Module 1.2: Data Quality-Standards

 Module 1.2: Data Quality-Standards

This weeks module is looking at positional accuracy of road networks. The data for the roads came from Albuquerque and TeleAtlas for Street Map USA. The Albuquerque dataset is considered pretty accurate while the TeleAtlas is distributed by ESRI in ArcGIS. We were supposed to look the different street views to see where they were similar and where they diverted from each other. It was interesting to see the ESRI Street Map USA seemed further away from the "true" data points. One would think the street views that ESRI produces would be very accurate. The way we put the "true" data points in were but doing a close up of the orthoimages to see where the streets seemed to be located. The image below shows data points and street views of both Street Map USA and Albuquerque data sets.

I did 40 data points because it was hard to get the sampling rules of 20% in each quadrant and >10% of diameter apart at the 20 data points. The way I did the accuracy assessment was using the table on Page 5 in the Positional Accuracy Handbook within excel. In order to make sense of the data I switched it to UTMs to get the Easting and Northing in meters. In the handbook we are supposed to see the difference between the test points and the "true" points of both the x and y coordinates. Also, we had to square both of the x and y coordinate differences. Then we would add together the squared differences and then add it all up to get the sum. After that we would get the average which is the sum/number of points. We would get the RMSE which is the average^1/2. Lastly, we would get the NSSDA which is 1.7308 * RMSE.

Once we got all that information we needed to write our final accuracy statement. This statement does change between a tested and complied to meet. The tested statement is used when the accuracy was determined by comparison with an independent data set of greater accuracy (Positional Accuracy Handbook, pp. 5). The compiled to meet statement is used when the data has been thoroughly tested and that method produces a consistent accuracy statistic (Positional Accuracy Handbook, pp. 5).  Based on the way these points were collected we would use the test statement. For the Streets Map USA the statement is Tested 306.9 meters horizontal accuracy at 95% confidence level and the Albuquerque Streets is Tested 16.7 meters horizontal accuracy at 95% confidence level.

By creating the different data points it helps know the positional accuracy of your shapefiles. It is important to try to get a higher accuracy data set to compare. After doing all the data points and getting the math correct it seems that the Albuquerque data set is more accurate than the Street Map USA data set. This is interesting because the Street Map USA dataset is used within ESRI ArcGIS and one would think that is more accurate then other datasets.