Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.
OCR for page 22
INTERNATIONAL GLOBAL NETWORK OF FIDUCIAL STATIONS: SCIENTIFIC AND IMPLEMENTATION ISSUES 2 SCIENTIFIC RATIONALE FOR GLOBAL NETWORKS Introduction Geodesy as a scientific discipline is in the enviable position of marrying fast-developing technology with rapidly expanding scientific goals. In a sense it is valid to ask whether technology development drives the science to the same extent as scientific requirements drive the technological innovations. As high-precision space-geodetic techniques become more easily accessible to the scientific community, the range of problems that geodesists can attack in earnest expands to incorporate problems previously deemed too formidable to be treated successfully. Examples abound as described in detail in several recent reviews, such as the Erice and Coolfont workshop reports mentioned earlier, and several recent reports by the National Research Council 's Committee on Geodesy (Geodesy: A Look to the Future, 1985; Current Problems in Geodesy, 1987a; Geodesy in the Year 2000, 1990a). These reports cover the opportunities offered by space-geodetic techniques and make specific recommendations concerning the areas in which to focus attention now and in the next decade or two. In this chapter we draw from these studies but adopt a somewhat more specialized point of view. Specifically, we endeavor to identify scientific problems that cannot be addressed properly in the absence of a global network, as well as applications of space geodesy that would benefit directly and substantially from such a network without requiring its existence. In addition, we recognize that precise global geodesy by itself is not a panacea that will produce the needed solutions, and we attempt to consider its contribution in the context of the broadly based, multidisciplinary environment advocated in Mission to Planet Earth (National Research Council, 1988). Although space geodesy is a relatively young discipline, it tends by nature to adopt a global perspective. For instance, SLR and VLBI networks (including the QUASAR network) have a global scope, although they do not provide uniform
OCR for page 23
INTERNATIONAL GLOBAL NETWORK OF FIDUCIAL STATIONS: SCIENTIFIC AND IMPLEMENTATION ISSUES coverage. More recent deployments, such as the DORIS and PRARE networks, are both global and quite uniform by design. The rapidly growing CIGNET GPS network exhibits similar characteristics, especially when considered in conjunction with other stations such as the tracking sites currently being deployed by NASA. Temporary global GPS deployments have been conducted successfully (e.g., GOTEX), and a global international campaign (GIG'91) was conducted in early 1991 under the aegis of IERS. Although our focus is on geodetic networks, other disciplines dealing with solid Earth science are actively building global networks of their own. These include, for instance, seismic networks (e.g., GEOSCOPE, IRIS GSN) and magnetic observatories (INTERMAGNET). Largely in recognition of the value of multidisciplinary measurements, the concept of the Permanent Large Array of Terrestrial Observatories (PLATO) was originally proposed as one of the two major elements of the Mission to Planet Earth. Scientific Goals It is self-evident that the major purpose of global geophysical and geodetic networks is to permit the collection of observations capable of constraining models of the planet as a whole. The qualifier global takes on a dual connotation in this context. On the one hand, such data are needed to study phenomena that operate on a global scale—instead of continental, ocean basin, or any of a hierarchy of scales down to the dimensions of grains that form the rocks. On the other hand, even when the scales under consideration are much smaller than the radius of the planet, so that a regional or even local description may be adequate, global coverage is still desirable to help understand the underlying physical processes that shape the planet. Long-wavelength features are then used as a background to analyze short-wavelength components. The scientific goals of global networks should reflect this duality (see also International Association of Geodesy, 1991): The dual scientific goals of global geophysical and geodetic networks are (1) to improve understanding of geophysical and geological phenomena that operate on global scales and (2) to provide a framework and boundary conditions to analyze phenomena that operate on smaller scales. A simple illustration is the spherical harmonic expansions commonly used to describe global characteristics of the Earth. The geopotential and geomagnetic fields are classical examples, but this type of representation is also used for the
OCR for page 24
INTERNATIONAL GLOBAL NETWORK OF FIDUCIAL STATIONS: SCIENTIFIC AND IMPLEMENTATION ISSUES three-dimensional internal structure of the Earth determined by seismic tomography, in models of convection currents in the mantle, or even to describe the kinematic plate velocity field at the surface. In some instances, global coverage—at least at sufficiently long wavelengths—is provided by observations from satellites in low Earth orbit. This is true of potential fields. However, for small-scale features, and for all structural, kinematical, and dynamical models of the Earth that depend on surface observations, the resolution is limited by the coverage provided by surface observatories. For instance, the determination of radial and lateral inhomogeneities in seismic velocities requires a global distribution of both seismic stations and seismic sources. Because Earth is a dynamic planet, and because its surface is covered by highly mobile tectonic plates, even measuring the rotation parameters of the planet—a global characteristic par excellence—requires adequate global coverage by a network of space-geodetic observatories. For that matter, even defining an International Terrestrial Reference Frame, on a planet where every piece of real estate is in constant motion with respect to every other piece, is a challenge (e.g., Boucher and Altamimi, 1989), and current realizations depend on data collected at a set of globally distributed sites. With geodetic precise positioning networks, these problems are sharpened by the lack of a natural and unambiguous way to interpolate observations. We do not enjoy the benefit of having potential fields or elastic waves to define a physical