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

Q. I have a GPS plus computer to record ata at a dixed location, and another GPS plus laptop in the field also recording data as I travel. Can I take the difference between the recorded location and the known location of the fixed point and use that as an error correction for the recorded data in the field? Would this reduce the problem of SA and get more accurate data? -D.R. Denver, Colo.

A. Charles Branch, Ashtech Inc.: Simple answer: Maybe.
      More complete answer: This wont yield the accuracies of more sophisticated methods of differential correction. Also, if using this method, you really should calculate both base and rover positions using the same GPS satellites at the same time. This is hard to do. If you dont do this, your corrections will be unpredictable and could fail to improve accuracy at all. You could get 5 meters of accuracy or you could get 100 meters. You may be more satisfied buying a software solution from a GPS manufacturer. Its up to you.
      What you are describing is called solution space differential correction. It is conceptually simple and so was among the first methods of differential correction offered by GPS manufacturers. You measure the difference between the base receivers known location and its GPS-derived location, record it as a vector (range and azimuth), time-tag it and subtract it from the position derived by the rover receiver at the same time. The problems are that under SA, GPS errors average some 40 meters and can change as rapidly as a few meters per second. This and the same satellite, same time issue make solution space differential correction a comparatively poor performer.
      What many GPS manufacturers recognized a few years ago was that they could yield much better accuracy for about the same amount of trouble using a different technique called measurement space differential correction. This is what you are getting when you buy a GPS mapping system from most GPS companies today. You should be able to get submeter accuracy quite reliably.
      However, if you want to try for 5 or so meters of accuracy by writing your own program and locking in both the base and rover receivers to the same satellites, go right ahead. One helpful hint: think 3D when calculating errors at the base and when removing them from the rover. Even if you are only interested in lat/long, taking elevation into account will improve your accuracy.
      Tom Damiani, Rockwell Semiconductor Systems: GPS position errors are caused by errors in measured ranges from the satellites to the users antenna. These range errors are caused by natural effects (such as the ionospheric delays for that particular satellite) or SA.
      In order for your proposed correction scheme to give the best results, it is necessary that both the fixed and mobile receiver be tracking the same set of satellites and use the same set of satellite ephemeris parameters.
      A good reference for various DGPS techniques and their advantages and disadvantages is contained in the following journal paper: Blackwell, E.G., Overview of Differential GPS Methods, NAVIGATION, Journal of The Institute of Navigation, Vol. 32, No. 2, 1985, pp. 114-125.
      Chuck Gilbert, Trimble Navigation: It is my assumption that you are trying to set up a base station and roving data collector using the difference of the two to eliminate Selective Availability (SA). While the concept is sound, there are many other factors which could cause this manual processing to fail much of the time. Unfortunately, you would never really be able to tell when this correction by hand process was a success and when it merely corrupted your data.
      In regard to SA there are many factors to consider. First, the positions being collected at the known location and the field locations must be collected simultaneously. One cannot collect one position in the morning, calculate the difference, then use that as the correction for all positions collected throughout the day. It is vital to consider the collection time of each position when applying corrections from one receiver to another.
      Also, there is a satellite issue. A major flaw in this process is that it doesnt seem to be able to account for constellation differences between the two receivers. Each one of the satellites has its own unique set of errors that are constantly changing. Therefore, both receivers must be obtaining their positions using the same constellation of satellites in order for them to have common errors. Any kind of temporary obstruction in the field, such as a passing tree may make some satellites temporarily unavailable. This obstruction would likely cause the two receivers to be computing the positions from a different set of satellites with a different set of errors.
      Naush Ladha, NovAtel GPS: The simple answer to that question is yes. You can take the difference between recorded position and known location and apply this, as a position correction, to your field data (mobile station). However, the correct and more standard way of computing this correction is to compute the range error to each GPS satellite being tracked at your fixed location and to apply these range corrections to the observations at your mobile station.
      The position corrections method is seldom used in industry. The drawback of this method is that computed corrections will vary depending on the location of the fixed station. The geometry between the fixed station and the tracked satellites is not accounted for. Also, position corrections at the fixed site are computed with a certain group of satellites while the mobile station is tracking a different group of satellites. In general, when the position correction method is used, the farther the fixed and mobile stations are apart, the lower the accuracy of the mobile stations position.
      The range corrections method is more commonly used in industry. The advantage of using this method is that it will provide consistent range corrections and hence field positions regardless of the location of your fixed station. You will only be able to obtain a good differential position if both the fixed and mobile stations are tracking at minimum the same four satellites.
      Both the position correction and range correction methods are valid and would significantly reduce the effects of SA. A user could choose either method depending on the application and the accuracy required.

Q. What is the real accuracy of the Coast Guard's DGPS as compared to the commercial DGPS? The Coast Guard claims a 10-meter accuracy for their DGPS. Some commercial DGPS vendors seem to offer 5-meter (or better) accuracy. Are the commercial vendors really supplying something more accurate than the Coast Guard signal? -C.M. Klamath Falls, Ore.

