6--Solid Earth Sciences and Topography

The purpose of this chapter is to delineate key science objectives in the solid Earth sciences for future NASA orbital radar observations. In order to identify a future role for U.S. radars, it is necessary to specify the types of data that are most needed by the community to answer current geological and geodynamics questions, and to describe the surface phenomena that are best studied using radar. In each instance, it is pertinent to lay out a program of radar observations and analyses that would maximize science returns at reasonable cost. The context for this chapter is to identify NASA's future role in orbital radar systems following the successful completion of the SRL-1 and SRL-2 Shuttle Radar SIR-C/X-SAR experiments. In particular, the relative merits of potential U.S. radars compared to the current and future radar systems to be flown by foreign agencies are addressed.

Research and discoveries in the solid Earth sciences over the past few decades have revolutionized our understanding of the Earth. For example, we now realize that sea floor spreading and associated continental assembly and breakup are primary modes by which the Earth releases internal heat. These processes also control the locations and nature of volcanism and earthquakes. Study of deep sea sediment cores shows that variations in solar insolation related to changes in Earth's orbit caused Pleistocene glacial epochs. The role of these variations on modulation of the long term changes in Earth's climate remains a hotly debated topic. Research has also suggested long term coupling between tectonics and climate, e.g., the emergence of the Tibetan Plateau modified atmospheric circulation in ways that led to enhanced Indian Ocean monsoons. These enhanced monsoons caused increased weathering on the Plateau, leading to increased consumption of carbon dioxide, and a Cenozoic cooldown of the Earth.

NASA's programs in geology and geodynamics have added significant results to our increased understanding of Earth. For example, GPS and VLBI measurements of the velocities of lithospheric plates are consistent with ages inferred from magnetic striping and geochronologies of oceanic crust. Recently, coseismic vertical displacements of centimeters have been measured using radar interferometry (Massonnet et al., 1993, 1994; Zebker et al., 1994). With regard to continental assembly and break up, 23 digital Landsat scenes were mosaicked, and field and isotopic data were used by Sultan et al. (1992) to reconstruct the Nubian and Arabian Shields before opening of the Red Sea. Results show that the great late-Proterozoic Najd transcurrent fault in fact extends into the Nubian Shield and that the Red Sea opening was accomplished with minimal crustal thinning.

Studies of the solid Earth have also contributed in the area of global climatic change and Earth systems science. For example, Arvidson et al. (1994), using SPOT stereogrammetric analyses, field work, and isotopic measurements, were able to model the evolution of the interbedded Quaternary deposits in the Eastern Desert, Egypt, and to show how eustatic sea level, tectonic uplift, and climate affected landforms and deposits. Brakenridge et al. (1994) have used ERS-1 radar as an all-weather system to monitor flood water levels for the great midwestern U.S. floods of 1993.

Lacking in many of these studies of the solid Earth is a detailed knowledge of the topography of the landscape, or the rate of deformation of the surface due, for instance, to earthquakes or active volcanism. It is with this background of solid Earth sciences that measurement objectives for SAR are delineated in the following sections. Because of the increasing national relevance of natural hazards, we also include objectives related to predicting and observing hazards and monitoring their aftermath.

The U.S. Global Change Research Program (GCRP) identified several key solid Earth processes that have a bearing on Global Change. These topics include several that can be investigated with orbital radars: coastal erosion, volcanic processes, surficial processes, and crustal motions and sea level change. The radar systems on board ERS-1, JERS-1 and, in the future RADARSAT and ERS-2, are generally adequate for mapping these phenomena over large regions of the Earth on a routine basis. Where an image in a timely manner is all that is required, these single-wavelength, single-polarization radars can be successfully employed for a wide variety of solid Earth applications, including mapping river flooding (Brakenridge 1994), mapping remote volcanoes that are frequently masked by clouds and long polar nights (Rowland et al., 1994), and paleo-drainage features in deserts (McCauley et al., 1982).

