If you were called upon to use static GPS methods to establish geodetic control that meets Federal Geodetic Control Subcommittee (FGCS) standards, there is a document, about 50 pages in length for you to follow [1]. There are no national standards and specifications for surveying using RTK. The California Department of Transportation (Caltrans) has a section in its Surveys Manual, Chapter 6B called "Real Time Kinematic (RTK) GPS Survey Specifications," and the National Society of Professional Surveyors (NSPS) is working on specifications for using RTK for boundary surveying. But, for the majority of surveyors, RTK is a new world of surveying where every point observed is a side shot (radial shot).

Coming from a surveying education background, it is easy to be critical of the way some people survey using RTK GPS receivers. When we as educators taught the first course in surveying, typically called "Introduction to Plane Surveying" or something equivalent, the field procedures were drilled into the students. Some of the more common rules were:

  • never run an open traverse
  • always double-run a level line if your closing mark is not an established bench mark
  • double all angles measured with a theodolite or total station
  • measure all instrument heights twice-the same for prism rod heights
  • always make certain you have a check on all observations
and the list goes on. When it comes to kinematic surveying, however, many surveyors seem to ignore the necessary standards to ensure the accuracy of their RTK work. This article will review the original procedure for kinematic surveying, as well as what standards and specifications are required for its use today.

Figure 1. Typical stop-and-go kinematic surveying procedure.

Kinematic Surveying

To review the concepts of kinematic surveying, I must start with the methodology used before real-time processing became available (the receivers used then were L1-capable only). Kinematics, as applied to surveying, was developed by Dr. Benjamin Remondi and described in a PhD dissertation at the University of Texas at Austin in 1984 [2]. Figure shows my rendition of kinematic surveying.

The requirement for starting a kinematic survey, at that time, was a short base line that had been observed using static GPS. Short usually meant less than 10 km. There had to be a minimum of five visible satellites (it would work with four satellites but accuracy degraded). A standard carrier-phase GPS receiver was positioned on a tripod over one end of the line at the base station. Another standard carrier-phase GPS receiver was positioned over the other end of the line at the initialization point; this receiver was called the "rover." The rover had an external antenna mounted on a prism or range pole, and this was positioned over the initialization point with a pole bipod or tripod; a cable ran from the antenna to the receiver. This receiver had to be carried by the surveyor, so it had to be in front of him or her so the receiver display could be seen. For this example, I'll describe the survey using a Trimble 4000ST receiver, which had a screen menu for kinematic surveying.

The observing procedure was as follows:

  • After the receiver antennas were positioned over the ends of the known base line, each receiver was turned on. Both receivers were set to record data every 15 seconds.
  • The purpose of the base station was to gather code and phase data, and log this data into memory. The station number 1000 was input into the receiver.
  • After the rover initialized, the surveyor pushed the "Kinematic Survey" button on the receiver, the receiver asked for a point number and the usual input was 2000.
  • Every 15 seconds, the rover would show the number of the epoch that was recorded (1, 2, 3"¦). Common practice was to record five epochs of data.
  • After the fifth epoch was recorded, the surveyor pushed a "Stop" button on the receiver. "Stop" did not mean the survey had ended-it meant that no additional data was to be recorded at the initialization point. The surveyor then picked up the antenna assembly (antenna, pole, and bipod or tripod) and walked (drove) to kinematic point No 1.
  • The move from the initialization point had to be unobstructed; at no time could the rover have less than four satellites visible. When the surveyor arrived at kinematic point No 1, he positioned the antenna assembly directly over the point, then pushed the Kinematic Survey button on the receiver; the receiver asked for a point number, and the input was 1.
  • After five epochs of data were collected, the rover moved to kinematic point No 2, then kinematic point No 3, and so on.
  • It was absolutely necessary to occupy at least one of the kinematic points a second time. As Figure 1 shows, if possible, users should always return to the original initialization point. Today, good field procedures require each point to be occupied a second time.
  • If at any time the rover had less than four visible satellites, the receiver would give a warning. It was more than a warning-it meant the rover had to reinitialize at a point that had already been surveyed. For example, when moving from kinematic point No 3 to kinematic point No 4, the receiver moved too close to a tree that caused the number of visible satellites to drop to three. The surveyor would move back to kinematic point No 3, place the antenna assembly over that point, push the Kinematic Survey button on the receiver and enter point No 3. Five epochs of data were recorded, and the move to kinematic point No 4 was by a route that had no obstructions.

