[Talk-us] How accurate is Tiger data vs Yahoo Aerial?

[Matthew Perry]

Alan,

On Fri, Feb 22, 2008 at 12:23 PM,  <am12 at bolis.com> wrote:
>  It looks like it is completely covered already by the Tiger data.  But
>  I've noticed many displacement problems where the Tiger data doesn't line
>  up with the Yahoo aerial photos.
>
>  What I am wondering is: which of the two is more likely to be accurately
>  placed?  Are the Yahoo tiles pretty well positioned?  They look like they
>  fit together fairly well to me, but I don't really know.
>
>  I find that in some sections, the Tiger tracks line up with the Yahoo
>  photos exactly.  But then go a mile over, and the Tiger tracks are off.
>  If I move the photo in Potlatch with space-and-drag, I can line them up
>  again for a section, but then they are off again elsewhere.
>

TIGER accuracy is highly variable. Same with the Yahoo imagery.

In my experience in santa barbara, ca, comparing to auxillary data
sources (USGS orthophotos from terraserver, local datasets, NAIP
imagery, GPS traces, etc) has shown that the Yahoo imagery and TIGER
data are about equal in terms of positional accuracy.

Some TIGER data I've seen suffers from horrible spatial accuracy.
Select areas have systematic coordinate shifts of hundreds of feet.
Other areas have very poor resolution; the sparse verticies lead to
wildly inaccurate maps at detailed scales. Still other areas are
missing crucial data (entire sections of long-established highways) in
addition to containing phantom roads that have no basis in current
conditions. On the other hand, many areas of TIGER are beautifully
accurate.

(see http://casoilresource.lawr.ucdavis.edu/drupal/node/433 for a
technique to fix systematically-shifted areas)

The bottom line is you need a third (or forth) source of data to be
sure but you'll soon get a feel for the most trustworthy source in
your area. It is worth taking the time to get this right ... nothing
worse than referencing all your data to the wrong base map.

>  I'm sure the correct fix is some real GPS readings, and align both the
>  photos and tiger tracks to them.  But I don't have a GPS so I'm just
>  working off the Yahoo photos.
>

Don't assume that the GPS is any more accurate than either the aerial
photos or TIGER. Often times it takes many tracks over the same area
to get a good lock on the average centerline.


-- 
Matthew T. Perry
http://www.perrygeo.net