SYNOPSIS OF PROCEEDINGS
Although GPS has been used to measure geophysical phenomena, such as the motion of tectonic plates, for more than a decade, static positioning at the millimeter level is a recent achievement. This level of accuracy has made GPS useful for static positioning applications, such as the study of crustal deformations related to earthquake and volcanic processes. However, for this application and others in the geosciences, sources of positioning error remain. These sources of error include unwanted motions of GPS monuments; unmodeled tropospheric signal delays; interactions between GPS antennas and the surrounding environment; variations in GPS antenna phase patterns; and antenna radome effects.
Roger Bilham discussed the stability of geodetic monuments and control points used in conjunction with the placement of GPS antennas.. He pointed out that although tectonic deformations are driven by stresses deep within the Earth, they are measured by GPS networks on the Earth 's surface. Therefore, surface noise caused by variations in groundwater hydrology, atmospheric pressure and temperature, and solar radiation can affect the accurate measurement of the tectonic phenomena. Surface instabilities and periodic variations in the upper layers of soils and bedrock caused by thermal stresses, the effects of fissures, and rainfall (soil expansion and possible stress on nearby bedrock) can also diminish the accurate measurement of deep tectonic signals.
These errors manifest themselves as horizontal and vertical displacements of GPS monuments. Monument instability and the resulting positioning errors can be minimized through proper design, deep installation (30 meters) in material such as bedrock, and the use of inclinometers and vertical extensometers to measure and quantify instabilities. Even antennas that must be mounted on towers and buildings can provide millimeter-level positioning accuracy if their displacement relative to a properly installed control point is accurately monitored.
Highly accurate GPS observations can be affected by differences in antenna types and mounts. In principle, however, these effects should be measurable in anechoic chambers and in the field. Test measurements were discussed in a paper prepared by a number of co-authors and presented by Chuck Meertens. The results showed that errors resulting from some combinations of antennas from different manufacturers could be measured and resolved to an accuracy of better than 12 millimeters in the vertical plane. Other combinations produced errors as high as 5 centimeters. These differences could be minimized, according to the authors, if manufacturers would move toward standard antennas for high accuracy applications. The tests also showed that antenna radomes could introduce errors as large as 10 millimeters in the vertical plane. The authors concluded that more work is needed to reduce antenna mixing and radome errors to the 1 millimeter level.
Even if all known errors attributable to the characteristics of GPS antennas could be measured and modeled out of a position solution, errors caused by interactions between antennas and the environment would remain. This class of errors can be divided into near-field effects (within a few meters of the antenna) and far-field effects or multipath (more than a few meters from the antenna). Unlike multipath, near-field effects, as Arthur Niell pointed out in his presentation, cannot be reduced by averaging observed data over time. Near-field effects caused by signal scattering from site monuments or snow are most prevalent when GPS observations are made at very low elevation angles. The resulting errors can be as large as several centimeters. Based on testing described in the paper, the author concluded that elevation-dependent near-field effects can be reduced by carefully evaluating antennas and monuments before they are installed in the field.
Like near-field scattering effects, multipath effects can be a significant source of error when low elevation GPS observations are made. Low elevation observations are important for estimating atmospheric signal delays, another source of error in static positioning applications, but as Chuck Meertens pointed out in his presentation, multipath errors are often mistakenly identified as atmospheric delays at low elevation angles. Multipath errors can be reduced by using antennas fitted with choke rings, which are concentric corrugations that reduce antenna sensitivity to ground-reflected multipath effects. Recent design enhancements to choke rings that have improved GPS positioning accuracy were discussed in a paper presented by Fred Solheim and Chris Rocken.
Requirements for data and positioning accuracy for the study of earthquake and volcanic processes were discussed in two papers, presented by Mike Watkins and Mark Murray respectively. After discussing specific monitoring activities in northern California, Mount St. Helens, Oregon, and the Kilauea volcano in Hawaii, both speakers reported the need for 1 to 3 millimeter horizontal accuracy and 5 to 10 millimeter vertical accuracy. They also noted that more accurate measurements would allow additional geophysical processes to be studied, such as nonsecular motions. This level of accuracy can only be achieved by minimizing the sources of error discussed above.
The monitoring of earthquake and volcanic processes requires a 30-second temporal sampling rate for reliable, autonomous “cleaning” of data (i.e., detection and removal of cycle slips and bad data). A decimated sampling rate can be used for determining position because of the temporal correlations in GPS phase measurements. The analysis centers of most GPS networks decimate data to 2 to 10 minutes. Data latency of 1 to 3 days is adequate, but there are times when near real-time dissemination would be useful (immediately following a large earthquake, for example). For studying postseismic deformations, sampling rates on the order of 1 second would be desirable, even if the resulting positioning accuracy is less precise.
WORKING GROUP DISCUSSIONS
Many of the issues discussed by the working group on networks, data sources, and static positioning applications were described in Chapter 2. However, some of the discussions focused on subjects covered by the presentations described above and were considered particularly important by geophysical researchers.
Data Sampling Rates and Latency
Although the group agreed that a 30-second receiver sampling rate should be the standard default setting for all GPS networks and reference stations, some applications may require that higher sampling rates be available on request. The group noted that the need for higher sampling rates may grow as ionospheric effects increase with the approach of the solar maximum. The group also agreed that real-time data are not required for most static positioning applications. However, postseismic studies related to earthquakes and volcanic eruptions could benefit from near real-time data. Data derived from networks like the U.S. Coast Guard's are already packaged in hourly files that are available a few minutes past the hour. However, rapid transfers of data with one hour latency may be too expensive for some sites. In addition, the need for data centers that compile older, previously received data sets that can be integrated with current data will continue.
Classification and Mitigation of Monument and Antenna Related Errors
The working group discussed the need to mitigate errors caused by monument instability, keeping in mind that some researchers actually gain useful information about surface conditions by observing antenna displacements. Even in these cases, however, attempts should be made to measure and characterize the stability of monument and antenna placements at as many GPS tracking sites as possible. Information for each site could be provided to users in the comprehensive network catalogue described in Chapter 2. To ensure that this information is accurate and consistent, standard methods of measuring monument stability will have to be developed. Otherwise, researchers will have difficulty separating the effects of monument instability from the effects of other error sources, such as multipath, radio-frequency interference, ionospheric and troposphere delays, and antenna phase center variations.
In order to satisfy the accuracy requirements of many static positioning applications, the group agreed that GPS observations should probably be collected by all sites down to at least 5 degrees elevation angle unless physical limitations, such as natural or man-made obstructions, make this impractical. Low elevation angle data are critical to measuring the signal delay and subsequent positioning errors caused by tropospheric water vapor. However, because errors related to interactions between GPS antennas and the surrounding environment are more prevalent at low elevation angles, characterizing reference sites for both near-field scattering and far-field multipath effects is critical. Information used to evaluate these effects, such as measurements of signal-to-noise ratio, could be made available to users. In addition, geological and geomorphological descriptions of sites and antenna configurations, including photographs and horizon mask information could also be made available. The need for continued testing of antenna phase center variations, especially for networks with mixed antennas was also discussed, as was the need for a better understanding of the effects of radomes on positioning accuracy.