Testing Positional Accuracy

This was a fun exercise for me because the data used in this weeks module was of my hometown, Albuquerque, NM.  The task was to test the positional accuracy of the two provided datasets against a reference dataset.  The data provided included centerlines of roadways throughout Albuquerque, one from the City of Albuquerque and the other from StreetMaps USA.  The reference dataset was generated by digitizing points at the center of intersections of 20 test locations.  Orthophotos of the Albuquerque metro area were also provided and utilized to create spatially accurate reference points.  By creating a shapefile of reference lines, I used them to trace the curblines across 4-way intersections, snapped additional reference lines across the center of the created box, and finally snapped a point feature from the reference dataset at the intersection of the reference lines.

Two additional shapefiles were made to capture point features at the centerline intersection of the tested datasets at the same test locations.  I also created feature templates for the feature classes to populate the created attribute field of the point features to correspond with the testing location ID as I plotted them.  To calculate ideal spacing (minimum of 20% in each quadrant and > 10% of diameter apart), I used the reference lines to create quadrants.  By snapping lines to the vertices of the corners of the study area, midpoints were used to reliably create the quadrants.  By measuring the corner to corner distance of the study area (9.3 miles across), 10% of that distance (0.93 miles) was used to create another reference line to measure spacing between reference points (shown in the NW quadrant).

After using the Add XY Coordinates tool to the point features from the three datasets, the X and Y coordinates were loaded into a horizontal accuracy statistic worksheet modeled after what was described in the Positional Accuracy Handbook (Minnesota Planning, 1999).  This was populated using an Excel file; rather than opening the data as a .dbf file, I used the Table to Excel tool to export the attribute tables of each dataset to a .xlsx file.  I then copied/pasted the X and Y coordinate values accurately into the worksheets.  After the formulas were loaded into the worksheet to obtain the RMSE of each dataset, positional accuracy can be estimated.

City of Albuquerque dataset - Positional Accuracy: Tested 22.3 feet horizontal accuracy at 95% confidence level.

StreetMaps USA dataset - Positional Accuracy: Tested 732.8 feet horizontal accuracy at 95% confidence level.