I've been experiencing this problem for over a year now, but it's become extremely bothersome now that I'm working on the "heads-up digitizing" of large extent areas. It seems that compared to accurately verified survey points and USGS image tiles, ESRI's base imagery simply does not align with the same accuracy. Through many projects, I have verified the proper projections of both survey data and georeferenced imagery downloaded from the USGS (both of which align perfectly ontop of each other), but turning on the Bing base imagery shows it can be mis-aligned as much as 10 feet in some instances - even when the data frame is projected in the same coordinate system as the survey data and georeferenced imager tile.
For small projects, this does not matter for me - since one or two USGS image tiles give the accuracy I need. With larger projects, however, extending hundreds of miles - I simply don't have the time or physical disk space to download accurate USGS imagery. Instead, I must rely on Bing base imagery to trace my features (in this case, railroad tracks).
The attached picture shows a new MXD, with simply the USGS image tile imported. This ensures ArcMap imported the projection of the image tile since it was the first feature added to the map. Then, I turned on Bing base imagery:
The image above shows the edge of a USGS image tile, shown over the background of Bing imagery in ArcMap. As you can see, the features don't line up - the railroad tracks, the right-of-way, the vegetation zones, etc. While this example is just east of Sidney, Ohio, USA - this seems to be a problem in all parts of the continental United States. I have rarely, if ever, seen the Bing imagery "spot on" with properly referenced imagery or survey data.
What am I doing wrong, if anything?
Suppose I created a complex network of features that were traced using the Bing base imagery. Is there a tool to reproject these features to a more accurate spatial location?
Thanks for any and all help!