Special Topics in GIS Module 1: Calculating Metrics for Spatial Data Quality

 

In this week’s lab, we learned how to examine map data to determine precision and accuracy. When working with map data, we can assume that the larger the values in our data, the less precise the map data will be. Similarly, for accuracy, we can assume that the larger the distances within our map data, the less accurate the data will be. We rely on the data from the attribute table to measure horizontal accuracy and precision. The distance values in the table are sorted by smallest to largest. The total number of values is multiplied by the desired percentile, (i.e.: 50%, 68%, 95%, etc.) to determine the index number. The index number will help guide us to which row to reference in the attribute table, by counting down to that row from the top of the table. We take the average of the distance identified in the index row and the row below it, and this average distance gives us our distance accuracy value.  

I measured the distance between the Average Location Point and the Reference Point to be 3.49 meters. I calculated the horizontal precision (68%) to be 4.29 meters. The difference of which is about 0.8 meters. The distances measured for this map were relatively small, so I did anticipate high accuracy. The results are 80% accurate. I would have guessed that the 68% distance index would have been off by half a meter or so, give or take, which is close to my results. It’s possible the accuracy could be better, and the data could be re-examined to improve the accuracy of the map.

I did find there to be a notable difference of 0.8 meters between the results for horizontal accuracy and precision. The values in the data sets are low, and the distances between the GPS waypoints are relatively small, so better precision and accuracy were anticipated with this dataset.

I calculated vertical precision to be 5.89 meters. My results for the average elevation compared to the elevation reported for the reference point were 5.88. I think these results indicate a high accuracy within the map for the elevation data because the difference between the two numbers was so close. 

I think that there may be evidence of bias for the horizontal data because the horizontal results were off by 0.8 meters. On a scale this small, I think the accuracy could be improved. However, I do not see evidence of similar systematic error with the vertical data results.