GPS Q&A: Industry experts answer reader's GPS questions

Q. How reliable are GPS heights? N.C. San Diego, Calif.

A. Henri Ayers, Leica Canada Inc.: GPS height differences are reliable to about 2 to 3 ppm (2 to 3 mm per km) with respect to the WGS-84 (GPS) reference ellipsoid. They become less reliable when they have to be referred to the Geoid model used to compute the local geoid separations for the GPS positions. Moreover, additional degradation to local geoid separations may occur in mountainous areas. In general, the assessment of GPS height reliability must be conducted on a local basis by comparing orthometric heights derived from GPS using the best available geoid model with the ones related to the vertical datum within the area. The smaller the area, the better the agreement will be, especially in mountainous areas (up to 1 decimeter).

John C. Bohlke, Sokkia Corp.: The reliability of GPS heights can be expressed through statistical calculations. A confidence level is associated with various statistical expressions to represent the reliability of the data. Processing the data and performing a network adjustment often leads to an accurate statistical estimate of vertical reliability. The best way to ensure reliability is by measuring multiple baselines to each point and using local vertical control points. The redundancy of these calculations will enable the adjustment program to identify any inconsistent measurements thereby establishing a very high confidence in the heights. If the statistical estimate represents only the confidence of the horizontal measurements, the value can be doubled in order to provide a reasonable estimate of the vertical reliability.

Kirk Burnell, NovAtel Communications Ltd.: Questions concerning GPS are generally about two key issues: accuracy and reliability.
      When a GPS position is computed, there are four unknowns being solved: latitude, longitude, height and receiver clock offset (often just called time). The solutions for each of the four unknowns are correlated to satellite positions in a complex way. Since satellites are above the antenna (none are below it) there is a geometric bias, therefore geometric biases are present in the solutions and affect the computation of height. These biases are called DOPs (Dilution Of Precision). Smaller biases are indicated by low DOP values. VDOP (Vertical DOP) pertains to height. Most of the time, VDOP is higher than HDOP (Horizontal DOP) and TDOP (Time DOP). Therefore, of the four unknowns, height is the most difficult to solve. Many GPS receivers output the SDs (Standard Deviations) of the latitude, longitude and height. Height will often have a larger value than the other two.
      Accuracy is based on statistics, reliability is measured in percent. When a receiver says that it can measure height to 1 meter, this is an accuracy. Usually this is a one sigma value (one SD). A one sigma value for height has a reliability of 68 percent. In other words, the error is less than 1 meter 68 percent of the time. For a more realistic accuracy, double the one sigma value (1 meter) and the result is 95 percent of the time). Generally, GPS heights are 1.5 times poorer than horizontal positions.

Chris Dietsch, Trimble Navigation: Based on the way the question was asked I am assuming that it is referring to the reliability of Autonomous Ellipsoid Heights generated using a C/A code solution. According to the U.S. Government Federal Radio Navigation Plan it is projected that under Selective Availability a predictable and repeatable vertical accuracy of 156 meters, at the 2 sigma confidence level, can be achieved. Don't let this statement mislead you, there are many techniques of positioning, navigating, surveying and mapping that yield centimeter level vertical results.
      The most common way of reducing the error in GPS heights is relative positioning, which requires that two receivers track four or more common satellites. If one of the receivers is set up over a point with a known or assumed ellipsoid height, the solution will then produce the change in ellipsoid height from the known point to the new point. This solution can be produced in real time or during post processing and its reliability depends on what GPS observables are used. If the pseudorange observations are used and differential corrections are applied, 2 to 5 meter reliability is expected. If the carrier phase observations are also used, the results normally yield centimeter level accuracies.
      In surveying and mapping, most applications require an orthometric height, the height above the vertical datum, for a point and not the ellipsoid height. The orthometric height is equal to the ellipsoid height minus the geoid height. So if the user wants to know how reliable GPS derived orthometric heights are, then one has to know how accurately the geoid separation can be computed in that area as these errors will combine with the errors in the ellipsoid heights.

Donald Meyer, Magellan Systems Corp.: There are several points that an average user should know when evaluating the reliability of GPS provided altitude.
      First, the user needs to make sure that the GPS unit is operating in 3D mode. This is usually indicated by an icon or other similar message on the screen. Only while in 3D mode is a unit capable of determining its altitude by using signals from the GPS satellites. Alternatively, when a GPS unit is in 2D mode it displays the same altitude value regardless of the real elevation.
      Most of the consumer receivers start in 2D mode, since this requires a fewer number of satellites, and then automatically switch to 3D when enough GPS satellites are found. After a unit has switched into 3D mode it may take several seconds before it determines the correct altitude. For users who use their GPS receivers at a constant altitude, say a lake where altitude is known, a 2D mode of operation is preferable since a receiver can use "extra" satellites to improved its horizontal position.
Most receivers have a setup function that allows users to chose between 3D and 2D only operation.
      While in 3D one should expect altitude variation of plus or minus 100-150 due to the Selective Availability error introduced by the Department of Defense. In general, the more satellites a unit tracks, the better its accuracy will be. However, without differential corrections one should not expect altitude accuracy better than 50-100 m. With real time differential corrections, altitude accuracy should be within 3-10 m depending on the quality of the differential service.

