GIS 5935 Module 2

This week we continue working with data accuracy. We were provided two data sets, each depicting the road system for the City of Albuquerque New Mexico. Our job was to use NSSDA data standards to determine the accuracy of the data that was provided and ultimate write an accuracy statement for each set of data provided. 

The first step was to designate at least 20 distinct test locations. These test locations would be the site of where each analysis of the roads would take place. We had to use at least 20 according to NSSDA standards in order to have a valid statistical analysis. Below is a map of my test points. I tried to spread my test points across the data set while finding intersections with 90 degree angles. The 90 degree angles make it a little easier to see where the actual intersection should be. 

After creating test sites we then created new feature classes and digitized the intersections of each intersection at the selected test site. Digitizing required placing a point at the intersection for each data set and a reference point that we selected for each test sight. Once the digitizing was complete we could then add XY coordinates to all of our digitized points. The XY coordinates were then placed in an Excel spreadsheet where all the calculations were carried out. The Sum, Average, Root Mean Square Error and NSSDA error were all calculated using simple formulas in the spreadsheet. Calculations were based on the reference points that we created using provided ortho images. 

The following are my accuracy statements for each of the datasets that were provided. 

Accuracy Statement for ABQ City Data:

Using the National Standard for Spatial Data Accuracy, the data set tested 4.93 meters horizontal accuracy at 95% confidence level.

 

Accuracy Statement for StreetMaps USA Data:

Using the National Standard for Spatial Data Accuracy, the data set tested 56.05 meters horizontal accuracy at 95% confidence level.

As is evident from the accuracy statements the ABQ City data was much more accurate than the SteetMaps USA data. Because we selected and created our own reference points on the fly there is likely some room for improvement but there would still likely be a significant difference between the provided datasets.