One of the most common questions we hear when providing drone mapping and drone surveying is “what kind of accuracy can we expect in a final product”?
In order to assure our deliverables meet required accuracy we utilize checkpoints randomly distributed throughout a project site to test the point-cloud-derived surface for accuracy. These checks are done in accordance with the ASPRS standard for a 10 cm Vertical Accuracy class, which is appropriate for a 1′ (30 cm) Class 1 contour, per the ASPRS 1990 standard.
Applying the 1990 ASPRS Accuracy Standards dictates that accuracy for spot heights in open terrain be 1/6 of the contour interval, so for a 30 cm (1′) contour interval, the required spot height accuracy, RMSEz, is 1/6 * 30 cm, or 5 cm (.16′). RMSE in one dimension is simply the average of the absolute values of the errors (sqrt(z^2)). We require an average RMSEz below 5 cm (0.16′) for our open terrain random check points in order to provide 1′ contours of a site.
According to the 2014 ASPRS Standard, the resulting surface, with a 10 cm Vertical Accuracy Class, has an RMSEz of 10 cm in Non-Vegetated Terrain, such as bare dirt, short grass, and paved surfaces. (The majority of the South West)
Vertical errors associated with vegetated areas cannot be represented by a normal distribution, so the ASPRS Standard uses the National Standard for Spatial Data Accuracy (NSSDA) methodologies to compute a 95% confidence level value of 19.6 cm for Non-Vegetated Vertical Accuracy (NVA) and 30 cm for Vegetated Vertical Accuracy (VVA) for the 10 cm Vertical Accuracy Class.