A. Damiani: The accuracy of all DGPS services depends on the accuracy of the corrections determined by the reference station receiver, the time delays involved in calculating and transmitting these corrections, and the error rates of the communication links. Because these parameters vary from service to service, it is not unexpected that the accuracies achieved by various services differ.
      The accuracy of the Coast Guard DGPS service can be determined by contacting the U.S. Coast Guard Navigation Information Service at 703-313-5900. They can also provide information regarding information services available electronically.
      Some commercial DGPS services are offered at accuracies below those advertised by the Coast Guard, but of course they charge a fee for their services. In general, they charge higher fees for more accurate services.
      Dr. Frank van Diggelen, Ashtech Inc.: The Coast Guards DGPS gives the best possible combination of large coverage with high accuracy. One meter accuracy is typical. It is technically possible for commercial services to offer higher accuracy but only over a small area.
      To answer this question fully we must consider the quality of the data, the area of coverage and reliability of the service. The U.S. and Canadian Coast Guards both chose their receivers after competitive bid processes that included all the major GPS manufacturers (and they didnt choose the cheapest either!), so the quality of the data is as good as anything produced by commercial DGPS stations.
      Apart from the initial data quality, your accuracy will depend on the latency of the corrections, your distance from the reference station, and the reliability of the service.
      The latency of the corrections is the amount of time it takes for the corrections to get from the reference station to you. As the corrections get old their accuracy degrades. This introduces a trade-off: high data rates result in less coverage but more accuracy; lower data rates result in more coverage but less accuracy. The U.S. and Canadian Coast Guards both use low-medium frequency transmitters that give a large coverage radius of up to 300 miles, and data latency of 1 to 3 seconds. The achieved accuracy is much better than the 10m specification, as shown by the tests reported below. A commercial service with higher data rates may have higher accuracy, but only in a small area, and the format may not be the RTCM standard used by the Coast Guards and compatible with all DGPS receivers.
      The distance from the reference station affects accuracy at about one part per million, that is, an extra decimeter of error for every 100 km distance. The U.S. and Canadian Coast Guards have the most reference stations of any service. The U.S. Coast Guard reference stations cover the entire coastline of the conterminous U.S., Hawaii, Puerto Rico, Southern Alaska, the Great Lakes, the Mississippi and Missouri River valleys. The Canadian Coast Guard stations cover the Canadian side of the Great Lakes, the Saint Lawrence Seaway, and much of the Canadian coastline. You can find out if one of these stations is near you by visiting the Web site: www.navcen.uscg.mil/dgps/dgps.htm
      Reliability is important if you want an assurance that the accuracy you achieve today will still be there tomorrow. Both the U.S. and Canadian Coast Guards employ doubly redundant systems with automatic integrity monitors, thus each Coast Guard station consists of two DGPS reference stations, plus integrity monitor. If a problem occurs with one reference station, it is taken off-line and immediately replaced with the other, and a technician is sent out to investigate the problem. It is unlikely that commercial services have the resources to provide the same level of service.
      Ladha: The real accuracy of the Coast Guards DGPS signal is more than likely better than 10 meters. However, there are a number of factors which are involved in determining the accuracy of a DGPS. These include:
      A users proximity to the base station which is transmitting DGPS corrections; the GPS receiver used by the Coast Guard; the GPS receiver used by the commercial DGPS services; the GPS receiver used by the user; and the statistical qualifier used in conjunction with the stated accuracy.
      If you were to compare the Coast Guard and commercial DGPS services under the same situations, e.g., base-user proximity of 1.0 kilometer and stated accuracy at 2drms (95 percent confidence), you would probably find that the Coast Guards DGPS is at least equivalent to, if not better than, commercial DGPS services.
      Also of note is that the Coast Guards DGPS service is available to all users (marine, land and air), similar to a public utility without any charge. In addition, the Coast Guards service acts as an integrity monitor, which provides an independent check of each GPS satellites signal and reports whether it is good or bad. Commercial DGPS vendors usually have a monthly or yearly subscription fee without any associated liability.
      All of the previous discussions have been dealing with code data. Some commercial DGPS services are now also starting to provide high accuracy carrier-phase data along with code data. With this type of data, depending on your equipment, you will be able to achieve decimeter and even centimeter level accuracies.

About the participants:
Charles Branch is a GIS marketing manager at Ashtech Inc. in Sunnyvale, Calif. He may be reached at 408-524-1603 (phone), 408-524-1500 (fax), or e-mail: [email protected] Tom Damiani is manager, key programs marketing in the Wireless Communications Division at Rockwell Semiconductor Systems in Newport Beach, Calif. He may be reached at 800-854-8099 (toll free phone) or e-mail: [email protected] Dr. Frank van Diggelen is a marketing manager, OEM and Navigation at Ashtech Inc. in Sunnyvale, Calif. He may be reached at 408-524-1508 (phone), 408-524-1500 (fax), or e-mail: [email protected] Chuck Gilbert is an applications engineer at Trimble Navigation in Sunnyvale, Calif. He may be reached at 408-481-8000 (phone), 408-481-6074 (fax), or e-mail: [email protected] Naush Ladha is an applications engineer in NovAtel GPSs Sales and Marketing Division, headquartered in Calgary, Alberta, Canada. Ladha may be reached at 403-295-4564 (phone), 403-295-4901 (fax), or e-mail: [email protected]

Back