December 25, 1994
Byron Tapley (tapley@utcsr.ae.utexas.edu)
Chair, Altimeter Study Group
George Born; Dudley Chelton; Robert Cheney; Kathryn Kelly;
Richard Rapp; and Carl Wunsch
1.1 Charge to the ASG
At the request of the EOS Program Scientist, the Altimeter Study Group (ASG) was formed in September 1994 by the Chair of the Oceans Panel of the EOS Investigators Working Group. The ASG's purpose was to evaluate the relative merits of two possible future radar altimeter missions, the second GEOSAT Follow-On (GFO-2) and the TOPEX/Poseidon Follow-On (TPFO), for flight in the late 1990s as the EOS Radar Altimeter mission (EOS ALT-R).
The ASG was charged with the following tasks:
1.2 Membership of the ASG
The eight members of the ASG, charged with providing these assessments, were:
Byron Tapley, Chair University of Texas, Austin George Born University of Colorado Dudley Chelton Oregon State University Robert Cheney National Oceanic and Atmospheric Administration Kathryn Kelly Woods Hole Oceanographic Institution Richard Rapp Ohio State University Drew Rothrock University of Washington Carl Wunsch Massachusetts Institute of TechnologyThe following ex-officio members were asked to provide technical information regarding the GFO-2 and TPFO missions:
Jay Finkelstein Space and Naval Systems Warfare Command Jim Mitchell U.S. Naval Research Laboratory Charles Kilgus Johns Hopkins University, Applied Physics Laboratory Robert Barry Space Systems Division, Ball Aerospace Lee-Leung Fu TPFO Project, Jet Propulsion Laboratory Philip Callahan TPFO Project, Jet Propulsion Laboratory Jean-Francois Minster Centre National d'Etudes Spatiales(CNES)
1.3 The Process
The group was formed by invitation on October 10th. On November 1 members were sent documents supplied by the GFO, TPFO, and EOS ALT-R projects defining their mission requirements, and pertinent sections of the EOS Payload Panel reports. Following a review of this material, the ASG and ex-officio members met on November 9-11, 1994 in Austin, Texas. The agenda provided for briefings by ex-officio members and their colleagues, ranging over the relevant scientific and technical issues. The discussion was vigorous and open. The relative advantages and disadvantages of both missions were debated. During the three-day period, the eight-member working group met in executive session three times to lay out the course of their considerations and to outline the report that follows. All eight members of the ASG participated in the writing and editing of this report through various means, including two telephone conferences in December to discuss the evolving drafts.
This document is the report from the Altimeter Study Group to NASA Headquarters and the EOS Payload Panel. Having delivered this report, the group is dissolved.
In Sections 2 through 6 of the report we discuss the issues of science objectives, the two missions, the measurement accuracies, sampling issues, and inter-agency issues. There are recommendations in each of these sections. In Section 7, we review and summarize our recommendations and findings in the framework of the charge stated in Section 1.1. Acronyms are listed in Section 8.
2. SCIENCE REQUIREMENTS FOR ALTIMETRY
2.1 Altimetric Measurements and Global Change
A large number of documents exist detailing the need for highly accurate and precise altimeter missions in the context of global change. The specific needs of the EOS program have been stated in the recent reports of the EOS Payload Panel. We recapitulate the central ideas.
All space-based sensors are limited to observing phenomena at the sea surface. Altimetry determines the surface elevation of the ocean relative to the Earth's mass center. With knowledge of the geoid (the gravity-induced equilibrium shape of the oceans), the ocean surface departure from the geoid can be used to infer the dynamic motion of the ocean. If the geoid is only poorly known, direct inferences can be made about the time rates of change of the ocean circulation. Uniquely among properties measurable from space, surface elevation is a direct consequence of water motions over the entire water column and can be interpreted in terms of the full three-dimensional movement of the fluid. Because of the vast expense and the logistical and operational difficulties of obtaining globally distributed in situ oceanic observations, altimetry has been identified as the central element of major programs such as the World Ocean Circulation Experiment (WOCE) aimed at understanding the ocean's role in climate. Because climate is a global phenomenon, it is unlikely that it could ever be understood without ongoing altimetric ocean observations.
The original discussions of some 15 years ago that led to the TOPEX mission and the TOPEX/Poseidon (T/P) mission design had determined that an ultimate system measurement accuracy near 1 cm was required to fully meet the oceanographic goals. Some perspective on the evolution of requirements since then is contained in Appendix B. The need for such accuracy and an appreciation for the impact of reducing to the 1 cm level such seemingly small errors as 4 or 5 centimeters can be understood in a number of ways. We give two examples:
Divergence of meridional heat flux: First, consider a major goal of WOCE: the determination of the heat flux divergence to and from the atmosphere. The value of this divergence and its variability over periods of weeks to years is extremely important for understanding the impact of the ocean on the atmosphere. It is believed that the most accurate such estimates are those computed from direct determination of the oceanic flow field and its corresponding temperature transports.