Almost all solid Earth sciences studies benefit from knowledge of topography, where the required resolution depends on the specific study, but generally in the range of 25-30 m spatial and 2-5 m vertical (Figure 6-1). The measurement of topography and topographic change with orbital radars has been demonstrated in a few limited cases (Figure 6-2), but in general this technique is still in its infancy and only limited research has been done on the validation of the data sets. With the exception of a one year time-series for the Landers earthquake in California (Massonnet et al., 1994), no long- term studies of ground deformation have been done with radar interferometry. However, we know from GPS measurements made at permanent sites (e.g., Figure 6-3) that surface displacements at rates of ~3 cm/yr. can occur in volcanic and tectonic environments. Topographic data can be used to calculate local slopes (Figure 6-4), the volumes of landforms (e.g., volcanic cones, river valleys, or sand dunes). Surface displacement may take place due to the rupture of a fault line or the growth of a volcano. In Solid Earth science it is the measurement of topography and topographic change that we consider to be NASA's role in imaging radars; these topics are the most innovative and technologically challenging, while at the same time they also offer the greatest scientific advances.

Radar also offers geologic information on surface properties through the strength of the back scattered radiation at whatever combination is used of frequency, incidence angle, and polarization state. The greatest degree of information is obtained with multiple frequencies, incidence angles, and the full Stokes scattering matrix (van Zyl, 1989). This extensive information is important in studies of surficial deposits and materials because it allows characterization of the scattering process and separation of dielectric constant from textural parameters. The multi-frequency component of radar studies in deserts is important because the different wavelengths penetrate to different depths beneath the surface, thereby providing a crude 3-D view of buried drainage basins in areas such as the Sahara Desert.

Scattering information is important in studies of surficial materials and processes in vegetated areas, e.g., recovery of scoured and sand-covered areas from the 1993 midwestern U.S. floods. In this latter case, herbaceous grasses are encroaching on the affected areas, as are stands of cottonwoods. Multi-temporal polarimetric SAR at C, L, and P bands allow mapping of the vegetation density and character, and show how the underlying surfaces are changing by wind and water-related processes (see Chapter 2-Ecology).

Topographic Mapping and Crustal Motions

Considerable excitement has been generated by interferometric SAR investigations of the 1992 Landers earthquake in California (e.g., Massonnet et al., 1993; Zebker et al., 1994). Indeed the detection and quantitative investigation of ground deformations resulting from an earthquake or a volcano holds great promise for the Earth science community (Mouginis-Mark, 1994). The ability to use orbital radars for topographic mapping and topographic change detection is viewed as the single most important role that a new U.S. radar mission could play, not withstanding other important contributions that the data from such a system could provide for regional mapping and quantitative studies of surface backscatter characteristics.

Topographic data have been generated by combining data for Long Valley, California, from the SRL-1 and SRL-2 missions (Figure 6-2). In this instance, the radar interferometric data can be used to aid in the visualization of geologic features such as mountain ranges and fault zones. The topographic data derived by radar interferometry are also inherently more useful than data derived from stereo air photography because the derived digital elevation model (DEM) is self-consistent. Moreover, the interferometry technique works well in areas of low topographic relief where classical stereo methods tend to fail because tie points cannot be identified.

To develop topographic change maps requires an additional level of knowledge of the surface and local meteorology. A preliminary comparison of SRL-1 and SRL-2 data at the Kilauea, Hawaii volcano area (obtained in April and October 1994) shows variations in phase at a scale of 20-50 km (Figure 6-5) that are currently interpreted as anomalies due to the local weather. This effect is more clearly seen in data collected during the last four days of the SRL-2 mission. Here, phase variations equivalent to surface displacement of a few centimeters appear on successive days, but are absent when data two or three days apart are viewed (Figure 6-6). Although parts of the surface can be seen to be entirely decorrelated due to the eruption of new lava flows (Figure 6-7), the regional trends in phase shift (Figure 6-6) have the morphology and size that are typical of rain cells in this area. Thus, local meteorology may have an important influence on the analysis of radar interferometry data. Such effects cannot be avoided, but they can be quantified by the use of a GPS array on the ground to provide the wet atmosphere delay, or by the inter-comparison of several radar scenes collected in a period of about 1 week.

