosaic dataset.  The solution to the problem is to modify the mosaic dataset’s mosaic method using ArcCatalog or ArcMap. Changing the mosaic method to “min” solved my problem by always displaying the pixel with the minimum value, which is the pixel value associated with the property boundaries.

Overlapping raster issue fixed. Click map to see larger image.

Overlapping raster issue fixed. Click the map for larger image.

Rasters Displaying At Various Scales

Another problem I experienced was that not all of the tax maps showed up at the same scale because not all of the tax maps have the same pixel size.  While this may be acceptable for most applications, I needed all tax maps to display at the same scales regardless of the source data pixel size.

The solution to this problem was modifying the MaxPS field in the Footprint attribute table.  The MaxPS field is automatically calculated when the mosaic dataset is created and is based off the HighPS field that exists in the same attribute table.  As described here, the MaxPS field specifies the minimum scale at which a given raster is displayed.  I modified the MaxPS field by making sure all rasters used in the mosaic dataset had the same MaxPS value.  This ensured that all tax maps display at the same scales.

Modifying the MaxPS field in the Footprint attribute table. Click the image for larger view.

Modifying the MaxPS field in the Footprint attribute table. Click the image for larger view.