College of Agriculture and Natural Resources
Center for Land Use Education and Research
Coastal 1934 Aerial Images

 

Connecticut Shoreline Change

Research

Rectified 1934 Image Dataset for the Connecticut Shoreline

Detailed Methods

The key with this project is precise image alignment, to make sure features within the 1934 images are spatially accurate when compared with locations in modern imagery.  To accomplish this we had to decide on a base map to work from.  We concluded the Statewide 2012 leaf-off ortho imagery provided the best available base for the alignment, it has detailed resolution and a high standard of spatial accuracy in Connecticut. 

We do not have the tools or elevation data from the 1930 to create true ortho photographs, instead we had to manually georeference the images to the 2012 layer.  ArcGIS versions starting with 10.1 have a projective option within their georeferencing tool that works well with vertical aerial photographs, removing distortions of pitch and roll with a minimal number of reference points.  This tool and the relative flatness of the CT shoreline allows for accurate alignments.  Areas with high elevation or complex topography showing misalignment with modern ortho imagery were deliberately ignored as sources for control points if their use degraded the alignment of coastal features. 

High resolution scanned TIFF files were downloaded from the MAGIC web server (http://magic.lib.uconn.edu/mash_up/1934_aerial_index.html).  Images were chosen if they covered a significant section of the coastline and tied to or overlapped with adjacent images.  In many cases there was enough overlap to allow the selection of the better of one or two images.  The selected image was then cropped to eliminate borders, stamped text, and to focus coverage on the areas of interest.  These scans were then brought into ArcGIS software above the state 2012 image and roughly fit to the area covered using the fit to view option in the georeferencing tool. 

Georeferenced points were selected with a focus on coastal alignment, attempting to choose all points from the same elevation plane across the image.  Aligning to relative sea level provided the best chance of matching features along the coast.  These included rocks exposed just above the water, the base of docks or seawalls, and features level with inland marshes were chosen to keep this alignment plane.  The projective option required at least four tie points with the modern images.  Care was taken to select well distributed features, a rare ideal would be one at each corner of the cropped image. 

Ground Control Point Sample

 

Once four points were selected the GIS software would calculate the approximate location of other points across the image as if the photograph were then stretched to best match the view along that plane.  Additional points were collected to verify alignments and tighten the RMS values if necessary.  In a perfectly flat landscape all of the other pixels would then appear where they should in a geographic map.  In reality features above or below the projective plane can appear to shift, ignoring the tops of buildings and tall hills distortions from elevation were minimal in most of the images.

Alignments across each photograph were checked with the slider and toggle features in the effects tool.  Once verified the projection was locked into place and the image saved to a folder of rectified images.  The rectified images were then linked into one ArcGIS Mosaic dataset.  Because the focus is spatial alignment no attempt was made to color adjust between the images.  To provide the best view scans determined to have the highest quality were listed higher in a draw level field to assure they display above overlapping images of lesser quality (poor focus or alignment).  The user retains the option to switch to a lower level if it provides a better result for their location.

 

Contact Joel Stocker for more information
joel.stocker@uconn.edu.

Back to Top