averaging of pointwise properties (such as density, velocity, and Poisson's ratio) of the Earth and, therefore, must resort to using a particular mechanical and rheological model of the Earth between sites to interpolate and interpret the data. This technique is often quite successful when the intersite distances are smaller than or comparable to scales of relevant inhomogeneities. However, for spatial scales that can be realistically sampled by global networks, this approach is fraught with danger, because large-scale crustal inhomogeneities are likely to invalidate smooth strain models between sites. Geodesists therefore have deployed a hierarchy of networks in which sparser regional networks serve as fiducial sites in the survey of denser local networks. With the advent of space geodesy, we can conceive of and actually implement a global high-precision network of sites that will serve as fiducial points tied to a common reference frame with uniformly high precision. The deployment of such a network will substantially influence the various uses of space-geodetic techniques. The first steps have already been taken toward a goal that is specific to geodetic global networks, namely (see also International Association of Geodesy, 1991):
OCR for page 25
INTERNATIONAL GLOBAL NETWORK OF FIDUCIAL STATIONS: SCIENTIFIC AND IMPLEMENTATION ISSUES Geodetic measurements with a precision of approximately 3 mm or better, in both horizontal and vertical components, should be made possible at any time, anywhere on the planet, on spatial scales ranging from a few kilometers to intercontinental distances, with an achievable temporal resolution of a day or better. Millimeter-precision geodesy is now possible even on very long baselines using VLBI (e.g., Herring, 1991). In other words, the technological know-how exists to achieve this level of precision between properly upgraded VLBI sites. The problem raised by this goal is therefore not so much to achieve very precise measurements, but rather to make them affordable and logistically feasible at unprecedented temporal and spatial densities. The equivalent strains detectable with such a deployment would be about 10-6 on local scales and better than 10-9 on global scales, with a temporal resolution of 1 day or better. On local scales such a capability obviously would neither duplicate nor supplant existing techniques such as observatory strainmeters capable of detecting strains of a few parts in 1010 over times shorter than a day (Agnew, 1987). Instead, the capability should be viewed as complementary, designed to investigate spatial patterns of rather large strains and their variability with space and time, an aim not achieved by most existing geodetic monitoring networks primarily because of their low time sampling rates. Although our goal introduces 3 mm as a target precision, it must be recognized that in many areas of the world significant scientific progress would result even from measurements made at a precision of 10 mm. This is true, for example, of areas where crustal motions are poorly known and where any geodetic constraint would help. As an illustration, crustal deformation in the India-Eurasia collision zone, which includes the large Tibetan Plateau, is a matter of considerable current debate among geologists and tectonophysicists (England and McKenzie, 1982; England and Houseman, 1989; Molnar and Tapponnier, 1975; Tapponnier et al., 1982; Peltzer and Tapponnier, 1988). In that case, reconnaissance geodetic surveys would be most valuable to help select among the various deformation models proposed in the literature. In other instances a precision of 1 mm for daily measurements would be very desirable. This is true, for example, of areas already well instrumented, where important geological problems are likely to require analysis and interpretation of rather subtle geodetic signals. Examples include the detection and analysis of postseismic, interseismic, and possible preseismic strains within plate boundary deformation zones such as the western United States or intracontinental seismic zones such as the New Madrid area in the central United
OCR for page 26
INTERNATIONAL GLOBAL NETWORK OF FIDUCIAL STATIONS: SCIENTIFIC AND IMPLEMENTATION ISSUES States. Averaging observations over longer times will of course improve the detection of small signals with long time constants, but it must be borne in mind that systematic effects do not cancel in such a procedure and that the rule of thumb is that a signal-to-noise ratio improvement of a factor of three constitutes a practical limit. Furthermore, especially at the millimeter level, monumentation issues are important (e.g., Wyatt, 1982, 1989), and geological interpretation is chancier. The goal stated above could be reached through a coordinated international effort by the middle of the decade and should be largely surpassed —in precision as well as spatial and temporal resolution—by the year 2000. As stated, it is compatible with several recommendations cited in Appendix A and incorporates some of the notions expressed in them. Although global networks are not discussed in detail in the Erice report, which focuses on geodesy, they are a centerpiece of the Coolfont report, which is permeated by the notion of global studies of the Earth. This includes not only global altimetric, geodetic, geological (e.g., soil and surface processes), and geophysical coverage (e.g., potential fields) but also studies of large-scale structures of the interior of the Earth, for which a multidisciplinary approach is essential. As an example, a recommendation of the Coolfont Plate Motion and Deformation Panel states: In order to study plate motion we need a global distribution of geodetic stations that measure relative positions to 1 cm over one day and to 1 mm over three months. In order to monitor regional and local deformation we need a terrestrial reference frame. Both of these objectives can be accomplished with a global distribution of space-geodetic observatories. Although this recommendation is couched in terms of a geodetic network, the concept that has been derived from it involves a set of globally distributed sites with a multidisciplinary vocation, described collectively as FLINN. As stated in the Coolfont recommendation, FLINN sites should also be tied to a geocentric reference frame with a precision of 1 mm averaged over 1 year and should serve as fiducial sites for dense regional networks (NASA, 1991).
