Module 1.2: Data Quality - Standards

    For this week’s lab, I assessed the accuracy of two Albuquerque, NM street layers based on the horizontal error of the intersections compared to a reference set. Below are the steps that I took to generate the estimated error and provide a formal accuracy statement for each shapefile.

• Determine the sampling values (C and C/10) where C is the length of the diagonal for the study area
• Identify the intersections to be sampled
• Ensure that the selected intersections meet the sampling criteria
•     Points must be evenly divided among the quadrats of the study area (min. 2% in each quadrant
•     Points must be no closer than the length of C/10
•     If they do not, reselect intersections to meet the sampling requirements 
• Below are the selected intersections for horizontal accuracy assessment (n=20)

• Create three feature classes - one for the reference points and one for each street layer that will be assessed (ABQ and StreetMap)
• Create the point features for each feature class based on the selected intersections - use unique IDs to differentiate each point
• The same IDs should be used in each feature class so the intersections can be cross referenced
• Use the Add XY tool to add the point’s XY coordinates to each feature class’s attribute table
• Export the attribute table as a dbf file
• Use the exported values to determine the horizontal accuracy of each street layer

    After completing the analysis, it was apparent that one layer was more accurate than the other. The ABQ Streets layer proved to be the more accurate of the two layers. Below are the formal accuracy statements for each layer.

The horizontal positional accuracy of the ABQ Streets dataset tested 21.1 ft at 95% confidence level using the NSSDA.

The horizontal positional accuracy of the StreetMap USA dataset tested 308.3 ft at 95% confidence level using the NSSDA.

Stay tuned for next week's lab where I continue to explore data quality.