I'm wondering if anyone has used the 2006 4-band NAIP imagery to construct vegetation models (eg NDVI). I noticed that the stretched value and pixel values match exactly for all reflectance bands. Can this possibly be correct?

I also noticed that the band numbers didn't make much sense. For example, band 1 on the color image is red, but band 1 is usually blue in 4-band imagery. Previously, I noticed that ArcGIS rearranges the band numbers to match the display properties during processing. For example, I had a Quickbird image that I tried to re-project. Not only did ArcGIS rearrange the bands, but it dropped the band that was not displayed.

Could both of these problems (strange band numbers and pixel values) be the result of processing steps in ArcGIS? Is there a way to download minimally processed 4-band NAIP imagery?