OCR for page 27
INTERNATIONAL GLOBAL NETWORK OF FIDUCIAL STATIONS: SCIENTIFIC AND IMPLEMENTATION ISSUES Objectives The primary scientific objective of a Global Network of Fiducial Sites is to provide an effective and economical means of acquiring, analyzing, and archiving the data required to solve global problems. The primary objectives are identified directly with research topics that cannot be pursued in any other way. In this category we place the study of complex multiscale problems such as global sealevel monitoring and postglacial rebound, for which global coverage is essential. A second category involves the precise measurement of tectonic motions, including plate deformation. The most interesting signals are departures from the motions predicted by the long-term geological models (e.g., Jordan and Minster, 1988a). Further, many signals of practical interest that take place on local and regional scales can only be interpreted properly if the large-scale motions are well understood. Such problems place very stringent demands on network operations and often require improved solutions to other problems with a global character, such as the determination of an improved geoid or the measurement of densely sampled, precise time series of Earth orientation and rotation parameters. Other objectives, no less important, are more easily defined in terms of support of various scientific endeavors: examples include the calculation of precise orbits for a variety of Earth-orbiting spacecraft and the realization of a precise global reference frame. Applications that do not intrinsically require a global network but that would benefit directly from it cover a wide range of geological problems, such as relative motions across specific plate boundaries and plate boundary deformation zones, earthquake and volcanic cycles, and altimetric surveys. Finally, with the advent of inexpensive, and lightweight space-geodetic techniques such as GPS, a suitably dense, carefully maintained global set of fiducial points would have a significant impact on almost all aspects of geodesy, including practical surveying considerations. For our present purposes, geological and geophysical objectives may be discussed from the point of view that they identify certain scientific users of global geodetic data, the scientific customers of the global network. Geological and geophysical objectives are primarily applications of the capability obtained through the global network. But most of these applications require some degree of densification of the network to achieve the spatial and temporal resolutions needed to make substantial contributions. Two major global problems are used as examples to support and focus our discussion. These are:
OCR for page 28
INTERNATIONAL GLOBAL NETWORK OF FIDUCIAL STATIONS: SCIENTIFIC AND IMPLEMENTATION ISSUES global sea-level change, response to surface loads, and postglacial rebound and tectonic motions and intraplate deformation. The rates of horizontal motions from these phenomena are comparable. The main distinction is that the regional vertical motions associated with the first are typically much greater than for the second. As a result, although each of the problems of eustatic (global) sea-level change, postglacial rebound, and intraplate deformation has its own intellectual focus, these problems are very tightly intertwined from an observational standpoint. Not one of them can be solved independently of the others. For example, present sea-level change cannot be inferred correctly from tide gauge records without first adjusting observations of relative sea level for postglacial rebound. Similarly, horizontal motions associated with postglacial rebound or surface loading caused by sea-level changes are comparable to those associated with departures from the predictions of rigid plate theory. The challenge is to separate the deformations associated with tectonics from those associated with surface loads. In order to meet this challenge, we must first meet specific geodetic objectives, which must subsume those listed by Knickmeyer (1990). These include: determination of precise orbits for spacecraft used in geodesy and geophysics, precise determination of Earth rotation and orientation parameters, realization of a precise global terrestrial reference frame, support of scientific orbital missions, and support of local and regional studies. The first four of these objectives might be identified with providers of information and data made available via the global network. The fifth objective pertains to end users of the global network. Because of the large community of professional users of the data products (such as GPS ephemerides) provided by the global network operation, the concept of a service is an important component of international organizations such as IERS or the proposed IGS. However, this concept affects primarily the issues of data processing, management, and distribution; it does not introduce new constraints on network design as high-quality orbits (covered by the first objective) would satisfy the vast majority of the needs of the professional community.
