Lab 1 - Calculating Metrics for Spatial Data Quality

Here we go! The first lab of GIS 5935: Special Topics in GIS, and the first lab of the fall semester. The lecture and readings went over the fundamentals of spatial data quality, including learning about precision, accuracy, bias, resolution, and the FGDC's National Standard for Spatial Data Accuracy (NSSDA).

The first part of the lab assignment consisted of determining the precision and accuracy of a set of waypoints collected by a GPS unit. Precision is measured by finding the number of observations that fall within a certain distance. For horizontal precision, I did a spatial join to create a distance field between the waypoints and the average waypoint location, then created buffers based on percentiles of those distances. For this lab, 68% was the precision threshold, or 4.4 m. Accuracy is looking at how close observations are to a true reference value. For horizontal accuracy, I measured the distance between a provided reference point and the average waypoint and compared that to the precision threshold to determine accuracy.

Comparing the horizontal and vertical precision and horizontal and vertical accuracy I found that on average, the waypoints are more horizontally precise as the horizontal precision value (4.4 m) is less than the vertical precision value (5.8 m). For horizontal and vertical accuracy, again, the data is more accurate horizontally, as the horizontal distance between the reference point and average waypoint location (3.8 m) is less than vertical distance (6.0 m) between the two.

I also looked at vertical precision, again using 68% as the threshold, and then used a provided reference point to consider accuracy of the waypoints. The second part of the lab looked at calculating various error metrics, such as RMSE and creating a CDF for comparison to the table metrics.

Horizontal precision buffers for 50th, 68th, and 95th percentiles of distance values for waypoints to average waypoint.