Increased Accuracy For GPS Heights

While the horizontal accuracy of GPS is well known, the vertical accuracies leave something to be desired. The National Geodetic Survey (NGS) recognizes this disparity and has been performing research to see if GPS-derived vertical accuracies can be improved. Dave Zilkoski, a geodesist with NGS, has headed up two GPS projects, one in Houston and the other in San Francisco. Professional Surveyor recently spoke with Zilkoski about these projects.

A cooperative effort between NGS and the Harris-Galveston Coastal Subsidence District (HGCSD) and the Fort Bend Subsidence District, the Houston project was initiated to look at widespread subsidence caused mostly by the removal of underground fluids. This subsidence has resulted in flooding of low-lying areas, particularly during times of heavy rain or storm surges. The Houston area has a large vertical control network, but it had become very expensive to run conventional differential levels to monitor the subsidence.

Other subsidence measurements were performed using borehole extensometers, which are precise instruments mounted on pipes driven deep into the ground to stable strata. These instruments can accurately detect downward motion, but are very expensive to install and maintain. In addition, each of the six subsidence monitoring extensometers in the Houston area could measure only the subsidence in its location. The objective of the project was to develop equipment and methods that could measure subsidence over large areas, accurate to within one centimeter.

Lower Cost and Portability

To lower costs and provide portability, HGCSD and NGS developed the PAM, or Port-A-Measure, a trailer-mounted device containing a GPS receiver and a cellular phone. Powered by solar panels, the trailer can be towed to a site and operated unattended, automatically transmitting GPS observations. Three of the six existing extensometers were fitted with GPS receivers to act as Continuously Operating Reference Stations (CORS). The information gathered at the trailers is post-processed to provide accurate heights, and from the height changes, subsidence. The PAMs have been in place since 1994 and an enormous amount of 24-hour datasets have been collected. In some areas, subsidence is approaching 10 centimeters since 1994.

Zilkoski said that during the course of the project, they found that GPS measurements of short duration are affected by the time of day the observations are made. He stressed that repeatability and accuracy were goals. The soution was to make observations at different times of day, thereby taking advantage of different satellite geometry. NGS experimented with extracting one-hour, two-hour, and three-hour observations. Based on these experiments, it was determined that the most economical approach was to make short observations twice a day on two separate days, thereby ensuring that different satellite geometry existed.

San Francisco Project

The knowledge gained in Houston has been moved to San Francisco Bay, where a cooperative effort between the National Ocean Service—of which NGS is a part—several California state and local agencies, and private maritime interests, is seeking to establish a set of products and services to meet local navigation and coastal management needs. The difference between the Houston and San Francisco projects is that one was designed specifically for the detection of subsidence and the other is for the establishment of an accurate control network. In both cases, the knowledge and techniques can be applied elsewhere, wherever there is a need for accurate and precise GPS heights.

The San Francisco Bay Demonstration Project has been a collaborative partnership between the Bay Conservation and Development Commission, the Port of Oakland and other Bay ports, the Harbor Safety Committee, U.S. Coast Guard, and other governmental and non-governmental organizations such as the California Coastal Conservancy. In addition to safer, more efficient navigation, key local issues included improved dredging operations, establishment of the coastal zone boundary through the determination of a new mean-high-water line, sustainable wetlands management, enhanced spill preparedness and contaminants monitoring. The National Ocean Service is using the project to demonstrate the integration of its maritime commerce and coastal management missions, and to match federal capabilities with local initiatives, authorities, and needs.

Docking Ships in Zero Visibility

Horizontal positioning for navigation, through the use of CORS, is well-established, but much remains to be accomplished to achieve equivalent vertical accuracies. Zilkoski described a need to be able to determine, in real-time, exactly where a ship is in all three dimensions. If this can be done, ships will be able to dock in zero visibility. Because the water level rises and falls with the tides, and because many ships have curved sides, if the captain knows exactly where the dock is and exactly where his ship is, the water level will not matter. GPS will provide latitudes, longitudes and ellipsoidal heights for dock facilities, the bottom of the bay and, from the tide gauges, the water level. Real-time DGPS and telemetry of tidal information will allow ship pilots to interface with digital nautical charts and know exactly where they are with respect to the dock and the bottom of the bay.

Zilkoski said California presents special problems due to crustal movement. Parts of California may have moved over a meter due to crustal movement and subsidence, and the Central California vertical control network has not been re-leveled since 1960. Zilkoski said that previous work with state and county governments had revealed that cities and counties normally need vertical accuracies of two centimeters, while five-centimeter accuracy will suffice for rural areas. Keeping that in mind, an approach was formulated that would provide answers for both needs. The GPS observations for each of the project's tasks were referenced to five local CORS.

All Measurements Referenced to CORS

The first task was to make static dual-frequency GPS observations to determine ellipsoidal heights for three long-term tidal gauges. The next task was to establish the relationship between these tidal benchmarks and the ellipsoidal heights of local sea level, more specifically, mean low water and mean lower low water. The third task involved static GPS observations to determine ellipsoid heights, to two-centimeter accuracy, on 16 NAVD 88 benchmarks in the project area and then to estimate GPS-derived orthometric heights for each benchmark. Using pseudo-kinematic techniques, the next task was to establish 3D positions, to five-centimeter accuracy, on selected piers, shoreline features and other photo-identifiable points. Zilkoski is now analyzing the data from all angles. From this analysis, NGS will issue guidelines for establishing GPS-derived ellipsoid heights to two- and five-centimeter accuracies.

Remaining tasks include the use of tide-coordinated, GPS-controlled photogrammetry to estimate the ellipsoid heights and horizontal positions of the shoreline and features that were positioned during the pseudo-kinematic survey; using GPS-controlled LIDAR techniques to estimate ellipsoid heights of sea surface, bathymetry and shoreline; and finally, to integrate the LIDAR, photogrammetry and geodetic survey results for use in the creation of electronic nautical charts and GIS maps. Zilkoski emphasized that NGS is not building a GIS, but rather gathering data that will be usable in any GIS.

Technique a Welcome Addition

As a result of these two projects, NGS has demonstrated that GPS can be effectively used to provide high-accuracy vertical control. To ensure plenty of data for the San Francisco project, NGS made static observations on the benchmarks of three hours in the morning and three hours in the afternoon on two different days. Through the analysis of the data, Zilkoski showed that 30-minute observations would have provided sufficient data and, under some conditions, even 10-minute observations might be appropriate for station spacing of five-to-seven kilometers. He stressed the importance of obtaining different satellite geometry and minimizing multipath. Zilkoski feels that the technique will replace traditional vertical control networks, but because of the observation requirements, it will still be cheaper to run differential levels over distances of less than five kilometers. The fact that short observations can be made, even if they need to be made on four separate occasions, makes this capability a welcome addition to the establishment of accurate and precise GPS heights.

Marc Cheves is Editor of the magazine.

» Back to our May/June 1996 Issue

Website design and hosting provided by 270net Technologies in Frederick, Maryland.