OCR for page 29
INTERNATIONAL GLOBAL NETWORK OF FIDUCIAL STATIONS: SCIENTIFIC AND IMPLEMENTATION ISSUES The global solid Earth scientific problems that can be addressed, at least in part, through a global geodetic network share one feature: they are characterized by a spectrum of spatial and temporal scales. As a result, one is strongly tempted to seek ever-improved resolution by densifying the network everywhere. One possible approach is to construct a large global network as an assemblage of properly connected regional networks, the operation of which must be carefully orchestrated and coordinated. This approach may well be viable in the long run, but it seems preferable first to discuss general global network design without introducing the complications of a practical implementation strategy. In the following discussion, several numbers will prove useful: The area of the Earth's surface is 5.1 × 108 km2, with a square root of 2.3 × 104 km. This means that a 25-station global network uniformly distributed on the surface of the Earth will have an average intersite spacing of ~4,400 km. This is a crude, but workable characterization of a GPS core network. A 100-station network would have an average intersite spacing of ~2,200 km, and more than 400 stations would be needed to achieve an average spacing approaching 1,000 km, assuming again a globally uniform distribution. Of course, such a geodetic network cannot be easily realized because of the presence of large oceanic areas without emerged lands. Geophysical and Geological Objectives Eustatic Sea-Level Change. Global sea-level change is one of the most easily imagined end results in the chain of effects hypothesized to come from man's interference in the global climate system (see, e.g., Sea-Level Change, National Research Council, 1990b; also Towards an Integrated System for Measuring Long Term Changes in Global Sea Level, Joint Oceanographic Institutions, 1990). Human agricultural activity and consumption of fossil fuels are causing a possibly unprecedented increase in carbon dioxide and other greenhouse gases in the atmosphere. This is expected to lead to increased retention of solar heat and thus to global warming. Two possible consequences are more rapid melting of polar ice caps and mountain glaciers and an increase in the volume of water in the oceans because of thermal expansion. Both would cause a worldwide (eustatic) sea-level rise. Such an event, even if it were not large enough to cause widespread inundation, would be disastrous for coastal regions if only because it would increase the rate of coastal erosion. On the other hand, the climate changes
OCR for page 30
INTERNATIONAL GLOBAL NETWORK OF FIDUCIAL STATIONS: SCIENTIFIC AND IMPLEMENTATION ISSUES associated with global warming might lead to increased precipitation over ice sheets, resulting in an increase in the amount of water trapped in polar ice caps and thus a decrease in eustatic sea level. While those who attempt to model the complex interconnection of climatic processes have generated great controversy without much predictive success, other scientists attempting to measure directly the temperature or sea-level manifestations of the indisputable increases in atmospheric carbon dioxide are also facing great difficulties. Relative sea-level (RSL) change would seem to be a fundamental observable that should be obvious over the years at any coastal location. And in fact RSL has been monitored fairly precisely at many locations. Many good tide gauge records go back more than 100 years. The problem is that these observations depend on many factors in addition to changes in the mass of water in the oceans. Factors affecting RSL across the whole range of time scales include: changes in water volume caused by changes in temperature and salinity; changes in area and shape of the ocean basins; changes in volume and spatial distribution of ice; oceanographic signals such as basin-scale circulation, local currents, and tides; atmospheric or meteorological signals such as winds and barometric pressure changes; elastic and viscous response of the solid Earth, including postglacial rebound and tectonic motions; local ground effects such as compaction or changes in the water table; solid Earth tides and ocean loading tidal responses; and changes in the inertia tensor and Earth orientation/rotation parameters associated with changes in the Earth's center of figure relative to the center of mass. These factors were the subject of an extensive review in Sea-Level Change (National Research Council, 1990b). Table 1 summarizes the factors contributing to relative sea-level changes at various time scales. Problems at shorter time scales associated with extracting a reliable rate of eustatic sea-level change from tide gauge records are illustrated by Figure 2. First, as the figure shows, annual deviations from the secular trend can be fairly large—the mean annual sea level varies at the level of centimeters. This variation is due to atmospheric and oceanographic (short-term) influences on RSL.