Q. Are all GPS receivers compatible with others on the market? R.R. Reston, Va.

A. Ayers: Leica receivers are compatible. All Leica GPS receivers can output raw measurements in RINEX (Receiver INdependent EXchange) format for mixed receiver data processing. Moreover, Leica GPS receivers can receive or transmit RTCM messages for real time GPS positioning.

Bohlke: Sokkia's GPS receivers are compatible with other receivers on the market in a couple of ways. Sokkia's processing software is capable of reading and writing RINEX files. A RINEX file contains GPS data in an industry standard format that can be shared and processed with data from other GPS receivers. Additionally, many of Sokkia's GPS receivers are capable of generating or reading the standard real-time differential GPS format, RTCM SC-104. Therefore, a Sokkia GPS receiver can be used with other receivers to generate or read the RTCM message.

Burnell: All GPS receivers output two solutions: position and time. The manner in which they output makes each receiver unique. Most geodetic and survey grade receivers output the position in electronic form (typically RS-232), which makes them compatible with most computers and data loggers. All NovAtel receivers have this ability. However, each manufacturer has a unique way of formatting the messages. A NovAtel receiver is not directly compatible with a Trimble or an Ashtech receiver (which are also incompatible with each other) unless everyone uses a generic data format.
      Fortunately for the user, there are several generic data formats available. For position and navigation output there is the NMEA format, differential corrections use RTCM or TRCA format, and receiver code and phase data use RINEX format. NovAtel and all other major manufacturers support these formats and can work together using them.
      The user must understand his or her post-processing and real-time software requirements. Good software will support a generic standard; bad software will lock the user into one brand of GPS equipment. For the most flexibility, insist on generic data format support for all hardware and software solutions.

Dietsch: Our GPS receivers are fully compatible with the following industry standard applications, RTCM-SC-104 (Radio Technical Commission for Maritime Services Special Committee 104), NMEA (National Marine Electronics Association) and RINEX. Our receivers generate standard output that conforms to RTCM and NMEA standards. These formats can be utilized by any receiver that accepts those standards as input. Conversely, our receivers will accept RTCM and NMEA input generated from any receiver that also conforms to those standards. Data collected for post processing by another manufacturer's receiver, that can be converted into the RINEX format, can be used in Trimble's software.

Meyer: Magellan GPS receivers are among the most compatible units on the market. The ProMARK X line of professional products comes with the MSTAR professional GPS software system, which provides a complete file conversion utility. Files can be converted to a variety of popular GIS and CAD formats. MSTAR also enables Magellan ProMARK X receivers to be fully compatible with the RINEX format, the industry standard for data exchange between GPS receivers of different brands. The RINEX standard clearly defines the file formats, data types, field sizes and other particular items related to interchange of GPS observable data.
      When Magellan-format files are converted to RINEX using MSTAR's file conversion utility, the resulting navigation and observation files are completely within the RINEX standard. There has been some disagreement in the past regarding certain items in the standard and their interpretation, and this may have led to some confusion in the industry. Two items have been most prevalently discussed: Doppler shift sign and integral-second epochs.
      First, the RINEX standard clearly calls for the Doppler shift observable to be expressed in Hertz, with the sign positive for approaching satellites and negative for receding. This is important because some software will not process data correctly if the Doppler shift sign is reversed, and some will not process it at all if the epoch time is not expressed at an integral second. MSTAR handles both cases. The software will evaluate the Doppler shift to determine if it has been formatted incorrectly, and correct it during processing if necessary. Second, nowhere does the RINEX standard call for integral-second epochs. MSTAR handles both integral-second and non-integral-second epochs.
      In both cases, Magellan's file conversion routines strictly follow the published standard. In fact, MSTAR will correctly process any manufacturer's data which is converted to RINEX, provided it complies strictly with the published standard.
      Thus, Magellan's professional product receivers, such as the ProMARK X, are fully compatible with other manufacturer's software in the context of the RINEX data format. Finally, ProMARK X data files are ASCII, so anyone with enough know how can write a conversion program for a special application.

About the participants:
Henri Ayers is a GPS specialist at Leica Canada Inc. He may be reached at 416-497-2460 (phone). John C. Bohlke serves as GPS technical product manager for Sokkia Corp. in Overland Park, Kan. He may be reached at 913-492-4900 or 800-4-SOKKIA in the U.S. (phone) or 913-492-0188 (fax). Kirk Burnell is an applications engineer in the NovAtel Communications Ltd. Sales and Customer Service Department in Calgary, Alberta, Canada. He may be reached at 403-295-4595 (phone), 403-295-4901 (fax), or e-mail: [email protected] Chris Dietsch is a product applications engineer at Trimble Navigation Ltd. in Sunnyvale, Calif. He may be reached at 408-481-8502 (phone), 408-481-6074 (fax), or e-mail: [email protected] Donald Meyer is a public relations manager for Magellan Systems Corp. in San Dimas, Calif. He may be reached at 909-394-5000 (phone), or 909-394-7050 (fax).

Back