“Positional Accuracy is the degree of compliance with which the coordinates of points determined from a map agree with the coordinates determined by survey or other independent means accepted as accurate.”Glossary of Mapping Science (ASPRS and ASCE, 1994).
It is a measure of how objects are positioned relative to each other. It uses sampling to estimate the discrepancy between a map and its true location on the earth’s surface. Therefore, it is a measure of maximum allowable deviations between two spatial objects on the map and on the ground.
To assess positional accuracy, two layers are required: the layer whose accuracy you want to evaluate and another layer that can be used as a point of reference. For instance, we can compare the location of roads in a feature class with their location in a TIFF image.
Factor affecting Positional Accuracy
There are several factors that affect the positional accuracy of the geo-referenced images. Some of them are:
- Distortion in sensor lens
- Tilting of aircraft carrying the sensor
- Topography on remotely sensed imagery
Due to the presence of these errors in georeferenced images, it is important to do an accuracy assessment. While doing a positional accuracy assessment, we try to characterize the accuracy of a geospatial dataset. Two things should be considered while doing a positional accuracy assessment, 1) the accuracy of the sampled data, and 2) the level of accuracy we want to obtain.
Approaches for Positional Accuracy Assessment
There are several approaches for assessing the positional accuracy. One of the best and widely accepted approaches is the mean root square of squared differences between the map and the reference point. This term is called the “root-mean-square-error” or RMSE. RMSE is estimated from a sample of the map and reference points. It measures the differences between values (sample or population values) predicted by a model or an estimator of the values observed. The RMSD represents the square root of the second sample moment of the differences between predicted values and observed values or the quadratic mean of these differences. In other words, it tells you how concentrated the data is around the line of best fit. Root mean square error is commonly used in climatology, forecasting, and regression analysis to verify experimental results.
All positional accuracy parameters are estimated by comparing reference coordinates or elevation to the map or elevations to the map or image coordinates or elevations of the data set being assessed at each sample location.