OCR for page 31
INTERNATIONAL GLOBAL NETWORK OF FIDUCIAL STATIONS: SCIENTIFIC AND IMPLEMENTATION ISSUES TABLE 1. Some mechanisms of sea-level change (from National Research Council, 1990b). Time Scale (years) Order of Magnitude of Change (mm) Ocean steric (thermohaline) volume changes Shallow (0 to 500 m) 10-1 to 102 100 to 103 Deep (500 to 4,000 m) 101 to 104 100 to 104 Glacial Accretion and Wastage Mountain Glaciers 101 to 102 101 to 103 Greenland Ice Sheet 102 to 105 101 to 104 East Antarctic Ice Sheet 103 to 105 104 to 105 West Antarctic Ice Sheet 102 to 104 103 to 104 Liquid Water on Land Groundwater Aquifers 102 to 105 102 to 104 Lakes and Reservoirs 102 to 105 100 to 102 Crustal Deformation Lithosphere Formation and Subduction 105 to 108 103 to 105 Glacial Isostatic Rebound 102 to 104 102 to 104 Continental Collision 105 to 108 104 to 105 Sea Floor and Continental Epeirogeny 105 to 108 104 to 105 Sedimentation 104 to 108 103 to 105
OCR for page 32
INTERNATIONAL GLOBAL NETWORK OF FIDUCIAL STATIONS: SCIENTIFIC AND IMPLEMENTATION ISSUES FIGURE 2. Regional averages of annual sea level anomaly (from Barnett, 1984; see also National Research Council, 1990b, p. 17). Second, the secular trend varies with location, with differences between local trends comparable to the amplitude of the secular trend itself. This effect is due mainly to continued viscous relaxation, or postglacial rebound, following the Pleistocene deglaciation ~10,000 years ago, although tectonic motions and elastic response to loading of the crust also contribute, as we shall see. Perhaps the best understood of the short-term effects is the inverse barometer effect: local sea level rises 1 cm for every 1-mbar decrease in barometric pressure. This effect is important in the midlatitudes, where large pressure fluctuations over
OCR for page 52
INTERNATIONAL GLOBAL NETWORK OF FIDUCIAL STATIONS: SCIENTIFIC AND IMPLEMENTATION ISSUES ground stations. Another low-orbit spacecraft that may carry GPS or other precise tracking equipment for global gravity field mapping is Gravity Probe-B (GP-B). GP-B, planned for the mid-1990s, will fly at a 600-km altitude and will conduct, as its primary mission, the Stanford relativity gyroscope experiment. Regardless of which high-precision system is used—GPS, PRARE, or DORIS—these upcoming or proposed satellite missions to map the global gravity field will be enhanced by a global network of fiducial stations. High-Resolution Imaging Systems Space-based imaging systems such as NASA's Landsat Thematic Mapper and France's Satellite pour l'Observation de la Terre (SPOT) satellite are having a major impact on scientific applications of remote sensing technology. This impact reflects the increasing spatial and spectral resolution of these instruments, increasing ease of data acquisition coupled with decreasing costs, improvements in computer technology and image processing algorithms that permit ever more sophisticated and rapid analyses, and improved models and understanding in the scientific community. Closely related to these changes is increased demand for more sophisticated data products. One example is the need for image data with better geometric fidelity and more accurate location knowledge. This implies the need for improvements in the areas of spacecraft location and orientation (pointing). Spacecraft designers usually distinguish between control and knowledge for both location and orientation. Control requirements are generally less stringent than knowledge requirements, the latter being computed after the fact with the benefit of accurate models and additional data. However, control is required in real time, which often makes this aspect more challenging. Earth observation missions with the most stringent location and orientation requirements can be divided into two classes: high-resolution spectral imaging systems and imaging altimeters. In the first application we require geometric fidelity such that a picture element, or pixel, will be accurately associated with a given spot on the ground. Pixel location knowledge currently planned for imaging sensors in the civilian areas implies that we need to be able to locate pixels to an accuracy of 3 m. Actual performance at present—for locating 20- to 30-m pixels from sensors with high spectral resolution—is considerably worse. Location knowledge is usually improved during postprocessing and image analysis, as the image itself allows location to be
OCR for page 53
INTERNATIONAL GLOBAL NETWORK OF FIDUCIAL STATIONS: SCIENTIFIC AND IMPLEMENTATION ISSUES determined by reference to ground tie points. Nevertheless, because it is desirable to reduce processing and analysis time, accurate a priori location knowledge would be very useful for automating the analysis. Imaging altimeters may actually have more stringent requirements; we will therefore use them to illustrate potential tracking system applications. An imaging altimeter is a high spatial resolution instrument for height measurement over continents and ice caps. The oceanographic counterpart estimates height from time-of-flight measurements of a simple radar pulse with a spherical wavefront, the echo of which can be assumed to arrive at the spacecraft from the exact nadir point. The imaging altimeter, in contrast, uses some sort of scanning or imaging system, such as a scanning laser or synthetic aperture radar imaging system. With this approach, location and pointing knowledge (and to a lesser extent control) are critical to accurate height estimation. Spacecraft location knowledge in the radial component must be better than 0.5 m. Three-dimensional location accuracy, including along-track and cross-track components, can be slightly worse but should not exceed several meters. The required pointing knowledge depends on spacecraft altitude, surface slope, and desired accuracy; for a spacecraft altitude of 500 km, surface slope of 25 degrees, and desired height precision of 1 m, pointing knowledge should be accurate to about 1 arcsec. Most current spacecraft designs for high-precision Earth imaging rely on a combination of sophisticated star trackers or star cameras, which can be supplemented with (inertial) gyroscopes for accurate orientation information (both control and knowledge). They also use relatively crude ground tracking information and/or orbit modeling for real-time location data and more sophisticated orbit modeling for improved location knowledge. Clearly, one of the applications of a global tracking network based, for example, on GPS would be to improve the quality of spacecraft location data, in terms of both control and knowledge. Moreover, this approach is relatively cheap in comparison to existing high-precision tracking techniques, which rely on numerous ground-based lasers (constrained by limitations on satellite visibility from the ground stations) and extensive after-the-fact modeling. The GPS approach has additional advantages where near-real-time location accuracy is required because GPS tracking is essentially geometric, in the sense that range is measured between the orbiter, whose position may be poorly known initially, and several GPS satellites whose positions are well known. Extensive orbit modeling is not required in this geometric approach, implying that high orbit accuracy could be available in near real time.
