georeferencing Sanborn maps

2276
5
11-02-2011 10:59 AM
ccain
by
New Contributor
I want to digitize buildings from a 1895 Sanborn fire insurance map to overlay current basemaps. The original maps are large scale with I guess local coordinates. (No known projection). All of the buildings are rectangular on the map as are most of the streets. If I georeference the map to a current Streetmap baseline prior to digitizing, the buildings are oblique. Is it better to digitize streets and the buildings from the Sanborn maps then spatially reference them? Or georeference the maps first then digitize? Any suggestions are appreciated.
0 Kudos
5 Replies
PeterStratton
New Contributor
We georeference Sanborns before creating any vector graphics or export rasters based on them.  How bad is the distortion and how many control points are you using?  Are you sure your control points are using the same locations?
0 Kudos
ccain
by
New Contributor
I am using road intersections based on the PennDot State roads layer from ArcGIS online. (It has a GCS.) The RMS is 0.0005 which I take as being very good but the buildings are not square. Perhaps when I digitize them and and project them into a projected coordinate system they will be?
0 Kudos
PeterStratton
New Contributor
how many control points?
0 Kudos
MelitaKennedy
Esri Notable Contributor
I think that's 0.0005 degrees, which is not that good. I would try two things: see if you can set the georeferencing/spatial adjustment to use a conformal/similarity transformation rather than an affine one. Conformal/similarity allows a single scale value in both x/y directions, so right angles are maintained. What might also help is to project the reference data to a conformal projection for the area--like the local State Plane zone and then georeferencing/spatial adjusting to it rather than to lat/lon.

Melita
0 Kudos
ccain
by
New Contributor
Thanks Melita for the tip, Projecting to a State plane looks a lot better. I was going to reproject after digitizing but this works better. I had also slipped a decimal point on the RMS it was 0.00005.

Charlotte
0 Kudos