How to do a GPS survey to centimetre accuracy

Site Survey using RTK

The method of choice for a centimetre-accuracy site survey is RTK (Real Time Kinematic).  RTK involves the use of 2 GPS receivers (stationary Base Station and a Rover) communicating together via a radio link.  The base station must be located such that it will have a clear view of the sky and a continuous line-of-sight to the rover.   The precision of the Rover position relative to the Base station is dependant on baseline-length so it is desirable to keep the baseline as short as possible (<10 km).

RTK performs Real Time Phase Differential and computes the 3D vector (D X, DY, DZ) between the rover and base antennas.  Base Station coordinates and both antenna heights need to be entered to compute ground coordinates at the Rover.

How does RTK work?

Data from the Base Station (either raw GPS data or RTK corrections; there are different RTK approaches) is sent in real time via radio to the Rover.  With a sufficient number of common satellites visible at both GPS antennas, a “FIXED” solution of centimetre-level precision can be calculated. 

If there are insufficient common satellites, a “FLOAT” solution of lower precision (a few decimetres) is calculated.  You can try waiting for a “FIXED” solution or re-initializing the system (consult your manual) but these may not be successful as most often the FLOAT solution is due to poor satellite visibility at the Rover.

RTK systems are available in dual-frequency and single-frequency versions.  Dual-frequency systems deliver greater precision, faster and over longer baselines than single-frequency systems.  Your choice in equipment will ultimately depend on your budget, expected baseline lengths and precision needs.  See manufacturer’s specifications.

Here are typical horizontal precisions (RMS) you can expect for various baseline lengths under ideal conditions:

Do I need to do post-processing?

It is recommended (but not essential) to have Phase Differential Post-Processing software.   If your Rover loses radio-communication with the base (due to an obstruction or to excessive distance) you could still produce corrected Rover positions by post processing.  Set your base station to record raw GPS data during the entire RTK survey.  If the rover loses the radio link, record raw data at the rover (as per manufacturer’s recommendations) and perform Phase Differential Post-Processing along with the base station raw data.

Phase Differential Post-Processing could be used alone instead of RTK to do a site survey (it’s the same GPS equipment minus the radio link) however the user would lose the advantages of RTK (feedback on equipment performance and final results in real-time).

How can I determine accurate Base Station coordinates?

1. One method commonly used has been to tie the Base Station to an ACP (Active Control Point) using Phase Differential Post-Processing.  

   

   Phase-Differential post-processing to an ACP

   Because precision is dependant on distance this method is recommended only if the ACP is nearby (within 30 km***).  Good Base Station coordinates could then be attained with a relatively short observation period. ***Please note that 30km is a general guideline only.  Distance/ Observation Time/ Precision will depend on the equipment and the software used.  See software manufacturers’ recommendations.  We know some scientific software is capable of providing a precision of a few cm’s over a 100-200 hundred kms if used properly.
   Base Station coordinates will be in the same datum as the ACP so the user must ensure the ACP coordinates and datum are correct.

2. A fast-growing number of GPS users are using single point positioning services such as NRCan’s CSRS-PPP.

   

   CSRS-PPP

   CSRS-PPP is an easy-to-use, free online service that can provide accuracies that rival Phase Differential Post-Processing.   CSRS-PPP ties you to the precise GPS satellite orbits and produces corrected coordinates of a constant “absolute” accuracy no matter where you are on the globe, regardless of proximity to a CACS station.
   Here are typical horizontal accuracies (RMS) you can expect with Static CSRS-PPP:

   Static CSRS-PPP Horizontal accuracy table

   	2-3 hrs of observation	12-24 hrs of observation
   Static CSRS-PPP (Dual frequency receiver)	<5 cm	<1 cm
   Static CSRS-PPP (Single frequency receiver)	<50 cm	N/A

When should I use CSRS-PPP, before or after the RTK survey?

It is preferable to compute your base station’s coordinates before starting the RTK work.  You could setup the base station ahead of time and collect raw data (for anywhere from 2hrs to 24 hrs depending on accuracy required).  Leave the base station in place.  Transform the raw data to RINEX and submit it for Static CSRS-PPP processing.  CSRS-PPP can process the data approximately 90 minutes after data collection.  Once you have your accurate coordinates from CSRS-PPP you can begin the RTK work.

If it is not possible to collect raw data and run CSRS-PPP ahead of time, the RTK survey can still be performed entering the Base Station’s approximate coordinates. Ensure that base station raw data is recorded (uninterrupted) for as long as possible.  Run Static CSRS-PPP after the RTK survey, compute the difference (between approximate and PPP coordinates) and apply it (a 3D shift) to the entire survey.  A minor drawback here is that for each ten meters of error in the Base Station position an additional 1ppm (1mm per kilometre) error is introduced in the baseline computation.

A popular methodology is to establish accurate coordinates for 2 points within the survey area (a fair distance apart and inter-visible).  Collect GPS raw data simultaneously at both points and post-process both using static CSRS-PPP.  Also process both files as a baseline (distance, azimuth) using your phase differential post-processing software.  From the CSRS-PPP output coordinates you can compute the distance and azimuth between points as well (program INDIR, inverse solution).  Comparing these results can serve as quality control plus you now have 2 geodetic control points to choose from.  Make one of them the main Base Station location; the survey will be anchored to this point.  The other point can be tied with the Rover during the RTK survey (as an additional check) and used as a back sight for traditional surveying.

What essentials do I need to know about CSRS-PPP?

• Raw GPS data is usually logged in a format (ascii or binary) that is proprietary to the GPS manufacturer.  All manufacturers should provide software to transform their raw data to the RINEX format (Receiver-INdependent Exchange, the recognized standard for raw GPS data). 
• You must ensure the RINEX file header contains the correct antenna height, instrument serial#, antenna serial# and antenna type.  These can be entered either during the reformatting process or manually using a text editor (RINEX is an ascii file).
• The most common errors relate to antenna height (error measuring antenna height or not properly identifying the antenna type in the RINEX header).
• CSRS-PPP processing can be done in two modes: Static or Kinematic.  Static produces one corrected averaged single point, Kinematic produces a corrected track.
• Output coordinates can be in either NAD83(CSRS) or ITRF/WGS84
• Minimum duration of observation:  Under ideal conditions it takes an initial period (typically 60 to 90 minutes without loss-of-lock) to get initial convergence.  Static accuracies of approx. 5cm (dual-frequency receiver) and 20-50cm (single-frequency receiver) can be attained.   Under less than ideal conditions, one should increase this period. 
• Longer observations:  With dual-frequency receivers (only) accuracy will continue to converge/improve if you observe for a longer period (make sure not to lose lock).  Dual-frequency receivers can attain approx 1 cm accuracy with 12 - 24 hours of data in static mode.  You don’t need a huge amount of data; keep your files small by logging at 5, 10 even 30-sec interval.
• GPS receivers are not all created equal so it’s impossible to generalize.  Test your equipment at a known location, compare your results with the published coordinates and look at the graphs for speed of convergence in all 3 orientations (lat, long height).