OCR for page 54
INTERNATIONAL GLOBAL NETWORK OF FIDUCIAL STATIONS: SCIENTIFIC AND IMPLEMENTATION ISSUES The situation is somewhat different with orientation data. Current star trackers are capable of delivering approximately 1-arcsec performance. Although star tracker technology is mature, the instrumentation can be expensive. Can three GPS antennae be used to establish both the location and orientation of a spacecraft in a cost-effective manner, thus eliminating the need for expensive star trackers? Consider first two GPS antennae (separated by a baseline that might range from 1 to 10 m in length for a typical spacecraft), feeding their signals to a common receiver. The lowest achievable standard error in position estimates from GPS data in the absence of ionospheric and atmospheric errors is about 0.1 mm (based on zero-length baseline tests with advanced digital receivers). We conclude that even with a 10-m baseline, the best possible GPS orientation performance (~2 arcsec) is not quite competitive with star tracker technology. However, a case can be made that if high-precision tracking with GPS is required on a given spacecraft for other reasons, a GPS-based orientation system might be attractive to provide redundancy with star tracker orientation systems. This might be appropriate once space-based tracking with GPS becomes routine and the cost of a space-qualified GPS receiver falls. The core network described above for location information would similarly be adequate for orientation applications. Support of Local and Regional Studies Although support of local and regional studies is perhaps the primary objective when defining the need for a service, such as the International GPS Geodynamics Service initiated by the IAG, it probably has only a relatively minor impact on the design of a global fiducial network. On the other hand, applications of the global network data are many, and the following are merely possible examples: Tie local and regional nets to the global network: The main application of the global network is to provide a precise and reliable reference frame. However, for a number of applications, having two or three global sites within 103 km of the survey area will simplify processing and logistics and often result in higher accuracy. In general, however, the existence of globally accurate orbits is likely to have the greatest impact, if only in the form of much simpler and more streamlined processing of local survey data. Precise local geodetic and geophysical surveys: The number of local and regional networks monitored precisely for geophysical purposes (e.g., earthquake
OCR for page 55
INTERNATIONAL GLOBAL NETWORK OF FIDUCIAL STATIONS: SCIENTIFIC AND IMPLEMENTATION ISSUES and volcano studies) is increasing. The support afforded by a globally operated network cannot be overstated. Local dense networks would typically be operated independently of the global network, although a small number of sites could well be incorporated into the global operations, thus providing an automatic high-quality tie. Note that a single global site on Kilauea would permit reliable prediction of rift events! Finally, it should be noted that current efforts toward developing ocean bottom geodetic systems depend critically on the availability of kinematic GPS techniques when tying seafloor monuments to land-based networks. Other applications: For most surveying applications (e.g., cadastral, highways), the global network and associated services (e.g., IGS) would contribute primarily orbit information, in addition to providing a global set of very-well-located primary sites. The data processing could then be streamlined, although questions arise about the timeliness of orbit solutions and data accessibility. In addition, civilian services broadcasting regional corrections to Selective Availability (SA) would clearly benefit from the global network. A global network of fiducial sites would not only benefit the global sciences but would also have positive effects on local and regional surveying operations and thus contribute to other scientific and engineering operations. In many countries points with accurately determined coordinates in a global coordinate system (typically longitude/latitude) are rare but are necessary for many surveying operations. Surveying and cartography are therefore hindered by the lack of accurately determined points in a global network. Errors in the control used are usually noticed when maps from two countries are joined or when engineering projects extend beyond national boundaries. The inaccuracies in the control surveys are detected in such projects because modern surveying instruments are very precise compared to those used in the control surveys. Typically, it is impossible to correct the flaws discovered, so that good new data are forced to fit the existing low-precision control. More very precise control points must be established using GPS, which would be useful for new cartographic or Geographic Information System (GIS) data collections. Locations determined with GPS technology often do not fit easily into the existing network because of the errors it contains. One may argue that the precision proposed in this report is not relevant for surveying (clearly millimeter precision is not necessary for boundary location). But the justification for keeping high precision in boundary surveys and other day-to-day surveying operations is the need for long-term maintenance. Property surveys must be maintained over long periods, so that new measurements must be merged with old ones. Imprecision hinders this process and leads to costly fudging.
