In this week's lab, we were tasked with determining the precision and accuracy of a set of recorded coordinate positions and subsequently generating a map of our results.
Horizontal accuracy is measured as the distance between an experimental recorded measurement and a known reference measurement. In this lab, as you will see, we were able to calculate our accuracy by creating an average representative point from a large volume of data and measuring its distance to a known reference point.
Horizontal precision is the proximity of measured values to one another. The closer the proximity, the greater the precision.
To determine the precision, I first uploaded the point feature class of all recorded measurements. I then generated a new point feature class which represented an average coordinate position (point) of all of the represented measurements. The value used was determined by finding the mean x coordinate value and mean y coordinate value of all provided points. This was then plugged in when creating a new point feature for our feature class.
Next, I needed to determine the distance between each point and the averaged point. I reprojected each feature class to ensure that the proper units were able to be used. I then performed a spatial join, with the join set to closest, so a new distance field representing the distances for each point from the averaged point would be added to its attributes.
From this I created a buffer after determining which values represented the desired typical standard deviation of accuracy: the 50th, 68th, and 95th percentile.
This would should us how many of the points resided in each stratification. 50% of all values would be held within one distance, 68% within another, and 95% within a much broader perimeter.
The values representing the horizontal precision were:
50th percentile= 3.1meters
68th percentile = 4.5 meters
95th percentile= 14.8 meters
With these values I created the buffers represented on the subsequent map.
We were then provided a reference measured point which was to be used as the baseline measurement for accuracy. able to measure the distance between the two points. It turned out that our average measurement for all recorded locations was off by 3.24 meters.
This accuracy is somewhat close to the distance 50% of our measurements are contained within (3.24m vs. 3.1m). This gives an indication that there may be a bias of roughly 3 meters.
Comments
Post a Comment