NivaarExam PrepOfficial exam papers ↗

18-Geom-A1 Surveying · May 2014

Question 5 of 7: Single-, Double- and Triple-Differencing in GPS Surveys

Nivaar worked solution (AI-drafted; not reviewed by a licensed engineer)

Notes on this paper

National Exams — May 2014 — 04-Geom-A1 Surveying. Closed-book; any non-communicating calculator, ruler and protractor permitted. Format: seven questions are given and any five (20 marks each) constitute a complete paper — all seven are solved below for completeness. Where a datum is implied, elevations are referenced to the Canadian vertical frame (CGVD2013) and azimuths to NAD83(CSRS); US-foot stationing is retained wherever the printed question uses it.

Reference texts: Wolf & Ghilani, Elementary Surveying: An Introduction to Geomatics (15th ed., Pearson); Ghilani, Adjustment Computations: Spatial Data Analysis (6th ed., Wiley); Hofmann-Wellenhof et al., GNSS — Global Navigation Satellite Systems (Springer, 2008).

Question 5: Single-, Double- and Triple-Differencing in GPS Surveys (20 marks)

Question text not reproduced: the examination questions are © Engineers and Geoscientists BC. Open the official past paper (linked at the top of this page) to read the question, then follow the worked solution below.

Given. Carrier-phase observations collected simultaneously by two receivers ($A$, $B$) tracking common satellites in a relative (differential) GPS survey.

Find. A sketch and a principle statement for each of single-, double- and triple-differencing.

Carrier-phase positioning is far more precise than code positioning but is corrupted by three dominant nuisance terms: the satellite-clock error, the receiver-clock error, and the integer ambiguity $N$ (the unknown whole number of cycles at lock-on). Differencing forms linear combinations of the raw phase observations $\Phi$ that algebraically remove these terms one layer at a time. The essential requirement is simultaneity: the same satellites are observed by both receivers at the same epochs, so the shared errors are common and cancel.

Single differencing (between receivers)

ABkSingle difference Φᵏ_A − Φᵏ_B (cancels satellite-k clock)
Figure 5a — Between-receiver single difference: one satellite $k$, two receivers $A,B$.

Subtract the simultaneous phase of the same satellite $k$ observed at the two receivers: $$\Phi^{k}_{AB} = \Phi^{k}_{A} - \Phi^{k}_{B}$$ Because both receivers see the identical satellite clock at that instant, the satellite-clock error cancels. Most of the correlated atmospheric (ionospheric/tropospheric) delay also cancels over short baselines. The receiver-clock difference and the ambiguity difference remain.

Double differencing (between receivers and satellites)

ABklDouble difference (Φᵏ_A − Φᵏ_B) − (Φˡ_A − Φˡ_B)cancels both satellite and receiver clocks
Figure 5b — Double difference: two satellites $k,l$, two receivers $A,B$ (four one-way phases).

Difference two single differences formed to two satellites $k$ and $l$: $$\nabla\!\Delta\Phi^{kl}_{AB} = \left(\Phi^{k}_{A}-\Phi^{k}_{B}\right) - \left(\Phi^{l}_{A}-\Phi^{l}_{B}\right)$$ The second subtraction removes the receiver-clock errors as well (they are common to both single differences). What is left is the geometry plus an integer double-differenced ambiguity $\nabla\!\Delta N^{kl}_{AB}$, which is a true integer and can be "fixed." The double difference is the fundamental observable of precise relative GPS.

Triple differencing (between epochs)

ABklTriple difference DD(t₂) − DD(t₁) (cancels N — flags cycle slips)
Figure 5c — Triple difference: the double difference at epoch $t_2$ minus that at epoch $t_1$.

Difference two double differences formed at two epochs $t_1$ and $t_2$: $$\delta\nabla\!\Delta\Phi^{kl}_{AB} = \nabla\!\Delta\Phi^{kl}_{AB}(t_2) - \nabla\!\Delta\Phi^{kl}_{AB}(t_1)$$ Provided no loss of lock occurred, the ambiguity $N$ is constant between epochs, so it cancels entirely. The triple difference is therefore ambiguity-free and ideal for an initial approximate solution and for detecting cycle slips (a slip appears as a spike), at the cost of higher noise and correlated errors — hence it seeds, but does not replace, the double-difference fixed solution.

MethodCombinationCancels
Single differencesame satellite, two receiverssatellite-clock error
Double differencetwo satellites, two receivers+ receiver-clock error (integer $N$ preserved)
Triple differencedouble difference across two epochs+ integer ambiguity $N$ (flags cycle slips)