OCR for page 56
INTERNATIONAL GLOBAL NETWORK OF FIDUCIAL STATIONS: SCIENTIFIC AND IMPLEMENTATION ISSUES Finally, current GIS technology relies heavily on coordinate values, which in practice are often determined very imprecisely. An accurate global network could provide the control for securing national geodetic networks and tying them to an international standard, thus helping to make data from different sources more compatible—for example, for the study of phenomena on a continental scale. A detailed review of many of these matters, including the requirements for precise control, is provided in the report Spatial Data Needs: The Future of the National Mapping Program (National Research Council, 1990c). In view of these general considerations, progressive densification of the global network beyond the GPS core network is required in the medium to long term. An expanded network, compatible with and tied to the core network, would consist of many more sites, possibly numbering in the hundreds, but not all of them would have to be monitored continuously. Spatial and Temporal Sampling The scientific matters—sea-level change, postglacial rebound, tectonic motions—that the global fiducial network is designed to address call for highly accurate determinations of radial and horizontal velocities (~0.1 mm/yr over ~1,000 km). Redundancy of measurements is crucial, both to reduce noise and to help eliminate systematic errors through recognition and editing of data that are suspect for either geodetic (e.g., instrument malfunction) or geophysical (e.g., site instability) reasons. On the other hand, financial resources are finite, so it is important to recognize the point of diminishing returns. We discuss below a rationale for determining optimal sampling strategy, both in time and in space. Because the forces acting on GPS satellites are much more difficult to model than those acting on a simpler satellite such as LAGEOS, continuous operation is necessary for GPS orbit determination. This means that the GPS core network must be operating continuously with high reliability. The core network also can adequately handle the determination of polar motion and Earth orientation, including high temporal resolution objectives. For most of the other scientific problems we have talked about, the geophysical processes are expected to occur on time scales of years to decades. The frequency of observations should be determined by the trade-off between geodetic accuracy and economic/logistic considerations. The more frequent the measurements, the
OCR for page 57
INTERNATIONAL GLOBAL NETWORK OF FIDUCIAL STATIONS: SCIENTIFIC AND IMPLEMENTATION ISSUES more accurate the (averaged) values, up to some point. For example, several measurements per year clearly give more reliable determinations of crustal motion over decades than do yearly measurements, but it is not clear that daily measurements give more reliable determinations than weekly measurements. On the other hand, the more frequently measurements are made, the more financially attractive continuous monitoring becomes. For example, weekly measurements could be extracted more economically from continuously operating stations than from field campaigns, but it is not obvious that the same is true of quarterly measurements. Similar comparisons can be made with spatial sampling: if most of the resources are spent traveling between sites, a permanent instrument equipped with a data transmission system (telephone or telemetry) may be more attractive. Finally, leaving a permanent installation in the field largely eliminates certain sources of error or noise associated with installation and removal of geodetic equipment. Assuming that high accuracy is the overriding consideration, how often should measurements be taken for optimum operation? Although our experience is still relatively limited, some relevant data are available from VLBI experiments. For example, Figure 6 shows the history of length determinations on the trans-Atlantic baseline from Westford, Massachusetts, to Wettzell, Germany, from 1984 to 1991. Individual length determinations scatter by up to ~50 mm from the best-fitting line (14.9 ± 0.4 mm/yr), even though the outliers otherwise appear to be “good” experiments from the point of view of their residuals. An unfortunate choice of individual measurement points could obscure the entire linear trend, which has a change of ~80 mm over the time shown. Averaging over 90-day windows reduces the scatter by an order of magnitude, but there is still nearly 5 mm of scatter. Examination of Figure 6 also reveals evidence of systematic variations at the noise level (~10 mm) in the original data, some of which survive the 90-day averaging and some of which do not. These variations are as yet poorly understood, although recent solutions indicate the presence of seasonal/annual effects. In addition, if the noise in the data averages followed the classical vN law, the 90-day averages shown would have had a root mean square (rms) scatter of 2.7 mm instead of the 3.2 mm shown. Furthermore, if we decimate the original data set to one point every 90 days (taking the closest raw data points to the middle of the 90-day windows), the VLBI rate estimate becomes 16.5 ± 1.4 mm/yr, which is not resolvable from NUVEL-1. Thus, frequent measurements offer a clear benefit, primarily in the ability to apply statistical techniques and time series analysis to the data. On the other hand, the errors from atmospheric effects are
OCR for page 58
INTERNATIONAL GLOBAL NETWORK OF FIDUCIAL STATIONS: SCIENTIFIC AND IMPLEMENTATION ISSUES FIGURE 6. Evolution of the Westford-Wettzell VLBI baseline length, 1984-1991 (courtesy of T. Herring, Massachusetts Institute of Technology, 1991).
OCR for page 59
INTERNATIONAL GLOBAL NETWORK OF FIDUCIAL STATIONS: SCIENTIFIC AND IMPLEMENTATION ISSUES probably correlated on time scales of 2 to 5 days, so little might be gained by computing solutions more frequently than about weekly. Another example of the potential benefits of dense temporal sampling are the 8 year records along three baselines in the San Francisco Bay area shown by FIGURE 7. Lengths for three geodolite lines in the San Francisco Bay area, showing possible anomalies precursory to Loma Prieta about 1 year before the event. The break in the plots in mid-1984 corresponds to the coseismic offset for the Morgan Hill earthquake. The solid line is simply a smoothed version of the data (from Lisowski et al., 1990).
