Making Projections Correct and Consistent

In order to make proper comparison between the different DEMs discussed above, I need to make sure that each of the files being compared has the same projection and the same resolution.  To do this requires a few steps.

Projections:
I have essentially three different data sets in different projections.

• ASP (parabola and bayesaaw) data are in "Albers_Conic_Equal_Area" with an undefined datum
• SETSM data are in Polar_Stereographic
• Airborne LiDAR are in "NAD_1983_Transverse_Mercator"

My first responsibility is to make sure that all the projection information is defined correctly for the files.  With the help of Clarie from PGC I was able to get a definitive answer on which projection each of the files had.  I then needed to define the projections for all the files.  Note that there are 380 tif files that require this treatment.  

I started the process manually, using ArcGIS's ArcToolbox and but found the right-click, batch Define Projection tool tedious as I needed to paste the projection information into each cell of the batch array.  Then I found a better way to batch define projections (or batch anything).  In order to use a recursive search tool (allowing the contents of subfolders to be searched and modified), I put together a simple ArcToolbox 'model builder' routine which allowed me to combine functions to do the job efficiently (thanks Frank Vignati for the idea).  After 2 minutes of building the model, I was able to process all the files within 3 minutes (and have a handy tool for the future).  Someday I'll learn Python and this will be even easier.

I am now confident in all three projections.  They are all now defined as:

• ASP
• NAD_1983_Alaska_Albers, WKID: 3338 Authority: EPSG
• SETSM
• WGS_1984_NSIDC_Sea_Ice_Polar_Stereographic_North, WKID: 3413 Authority: EPSG
• LiDAR
• NAD_1983_Transverse_Mercator, Authority: Custom (defined by mapping contractor)

Consistent Projections?

The next step is to decide if it is better to reproject all the files to have the same projection (over 100Gb of data = slow/tedious) or whether there is a better way.  I think for this first analysis I will use model builder to clip my lidar data to the extents of the 3 domains discussed above, reproject them to match the Stereo DEM data and then resample the LiDAR data to match the Stereo DEM.  Next I will use the extent of the clipped/projected/resampled LiDAR data to clip the Stereo DEM data.  At this point I will have two rasters ready for comparison.  The Model Builder workflow looks a little like this:

Now that a plan is in place...let's see how this all works.

After multiple iterations, I have improved on the model to produce the outputs I am looking for.  I am sure it could be improved (automate adding to map, grouping, symbology), but this is beyond my skills and time.