An important difference in approach to the exploitation of the radar interferometry method for studying ground deformation exists within the Solid Earth Panel. For dry environments, it is clear that the 3-day repeat orbit of ERS-1 was very useful as a demonstration of radar interferometry over arid areas and dry glaciers. This very short time meant that radar data could be correlated even in vegetated regions. In regions where fast growing vegetation is found, a time gap of 8 days appears to be the longest interval between observations that can be tolerated. However, unpublished work by Zebker et al. (1993) with ERS-1 data for volcanoes in Alaska has failed to produce the radar correlation even using 3-day repeat coverage. Snow and ice, and potentially wind disturbing the surface, conspire to prevent ERS-1 data being used in this setting. In addition, the 23deg. incidence angle produces severe lay-over on steep topography. In arid zones, however, the 35-day repeat period (achieved during the non-ice mapping phase of the ERS-1 mission) and much longer repeat cycles are tolerable for interferometry. An ERS-1 quality radar is adequate to detect such movements, provided that before- and after-event images are available. Furthermore, while the 3-day repeat of ERS-1 is useful for interferometry experiments, this mode of operation is a compromise concerning coverage and repetition. Only a small percentage of the Earth is covered in the 3-day repeat mode (~10% coverage compared to that achieved with the 35- day repeat). In addition, the "ice mapping phases" of ERS-1 have been limited to the period between January and March, so that the interferometry coverage available for other parts of the world may not have the appropriate temporal coverage. The ability to collect topographic data from radar interferometry in one pass using a pair of antennas (which is the method used in the TOPSAR airborne system) is an excellent way to avoid the problems of temporal decorrelation of the surface: the interferogram is constructed from data collected in a single pass. There are exciting possibilities in using a boom-mounted second antenna on the Space Shuttle to collect near-global topography during a single mission (SRL-3). This technique is described in more detail below.

Numerous issues remain regarding the type of radar observations that are needed to enable radar interferometry experiments to be successful. Unless a two-antenna interferometer is flown, the temporal frequency of the repeat- pass, the errors due to ionospheric perturbations and atmospheric water vapor, and the surface characteristics are all believed to be important. None of these issues have been rigorously investigated. We can draw on the experience with the ERS-1 and SIR-C/X-SAR missions to make some recommendations for the use of future radars, including the possible reflight of SIR-C/X-SAR as SRL-3. However, it is clear that a combination of orbital radar observations and GPS measurements in the field are both required for geodynamics and topographic analyses for two reasons. First, as was demonstrated with the analysis of the Landers earthquake data (Zebker et al., 1994), the radar-derived deformation map provides only line-of-sight changes with poor temporal resolution. GPS data from well monumented sites are needed to provide the 3-axis ground movement and to provide a continuous record of ground deformation against which the radar data can compared. Second, the neutral atmosphere will have a significant effect on the radar interferometry because the moisture in the atmosphere can induce a time delay equivalent to ground movement up to ~30 cm, which is far in excess of the ground movement that is typically expected along fault lines (except during catastrophic rupture) or dormant volcanoes. New techniques have been developed that use the time delay information contained within the GPS signal to determine the water vapor content of the atmosphere (Bevis and Bussinger, 1994 unpublished data). If such measurements were to be obtained concurrent with imaging radar data for an interferometry experiment, it would be possible to remove this uncertainty in the resultant radar deformation map.

Regional Mapping

There is considerable interest in the geology community for a radar mission that is devoted to mapping as much of the Earth's surface as possible. Radar's ability to penetrate thin sand layers in desert environments has been known since the first Shuttle radar experiment (SIR- A) back in 1981. In desert environments, the applications of a radar (whether multi-wavelength and polarization or single wavelength and polarization) are numerous, and include ground water hydrology (both its development and remediation), mineral exploration (including deposits of economic value such as gravels and clays), paleo-drainage mapping that can serve as indicators of climate change, and archeological research (Figure 6-8). In many other types of geologic settings, comparable mapping mode radar data would also serve to characterize the Earth's environment for comparison with future radar and (in areas that are not frequently cloud covered) Landsat data. In this way, a radar data base would be established that is equivalent to the 20-year record of Landsat images, except that many new areas, such as Indonesia, West Africa, and the Kurile/Aleutian Arc would be imaged.

This approach could be particularly beneficial for inter-disciplinary studies related to glaciers and the hydrologic cycle. Preliminary results for the two SIR-C/X-SAR experiments have shown that numerous features associated with glaciers in the Andes can all be mapped. Through the identification of moraines, pressure ridges, outflow channels, and the snouts of the ice sheets as a function of time, the water balance of the glacier can be determined. This in turn helps in the analysis of the hydrology of the region and has importance for the analysis of climate change, since the glaciers are sensitive indicators of the net loss or build-up of ice due to changes in the local climate.

