ABQ vs StreetMapUSA: Which Roads are More Accurate?

For this lab I compared the positional accuracy of two road datasets, the ABQ_Streets_Sample and the StreetMapUSA_Sample. The reference data for this comparison was the 2006 orthophotos.

I started by creating 20 test points at road intersections that were spread evenly across the study area. To make sure the points were balanced, I divided the map into four quadrants and placed about five points in each section. I then added matching points to both the ABQ streets layer and the StreetMapUSA layer. After that, I digitized the true locations of the same intersections using the orthophotos.

Once the points were created, I exported the X and Y coordinates into Excel. I calculated dx and dy for each point and then used those values to find the error distance between the test points and the reference points. After calculating all 20 errors I used them to find the RMSE and then multiplied that by 1.7308 to get the 95 percent accuracy value.

The results showed a big difference between the two datasets. The ABQ_Streets_Sample had an RMSE of 12.8 feet and a 95 percent accuracy of 22.2 feet. The StreetMapUSA_Sample had an RMSE of 162.7 feet and a 95 percent accuracy of 281.6 feet. The city dataset was much more accurate than the StreetMapUSA dataset. This shows why local government data is often the better choice for projects that require higher positional accuracy.