OCR for page 60
INTERNATIONAL GLOBAL NETWORK OF FIDUCIAL STATIONS: SCIENTIFIC AND IMPLEMENTATION ISSUES
OCR for page 61
INTERNATIONAL GLOBAL NETWORK OF FIDUCIAL STATIONS: SCIENTIFIC AND IMPLEMENTATION ISSUES FIGURE 8. Top panel: Laser strain record from Piñon Flat Geophysical Observatory, compared with Geodolite (open squares; courtesy of W. Prescott, J. C. Savage, and M. Lisowski, U.S. Geological Survey) and 2-color EDM (solid squares; courtesy of J. Langbein, U.S. Geological Survey) measurements of geodetic arrays near PFO. Geodolite array dimension is 15 km, 2-color EDM baselines are 4 km, and laser strainmeter length is 0.732 km. The dashed line gives the long-term strain rate found from the Geodolite network over 15 years (0.03 ± 0.013 µe/yr), in good agreement with the NW-SE strainmeter rate (0.03 µe/yr). The residual strain record highlights abrupt changes due to earthquakes and locally induced deformation. This series is formed by removing Earth tides from the monument- and laser-frequency-corrected strain record. The zero level for all series is arbitrary. Bottom panel: Comparison of absolute gravity (converted to elevation changes) and VLBI. An interesting correlation is evident in 1987, but since there is no anomaly in the PFO strain or tilt data, this may be a coincidence of instrumental effects (courtesy of D. Agnew, H. Johnson, F. Wyatt, and M. Zumberge, University of California, San Diego). Lisowski et al. (1990) to contain signals possibly precursory to the 1989 Loma Prieta earthquake. Lisowski et al. document what they label a “marginally convincing” strain anomaly approximately 1 year before the event, over spatial scales of 30 to 50 km. The time series in this study are shown in Figure 7. It seems believable that there is more to these time series than just random noise superposed on a steady trend, and it seems clear that, without the dense set of points shown, the possible anomalies would certainly be missed. Figure 8 shows a comparison of strains recorded continuously at the Cecil and Ida Green Piñon Flat Geophysical Observatory, California, with several types of geodetic measurements by both ground-based and space-geodetic techniques. Although the various trends agree reasonably well, it is abundantly clear that occasional geodetic occupations will not tell the whole story. But given the apparent scatter in the geodetic solutions, increasing the frequency of occupations to the point of continuous presence may not be scientifically justified. The other side of this issue is, of course, the economics. The economics of continuously operating stations versus occupation by field campaigns depends on the number of campaigns per year, the number of days per campaign, the cost of collecting data continuously, and the cost of returning data to the central archive (Appendix D). For the cost estimates in Appendix D, it was assumed that 20% of
OCR for page 62
INTERNATIONAL GLOBAL NETWORK OF FIDUCIAL STATIONS: SCIENTIFIC AND IMPLEMENTATION ISSUES the expense of maintaining dedicated laboratories can be attributed to the global network; the break-even point is about six campaigns per year, or bimonthly occupations. As discussed by Melbourne (1990), the cost of GPS receivers is decreasing rapidly and the newer generations of receivers are very likely to require much less maintenance in the field than earlier instruments. The result is that the cost of running the global network will be dominated by data processing and communications costs. In this respect, participants in the development of the global network will face the choice of either occupying network sites during temporary campaigns or equipping such sites with permanently installed receivers with a telemetered data stream. Both modes can be scientifically valuable, although we have argued that, at least in some cases, very frequent sampling should be a design constraint. Clearly, the choice of operation mode must depend in part on the characteristics of the site (e.g., availability of power, communications, personnel). However, with a substantial drop in the price of the equipment itself, labor-free, unattended operations with permanently installed equipment will become more and more economical at many sites. The obvious trade-off with the desired sampling rate yields a crossover point where permanent occupation is more cost effective than a temporary campaign that is rapidly shifting such that even fairly infrequent measurements might justify a permanent installation. On the other hand, from the point of view of site selection, the requirements for sustaining data logging and telemetry become correspondingly more important. Considerable economies could result from coordination with other activities with a global perspective, as we shall argue. Spatial sampling requirements are governed primarily by the scientific goals of the network. A number of the objectives we have set forward can be met with a GPS core network of approximately 30 sites, but other objectives, such as the identification of deformation patterns, require a mean intersite spacing small enough to avoid spatial aliasing. The answer is problem and region dependent, but, again, economic considerations rapidly become the main issue, since the number of sites is a quadratic function of the wavelengths to be resolved.
Representative terms from entire chapter: