GPS Accuracy Comparisons

Get fGIS now...click here

Comparisons of Independent, WAAS, Real-Time, and Post-Processed GPS Accuracies in Minnesota Forests

The following synopsis is derived from an unpublished draft report, used with permission of the authors.

In the autumn of 2003, Paul Bolstad, Jon Berkin, Kevin Horne, Department of Forest Resources, University of Minnesota and William H. Reading, North Central Research Station, U.S. Forest Service conducted research comparing a variety of GIS-resource grade and recreational grade GPS receivers. They looked at a range of GPS receivers and configurations for collecting data in the open and below northern forest canopies.

Bolstad and his associates compared recreational receivers in Wide Area Augmentation System (WAAS) mode, and expensive receivers optimized for spatial data collection (GIS receivers) in autonomous, WAAS, real-time differential, and post-processed differential modes. Data were collected over accurately surveyed open and sub-canopy locations. Individual position fixes were logged for varying time periods, and corrected using appropriate methods with results statistically analyzed.

Examples of the research results, comparing two of five GPS receivers used in sub-canopy forest conditions.

Their findings indicate:

  • Significant differences in the mean positional error due to receiver type under forest canopies, but no statistically significant differences under open locations.

  • Mean errors averaged between 2.9 and 7.2 feet for points in the open and between 8.2 and 23.4 feet below a forest canopy depending upon the grade of receiver (the recreational grade units showing the poorer performance).

  •  Mean errors were significantly higher for the recreational receivers than for GIS-resource grade receivers.

  • There was no difference between differentially corrected and uncorrected data when using the GIS receivers.

  • A higher number of fixes increased accuracy for two out of the eight receiver/configurations tested, but did not affect accuracy for the six others tested.

  • Recreational receiver accuracies were much less consistent, with higher frequencies of large errors. 

  • Sub-canopy tests in northern Minnesota indicate WAAS signals were only available between 8 (moving) and 23 (stationary) percent of the time for the recreational receivers, and between 22 (moving) and 33 (stationary) percent of the time when using GIS-resource grade receivers.

The tests evaluated five different Course Acquisition (C/A) code receivers including two inexpensive ($200) recreational grade GPS receivers (Garmin 76 and Deluo) and three expensive “GIS” grade systems (Trimble Geoexporer 3, Trimble GeoXT and a  Leica L5). The GIS grade units (costing 10 to 25 times the price of recreational grade receivers) are specifically designed for accurate data collection under a range of conditions.

Replicate tests were conducted for each receiver. A test consisted of standing over a known point with the receiver, beginning acquisition, obtaining position fixes until an established number was collected, noting the average position location, and if appropriate, saving the collected data to a file for further processing. Receivers were held between chest and head height, approximately four to six feet above the ground, without an external antenna.

The researchers collected a number of position fixes and averaged these to estimate coordinate location at each point. They collected 1, 10, 50, 100, 200, and 300 fixes each time a point was visited. A total of seven replicates were collected for each known point.

Accuracy differences among the more accurate receivers (Trimble, Leica) and poorer subcanopy receivers (Deluo, Garmin) are attributed to differences in their ability to reject multi-path measurements, or in the data reduction and position calculation algorithms built into each receiver. Some manufacturers invest heavily in research and development and have patented, proprietary signal processing methods. These both increase the accuracy of measurement and identify degraded or multli-path signals. Multi-path signals are received when transmissions are reflected from trees, soil, or other solid objects between the GPS satellite and the field receiver.

The researchers made a couple of ancillary observations while conducting their measurements:

  1. Raising the antenna can increase signal acquisition and reduce data collection times. Pauses during field data collection due to loss of lock on satellites lasted from a few seconds to several minutes. Raising the receiver to six to eight feet above the ground substantially shortened the duration of the pause for all receivers. Although not quantified, it was an improvement noted by all observers. Our experiences in this and other studies suggests a telescoping pole and external antenna are warranted when collecting GPS positions sub-canopy, and may provide substantial gains in efficiency, and perhaps accuracy. 

  2. Although the researchers did not test the importance of antenna orientation, this may also have an impact on data collection accuracy. Both quadrifilar (Garmin 76) and patch antennas (all others) were used. The patch type typically works best when horizontal, and the quadrifilar when oriented vertically. In past work they noted an apparent decrease in accuracy with improper orientation, but did not test its affects here. While the researchers were careful to maintain proper antenna orientation in the study, past experience showed an apparent decrease in accuracy with improper orientation.

The investigators would like to better quantify impacts of both antenna height and orientation across the range of available equipment, perhaps as a future project.

Although there are differences in accuracy between recreational and GIS grade GPS receivers, the lower accuracy of the recreational grade units may be acceptable depending upon the work being done.

If a majority of the work is in open conditions, then the relatively high accuracies and lack of difference among types suggest the recreational grade receivers are preferred. Errors of three to seven feet are much better than those obtained using pacing, chaining, or photo delineation for decades, and are generally more than adequate for most forest management activities. (Selection of receiver type and methods may also be based on other criteria, such as cost, size, ruggedness, associated software, battery life, expandability, or software capabilities.)

Selecting the appropriate receiver and methods might be more complicated when a major portion of GPS data collection will be below forest canopies. If single-fix errors are acceptable that average above 20 feet and are frequently above 60 feet when working below a forest canopy, then the less expensive recreation grade receivers may be the best choice. These errors are smaller than those obtained using previously acceptable paper and pencil methods, prior to the advent of GPS, and may meet the accuracy requirements of many resource management activities. Basic receivers such as the Garmin 76 or Deluo may also be appropriate when time allows multiple fixes at a point, e.g., when measurements at a cruise point lasts several minutes, or when cost considerations are primary.

Recreational grade receivers may also be used when digitizing linear features or area boundaries where common errors on the order of 30 to 60 feet are acceptable. This may be the case when the area to be measured is large relative to the perimeter, e.g., tens to hundreds of acres in a shape that is approximately round or square.  However, when the line, or area features need to be located to within 10 feet or less, e.g., when the areas to be measured are small, or when the areas are long and thin, then the higher end, more consistently accurate GIS receivers may be preferred.

-------------------

March 20, 2004

The preceding information is used with permission of Paul Bolstad, Jon Berkin, Kevin Horne, Department of Forest Resources, University of Minnesota and William H. Reading, North Central Research Station, U.S. Forest Service. A complete report of the GPS comparison research is expected to be published in 2004. Watch for it!

About

- Copyright © 2007-2011 ForestPal.com -
Web Statistics