In each of these instances, it is hard to argue for the need for a new U.S. radar. The issue is really one of regional data coverage, rather than the specific attributes of a SIR-C/X-SAR class radar. Many parts of the world have not been imaged by radar because the ERS-1 spacecraft lacks a tape recorder, and so has to rely on a series of ground stations for direct broadcast of the data. With the advent of RADARSAT in 1995, such a limitation will disappear. In addition, it would be far more economically practical, if the community were to rely on ERS-1 and its successor ERS-2, to establish new ground stations (which cost <1% of a new satellite) to capture additional data over parts of the world where radar data are currently lacking.

Quantitative Lithologic Mapping with SAR

The effects of long- and short-term geologic processes on the Earth's surface are expressed as gross compositional (lithological) differences between rock types, as more subtle within-lithology variation (sedimentary facies, sedimentary structures, facies migration, igneous phases, and hydrothermal alteration), as weathering products and soils, and ultimately as differences in the geomorphic configuration of the land surface (Sparks, 1971). Lithologic mapping potentially can provide information crucial to diverse issues such as desertification, natural resources (oil and gas, minerals, and water); and natural hazards such as volcanoes, earthquakes, and environmental contamination.

Traditionally, optical remote sensing has been used to directly characterize lithology on the basis of composition. There are numerous successful case studies in the published literature. Most of these studies, however, only consider composition at the molecular level, and do not take into account the gross geomorphology of the surfaces being studied. In addition, single-frequency SAR data has long contributed to structural and geomorphic mapping (see Sabins, 1987, for examples). Approaches for the extraction of lithological information usually consist of photo interpretation (including stereoscopic viewing when available). While lithology typically can not be mapped directly without additional information, photogeologic methods still present one of the most valuable approaches towards extracting geologic information from radar data.

The new generation of polarimetric, multi-frequency SARs such as the JPL AIRSAR and SIR-C/X-SAR systems, provide more powerful tools for extracting lithologic information from radar data and for supplementing optical remote sensing determinations of lithology. Because SAR systems interact with the Earth's surface on the 1 to 100+ cm scale, rather than the molecular scale, direct mapping of lithology based on composition is not possible (Elachi, 1987). Radar return is primarily controlled by the relationship between the radar wavelength and the scale of surface roughness with respect to the radar signal (cm-scale relief). Fortunately, the surface morphology of geologic materials is often related to the composition and weathering characteristics of the materials forming the surface. Figure 6-9 shows an example of this, where the roughness of alluvial fan surfaces may be related to the parent materials and the climatic conditions under which the fans formed. If a fan has an average vertical relief of 6 cm, then it might appear smooth (dark) to an L-Band radar system or rough (bright) to an X-band system. Additionally, the multi-frequency nature of these systems allows for theoretical modeling and extraction of quantitative surface roughness based on inversion of the models. Several investigators have used AIRSAR data and model inversions to identify areas with varying surface roughness (van Zyl, 1989; van Zyl et al., 1991; Evans et al., 1992; Kierein-Young, 1993; Kierein-Young and Kruse, 1992). The multiple wavelengths acquired by the AIRSAR and SIR-C/X-SAR systems are critical to successful lithologic mapping using surface roughness parameters, and such an analysis would not be possible with any of the current or planned foreign radar systems.

Analysis of Geomorphic Processes

Quad-pol radar data are also needed to investigate the relationship between radar backscatter and aerodynamic roughness. Although the original motivation for this study was the analysis of sand and dust transport (still valid), the method can also be extended to include climate modeling. In particular, general circulation models (GCMs) currently use only 2 or 3 (e.g., land and ocean) values of aerodynamic roughness, simply because there have been no easy means for getting good values. Yet Sud et al. (1988) have shown that by adding more values (e.g., land, ocean, and desert), GCM predictions change drastically in terms of precipitation, wind regime, etc. Consequently, the work of Greeley and coworkers holds promise as a technique for mapping large areas to derive aerodynamic roughness values for input to the next generation of general circulation models.

