GNSS (GPS) accuracy explained

September 2011

GNSS (Global Navigation Satellite Systems) such as GPS allow for fairly simple location-based data collection. But because of the way these systems work, the statistical accuracy of the location-based data collected through such systems can vary significantly.


For the United State's GPS (Global Positioning System) in particular, some information in this regard is provided at the following web page.


Accuracy of GNSS (GPS) data at the time of collection can vary significantly due to several reasons, including but not limited to the following.


  • Atmospheric (mainly Ionospheric) signal interference
  • Number and angle of supported constellation satellites in view
  • Signal "Multi-path" propogation / reflection
  • Differential correction sources used (such as SBAS)
  • Connected antenna location and configuration


The specifications (data sheet) for GNSS receivers usually state accuracy using terms based on statistical probability or confidence, somewhat similar to as listed in the following basic table.


Term Definition Accuracy
Probability / Confidence
CEP Circular Error Probable 50%
RMS Root Mean Square 63–68%
2DRMS Two times the distance of RMS 95–98%
R95 Radius 95% 95%


To explain further, if the specifications on the data sheet of a GPS receiver state a horizontal (distance) accuracy of 3 - 5 meters CEP, this means that during testing the location coordinates of data points collected with the receiver were found to be within 3 - 5 meters 50% of the time. Though in such testing the manufacturer may have compared data points to the averaged coordinates of several previously collected data points, to the certified surveyed coordinates of a physical marker, or to some other reference. It depends on the test criteria of the manufacturer.


For example, the test criteria for the Hemisphere GPS receivers (such as the XF101 for the Archer Field PC) is provided at the following web pages.



So it needs to be understood that the stated accuracy of a GNSS (GPS) receiver is in relative terms. When comparing the accuracy of different receivers from different manufacturers to each other (particularly those that state to provide higher accuracy such as sub-meter or better), it is important to understand the test criteria of the stated accuracy for each receiver. It is also always beneficial to simply perform your own testing with each receiver in your own applicable operating conditions.


The stated horizontal accuracy (again not necessarily directly comparable) of several common receivers is listed in the table below.


Receiver Stated Horizontal Accuracy
Hemisphere GPS XF101 (Crescent OEM P102, for original Archer) <0.4 meters RMS (DGPS)
<1.8 meters RMS (autonomous)
<0.3 meters RMS (DGPS with external antenna)
NVS NV08C-MCM (built-in to previous Archer 2 and Allegro 2 units) 2.5 meters RMS (autonomous)
1 meter RMS (differential mode (RTCM))
GlobalSAT BC-337 (for original Archer) 10 meters 2D RMS (autonomous)
5 meters 2D RMS (DGPS using WAAS)
uBlox NEO 5Q and 6Q (built-in to Mesa and AMXU pod) 2.5 meters CEP (autonomous)
2.0 meters CEP (DGPS)
IkeGPS <0.6 meters RMS (DGPS)
<1.5 meters RMS (autonomous)
Older units, <2.5 meters CEP (autonomous)


  • Autonomous = Using GNSS (such as GPS and/or GLONASS) only for positioning
  • DGPS = Using GNSS with a correction source such as SBAS (like WAAS or EGNOS)


Most receiver manufacturers do not provide a stated vertical (altitude) accuracy specification. But as an observed measure, vertical accuracy of GNSS (GPS) receivers seems to typically be 1.7 times that of the stated horizontal accuracy. So using the previously stated example, a receiver with specs that state 3 - 5 m CEP horizontal accuracy would likely provide 5.1 - 8.5 m CEP vertical accuracy. But again this is just a general observation in brief encounters with many different receivers, and is not based on thorough testing of any one particular receiver make and model.