6.3 GNSS Accuracy
There are different levels of accuracy of GNSS measurements, and these are directly related to the technique and equipment used, as well as the length of the observation times.
SP1
In Australia, the way we describe and determine accuracy is outlined in the Intergovernmental Committee for Surveying and Mapping (ICSM) Special Publication 1 document, which is referred to as SP1. ICSM is a group made up of surveyors and other spatial science professionals from each of the State and Territory Governments, along with representatives from Geoscience Australia.
SP1 is the document that sets out the standards for survey control (permanent survey marks) in Australia, and has six guidelines that cover the different methods of surveying that can be used to generate coordinates for permanent survey marks. These guidelines provide information on the methods and techniques needed to achieve different levels of accuracy, which is defined by the principle of uncertainty.
Uncertainty
Uncertainty is quite a simple idea – it is the measure of how wrong a value (in the case of GNSS, the coordinates) may be. It is expressed in the same units as the values, so in the case of GNSS it is usually expressed in metres.
For example, if the plane projection coordinates of a permanent mark were 1000m E and 50,000m N, with an uncertainty value of 0.05m in the horizontal that would mean the coordinates of the mark could be wrong by up to 0.05m.
Uncertainty is expressed as a standard deviation, but one that has been expanded to the 95% confidence interval. Remember from Module 0 the area under the curve of the normal distribution is about 68% at one standard deviation from the mean.
What 95% confidence interval actually means is that we are 95% confident that a sample of data will contain the true mean of the population of the data, even though we don’t know what the whole population looks like.
We achieve the expanded confidence interval by multiplying one standard deviation by a coverage factor, which is basically a scaling factor to take one standard deviation (which is about 68%) up to 95%. The coverage factor is represented by k.
The formula for uncertainty is:
95% CI = 1 x k
Where 95% CI = 95% Confidence Interval
= standard deviation of sample
1.996 = coverage factor for one dimension
Uncertainty in two dimensions
For horizontal coordinates, we have two dimensions, so we can describe uncertainty as an ellipse, where one axis is representing the uncertainty of one horizontal direction (such as Easting), and the other ellipse axis represents the other horizontal direction (such as Northing).
Because we are dealing with data in two dimensions, and each has its own standard deviations, we have two normal distribution graphs that combine to mean the area under them is three dimensional. The easiest way to visualise this is to imagine two normal distribution curves are placed at a right angle to each other, their x axes are now the axes of the ellipse, like in f