The ability of wind to initiate dust and sand storms is dependent on the roughness of the surface at the sub-meter scale, measured by the aerodynamic roughness (Z0). This parameter is a measure of the drag imposed by the surface on the wind and is therefore important also in general circulation models (GCMs). For dry, unvegetated and relatively flat surfaces, the radar backscatter coefficient (sigma-0) is a function of surface roughness at a scale comparable to the radar wavelength. Studies by Greeley et al. (1988, 1991) and Blumberg and Greeley (1993) have shown that there is a correlation between sigma-0 and Z0. Blumberg and Greeley (1993) developed a model to estimate Z0 from radar backscatter coefficients using AIRSAR data and found that the best correlation is with L-Band cross-polarized data. Concurrent measurements with specialized micro-meteorology masts (Figure 6-10) to measure the wind speed, wind direction, and temperature through the lower portion of the atmospheric boundary-layer, calibrated data from SRL-1 for Stovepipe Wells, Death Valley, were analyzed. These models were then applied to SRL-1 data, and an estimate of Z0 was computed. Results show that the models could predict Z0 very closely when using calibrated data (Figure 6-11).

These observations are important because they demonstrate the importance of obtaining calibrated, multi-wavelength and multi-polarization radar data. Use of ERS-1 or RADARSAT data (both C-Band single polarization systems) would not have enabled the same degree of characterization of the eolian roughness of the surface to have been determined.

Orbital radars have also been used to map river flooding and have been recently demonstrated by Brakenridge et al. (1994). In late spring and summer of 1993, the Upper Mississippi River experienced record-setting heavy rainfall. Wisconsin, Iowa, and Illinois experienced the wettest June-July period on record since 1895. In Iowa, measurable rainfall fell somewhere in the state for 33 consecutive days from June 22-July 25. Unusually heavy rains occurred in July, a month that is more commonly characterized by regional moisture deficits.

The radar was used to delineate the flooded agricultural fields along the valley floor and areas of unflooded agricultural land and flooded riverine forest. However, the flood stages determined from ERS-1 scenes lack close-interval sampling in time, and they are not as accurate as in-situ recording gages. The satellite data nevertheless offer an important spatial perspective for particular moments in time. Instead of a continuous local record of rising or falling stage, a single ERS-1 image is a time-instantaneous, spatially continuous portrait of flood stage along as much as 100 km of valley reach. Given favorable flood plain morphology for measuring flood stage, longitudinal profiles of the instantaneous flood surface can be constructed. For this type of investigation, short repeat-time coverage is crucial. This coverage might be achieved using different satellites (e.g., ERS-2 and RADARSAT) or a spacecraft in a short-repeat orbit (although the 3- day repeat of ERS-1 will leave many geographic areas out of the field of view for this part of the mission). RADARSAT could also provide rapid site-revisit capability, provided that the selection of identical viewing geometries is not a requirement.

Capabilities of Current Radars

For many aspects of the Solid Earth research, mapping from single radar images is adequate for the identification of spatial information and general lithologic characteristics. Given the continuing series of radars that will be launched through 2002 (i.e., ERS-2, RADARSAT, ENVISAT ASAR, and ALOS), it does not appear to be cost effective for NASA to fly a comparable radar system. Increasingly, it is appropriate for U.S. scientists to draw upon international partners to provide these imaging radar data sets.

However, these foreign radars have limited capabilities; this means that they are not ideal for making interferometric measurements for three main reasons:

(1) The orbital characteristics of these radars are not optimized for the collection of interferometric data. No foreign radar is planned that would perform the single-pass interferometry for topographic mapping (comparable to the ideas for SRL-3 described below), so that exact-repeat orbits would have to be used. In this case, the exact-repeat frequency varies from 3 days (for selected areas) with the ice-mapping phase of ERS-1 to 24 days for RADARSAT and 35 days for the mapping phase of ERS-1. In the short-term (for about one year), this situation will improve because of the tandem flight of ERS-1 and ERS-2. Exact-repeat coverage will be achieved within one day, but the site-revisit time, within which a second set of measurements can be made, will still be either 3 days (limited areas) or 35 days (global coverage). These long-repeat times mean that surface decorrelation may take place between observations (precluding topographic or ground deformation mapping) and that transient events may be missed entirely. A further issue is that the orbit of the spacecraft may not be known to sufficient detail that the baseline distance between successive orbits can be calculated for the interferometry experiments.