Figure 2. Kinematic observed lines.
After all kinematic points had been occupied, the rover was turned off, the base station receiver retrieved, and both receivers were taken to the company office for data reduction. With the advent of kinematic surveying, data processing was not easy. Sometime prior to 1990, I spent four days at a training session in Sunnyvale, Calif., learning how to process kinematic data. This was new technology, and I believe there were only six people in the class. The software was DOS-based, and a tremendous amount of keyboard strokes was required. The output from a kinematic GPS survey is the same as for a static survey, which is the vector from the base station to the rover (shown in Figure 2). The positions of kinematic points 1 through 5 are all determined from the base station; i.e. you have taken a side shot to each point. That is why it is important to occupy surveyed points more than once.

Figure 3. Real-time kinematic observed lines.

Real-Time Kinematics (RTK)

It took four different improvements in GPS receiver technology to get to where we are today with RTK:

  1. Manufacturers of GPS receivers developed the technology to pick up the L2 signal. (Note: By L2 signal I mean the L2 carrier modulated with the navigation message and the P(Y)-code.)
  2. The ability to transmit by radio all code and phase data gathered at the base station to the rover. The rover was equipped with a radio receiver and a data collector/computer that could compute the position of the rover in real-time.
  3. On-the-fly (OTF) ambiguity resolution (see Figure 3).
  4. Wide laning (see Figure 3).

Referring to Figure 3, notice that a known base line is not required to initialize the survey. What was the initialization point in Figure 1 is the base station in Figure 3. The observing procedure is as follows:

  • Position the base station antenna directly over the control point that was established by static GPS. The radio transmitter with antenna, etc. must also be in place.
  • The surveyor with the rover moves back a short distance from the base station, about 10 to 15 feet, and turns on the receiver. The receiver uses wide laning, which is L1 minus L2, to get a measurement wave of 86 cm wavelength. With a measurement wave of 86 cm, a base line vector from the base station to the rover can be determined in just a few seconds. The survey is now initialized and the surveyor moves to kinematic point No 1.
  • The surveyor positions the rover antenna over kinematic point No 1 and collects several epochs of data. In this case the data collection rate is more frequent than the example given in Figure 1; it may be more than one epoch per second to one epoch every several seconds. The standard produced by Caltrans specifies data collection rate.
  • It is not necessary to have a clear view of the sky when moving from the initialization point to kinematic point No 1, or from point No 1 to point No 2, etc. In fact, the surveyor can carry the antenna pole horizontally as he moves from one point to another. That is because the receiver has OTF ambiguity resolution, the ability to resolve ambiguities without reinitializing on a previously surveyed point.
  • Because each position determined is a side shot from the base station, it is mandatory to make repeat observations on most if not all occupied stations.

Table 1. Caltrans third-order RTK survey field specifications.

Standards and Specifications

I would encourage anyone working with RTK to look at the section in the Caltrans Surveys Manual on its website at www.dot.ca.gov/. Click on "Manuals and Forms," then on "Surveys Manual." Chapter 6B is "Real Time Kinematic (RTK) GPS Survey Specifications." Table 1 is an abstract of a table contained in Chapter 6B. Figure 5.1 in the Caltrans Surveys Manual lists the applications for third-order RTK as: topographic surveys (data points); supplement design data surveys; construction surveys (staked points); environmental surveys; GIS data surveys; and right of way flagging.

For third-order RTK surveys, each new station should be occupied twice. The second occupation of a new station should use a different base station location. If the new station is being elevated by RTK methods, the second occupation of the new station should have a minimum of three different satellites in the satellite constellation. This is generally achieved by observing the second occupation at a time of day that is either four hours before or four hours after the time of day of the first occupation.

My concern in writing this article is that many surveyors using RTK don't want to see the complexity of these specifications. They are satisfied with taking single shots and, sometimes, not checking back on checkpoints. As can be seen from the above information, RTK is not held in high esteem with respect to accuracy. Conscientious surveyors will recognize the proper applications of RTK surveying, and will carefully observe the procedures required to gain the highest level of accuracy when using this method.