(2) Single wavelength/single polarization radars (including all the planned foreign radars) cannot be used to routinely determine surface topography and topographic change due to the unknown influence of the ionosphere on the time-delay of the radar signal. As is the case with VLBI measurements, the ionospheric delay can be uniquely resolved with two wavelengths. This effect can be reduced by collecting the radar data at night, but this would limit the coverage and site-revisit frequency. In addition, multi-wavelength measurements enhance the probability that radar interferometry experiments can be conducted because they allow a wider range of baseline distances to be considered (X-band measurements require a separation of a few tens of meters, C-band a few hundred meters, and L-band about 1 km).

(3) From the analysis of the SIR-C/X-SAR data, it is clear that for many parts of the world the cross-pol radar data (at both C-Band and L-Band) are much more useful for the analysis of the surface morphology than like-pol (HH or VV) data. No foreign radar has this cross-pol capability, so that lithologic mapping with ERS-1, RADARSAT, or JERS-1 is more difficult than with SIR-C/X-SAR.

It is in this context, where no foreign radar is ideal for interferometric studies, that the idea of flying SRL-3 with an attached boom to conduct single-pass interferometry is so exciting. Indeed it is believed that flying this mission concept would dramatically advance the Earth Science Community's understanding of the planet. While the details of the capabilities of a boom-mounted system are still being investigated, the inherent height accuracy of the proposed X-SAR radar interferometer would be better than 5 m vertical at a spatial resolution of 25 m/pixel. What is truly exciting is the opportunity to collect an almost global (equator-ward of 62deg.) topographic map using the wide-swath (SCANSAR) mode, which provides 50 m/pixel resolution and a height accuracy better than 20 m. In addition, a complete radar image of the world between 62deg.N-62deg.S would also be obtained, thereby providing a benchmark data set against which future global change could be measured. The combined use of X-band at high spatial resolution and C-band in SCANSAR mode would enable the quality of the two DEMs to be tested against each other, as well as provide the regional framework (SIR-C) for detailed local (X-SAR) studies.

A major advantage of the SRL-3 topographic mission is that all of the topographic data would be referenced to a single topographic datum. Until now, the reference surface for maps from one continent do not match the datum for another continent (Figure 6-12). Particularly for long-wavelength (>1,000 km) studies of the gravity field and structure of the Earth, it is crucial to use a common datum (typically the geoid). The SRL-3 data would naturally be referenced to the Earth's geoid. Thus, a whole new range of geophysical and geodynamics studies would be possible that currently cannot be rigorously attempted.

It is our opinion that there is a major leadership role that NASA can play by flying the SRL-3 mission using a second boom-mounted antenna to perform single-pass interferometry. The data set so collected could be one of the most valuable and widely used data sets NASA has ever collected, and would serve as a benchmark data sets for decades to come. In particular, this data set would meet and exceed NASA's commitment to provide a global digital topographic data base in support of the Earth Observing System (EOS) with the added bonus that these data could be made available to the community prior to the launch of the EOS A.M.-1 platform. Synergism between the topographic data set and other information (e.g., Landsat, MODIS, ASTER images) would greatly facilitate studies dedicated to hydrology, ecology and land use.

Development of a New U.S. Radar Satellite

Development of the future U.S. radar system should be seen as incremental in implementation, starting from the existing aircraft TOPSAR system and the SRL-2 interferometry experiments, and ultimately leading to a long-duration free-flyer. Logical steps to achieve this goal are:

(1) Continue the experimental studies of radar interferometry for topographic mapping using the aircraft TOPSAR system. Much remains to be understood about the accuracy of topographic measurements at the meter to tens of meter scale, and the effects of different ground covers (e.g., vegetation vs. bare rock) and regional slopes. A comparison between the topography determined at more than one wavelength (including the effects of atmospheric water vapor on the time delay of the signal) can best be investigated under the controlled conditions of the TOPSAR experiments. In addition, the science community needs a short-term data source to develop expertise in the analysis of high spatial resolution topographic data sets.

(2) The use of a boom-mounted antenna on the SRL-3 Shuttle mission. As described above, this new technology provides the capability to map the Earth's topography (and provide an image) for all the land between 62deg.N and 62deg.S using single-pass interferometry.

(3) A free-flying radar could be constructed from the existing SIR-C/X-SAR hardware, and placed into orbit by the Shuttle for a 1-2 year mission. This mission concept has the advantage that most of the radar technology has already been developed, but the disadvantages are that the orbital inclination will be 57deg. due to the Shuttle launch, and the altitude will be relatively low (resulting in a short-duration mission). However, this orbit enables the ascending and descending orbits of the Shuttle to observe the same point on the ground from almost orthogonal aspect angles. The complete 3-D topography could therefore be mapped: Slopes will be well determined in at least one of the two pairs of scenes. This capability will not be available through the ERS-1 and ERS-2 spacecraft because the near-polar orbits means that the ascending and descending ground tracks intersect at narrow angles.

(4) The ultimate goal should be to launch a free-flying radar into a near polar (for global coverage) with a short (<3 day) exact repeat orbit for rapid site-revisit capability. This system would augment the data base collected by the SRL-3 mission or, alternatively, if SRL-3 is not flown, would establish a baseline topographic data set in its own right. The key difference would be that the coverage would be global as the spacecraft would be placed in a polar orbit, and that the spatial resolution would be uniformly high (as opposed to some areas having the 50 m/pixel coverage provided by SIR-C operating in SCANSAR mode).

In order to avoid surface decorrelation, this free-flying radar should have two antennas so that single-pass interferometry can be achieved. The parameters of this radar should include at least two different wavelengths (to compensate for the ionosphere) and have a variable incidence angle (to permit rapid retargeting of the antenna, as well as reduce lay-over in areas of high topography). Mission duration should be at least 3 years to enable the derivation of a global topographic map followed by monitoring for ground deformation at selected sites.

This mission concept, with two antennas flying on co-orbital platforms, has a great advantage. Interferometry experiments can be conducted for the Solid Earth investigations, and the same spacecraft can also be flown to conduct along-track interferometry that will be of great value for the analysis of ocean surface phenomena. We therefore see strong synergism between the Solid Earth Sciences/Topography and Marine Applications recommendations.

(1) Flying SRL-3 with a boom-mounted second antenna to perform single- pass topographic mapping would not only produce a quantum leap in our understanding of the world's topography, but would also represent a significant technological challenge in radar interferometry. To have the greatest value, SRL-3 should include both X-band and C- band interferometers to collect high and low resolution data, respectively. It is clear that the imaging data will also have great long-term value for global change studies, independent of the topographic mapping. This mission concept offers outstanding possibilities for enhancing NASA's standing in the international radar community at relatively low cost and in a short time frame (the SRL-3 mission could fly within two years).

(2) Continue to fly the NASA airborne TOPSAR/AIRSAR. This airborne system is crucial for continued interferometry algorithm development; the TOPSAR hardware is seen as the vital test-bed to develop new interferometric techniques that ultimately will be tried on spaceborne systems.

(3) NASA needs to support a vigorous data analysis program using the SRL-2 exact-repeat data set. Data obtained during the last four days of the mission will enable proof-of-concept studies in interferometry over an adequate variety of targets. More importantly, different radar baselines and the effects of local meteorology on the phase data can be investigated with this short-repeat coverage.

(4) Utilize data from non-US radar systems to pursue science objectives to the extent that these systems acquire relevant measurements, cause they will be there and we can do some good science. This capability would be significantly enhanced if there are additional ground stations established around the world to take advantage of the tandem mission of ERS-1 and ERS-2. A crucial component of this work with foreign radars is the need for a U.S. investigator team to be funded to work with the data.

(5) NASA should begin to build a free-flyer satellite with interferometric capability and cross-polarization and multiple incidence angle (nominally 20-45deg.) capabilities, to maintain U.S. involvement in unique and important ways. Results from this type of experiment promise to revolutionize several fields of study related to topography and topographic change. Single-pass interferometry is seen as the only way to avoid meteorological errors in the phase data, and a two wavelength SAR is needed to remove ionospheric effects. This same instrument has a very high probability of making quantum leaps in our understanding of Solid Earth and Oceanography sciences, as well as serving as an excellent tool for hazard mitigation studies.

Part of such a mission should have strong international participation so that more than one spacecraft could be used to reduce the time interval between observations. A significant component of such a mission could be dedicated to disaster-monitoring once a baseline data set has been constructed. The ability of the radar to have a wide swath width (up to 500 km at degraded spatial resolution) and the capability to image any point on the Earth's surface through the use of a steerable antenna enhances the possibility of imaging the surface during the event (such as river flooding), rather than just obtaining data once the physical process is over.