With the present T/P mission, sea surface elevation differences have errors at the 5 cm level. Over 2500 m of water, at mid-latitudes, an erroneous elevation change of 5 cm with respect to the geoid corresponds to a mass transport error of about 13 Sverdrups (13x106m3 s-1). Suppose, as is roughly representative of the Atlantic, warm water in the upper 2500 meters moves northward, and water 10 degrees C colder moves southward in the lower 2500 meters. Then the net meridional heat flux error is about 5x1013 W. If such errors are incurred in each of two estimates at two latitudes about 10 degrees apart (as has been the case for heat flux estimates in the Atlantic at 25 and 36 degrees N), the heat flux error is about 7x1013 W which, when divided by the approximate area (for the Atlantic) between the two sections of about 5x106 km2, gives a heat flux divergence error of about 14 W m-2. For comparison, the thermal forcing owing to a doubled greenhouse gas concentration is believed to be about 4 W m-2. Thus a reduction in the altimetric system errors from near 5 cm to near 1 cm would serve to reduce the present errors in estimates of time rates of change in oceanic heat flux divergence to values close to those anticipated for greenhouse gas increases. (Errors in the existing geoid estimates preclude such accuracies for the absolute values except where the elevation changes take place primarily over the very largest spatial scales; such errors will probably persist until a gravity measuring mission is flown.)
Mean sea level: A second example of the need for extremely high altimetric accuracy is the measurement of mean sea level. Mean sea level changes have always been regarded as both an indicator (symptom) of climate change and as a consequence of such change. Such changes will have huge economic impacts in coastal zones around the world. Using tide gauges and complex and uncertain corrections for tectonic motions of the gauges, estimates exist suggesting that mean sea level has been rising at about 1 to 2 mm yr-1 for roughly the last 100 years. But the sparsity of tide gauges, their poor distribution (because most are located on continental coasts inside harbors and estuaries), and the uncertainty in continental uplift and subsidence, render the estimate extremely uncertain, even as a multi-decadal trend. Modeling studies suggest that this sea level rise may begin accelerating over the next decade.
Recent preliminary estimates from T/P suggest that altimetry has become sufficiently accurate to observe trends of 2 mm yr-1 in global mean sea level by averaging instantaneous 1-cm-precision estimates to obtain sub-millimeter accuracies over extended periods. One must be cautious about acceptance of the conclusion, both because the analyses are preliminary, and because two years is too short to claim a true secular trend. But if the conclusion holds up under further analysis, it should become possible, with systems having T/P-like accuracy, to determine these trends, and any changes in their rates of change, on time scales of a few years, rather than over many decades. Improved or degraded system accuracies and precisions translate directly into corresponding capabilities to detect this small, but immensely important signal.
Other examples for which 1 cm accuracy is important include monitoring annual and semi-annual sea-level variations and the upper ocean heat content, and instability waves related to the El Niño-Southern Oscillation.
We recommend that NASA continue to press toward the goal of one centimeter accuracy for altimetric observations of sea surface height in support of global change scientific objectives and that any follow-on mission should achieve at least the demonstrated accuracy of the T/P mission.
2.2 Changes in Science Requirements
Because the technical capability has been evolving rapidly as the T/P data have been examined and analyzed, specifications fixed several years ago no longer reflect either the desired or realized capabilities. We support a continuing tightening of requirements as technology matures toward the goal of one centimeter measurement accuracy. As a basis for evaluating the candidate altimeter missions,
we reaffirm the Requirements for EOS Satellite Radar Altimetry for Oceanography, attached as Appendix A. We note below where these requirements should be tightened, motivated by the conviction that all EOS ALT-R requirements should be as stringent as the performance being achieved by and anticipated for T/P.
Requirement #4 concerning bias and calibration should be tightened to at least the current T/P performance of 0.5 cm rms absolute calibration with a knowledge of the bias drift rate to an accuracy of 1 mm/yr based on 5 years of data. Past mission accuracy requirements have always been a compromise between the ultimate scientific objective and what seemed feasible in an engineering sense. Thus the radial orbit accuracy requirement #10 should be commensurate with the accuracy currently being obtained in the TOPEX/Poseidon mission, rather than the TOPEX/Poseidon pre-launch specifications. We interpret requirement #14, which states that the tidal frequencies should not be aliased into decadal, annual and semi-annual frequencies, to mean that the orbit should minimize the aliasing into these frequencies and should not alias the dominant M2 and S2 constituents into these frequencies; even the 10-day repeat TOPEX/Poseidon orbit necessarily aliases one tidal constituent (K1) into the semi-annual frequency.
3. GFO-2 and TPFO Missions
GFO-2 and TPFO are continuations of previous altimeter missions flown by the Navy and by NASA. Because the agencies' objectives differ, the missions have been designed with different characteristics.
3.1 TPFO Mission Concept
TPFO, the follow-on to TOPEX/Poseidon, is a NASA/CNES mission whose major objective is to determine the general circulation of the ocean and its variability with sufficient accuracy to allow a quantitative assessment of the ocean's role in the Earth's climatic, hydrological, and biogeochemical systems. The probable launch date is 1999, and the mission design life is 3 years, although the satellite will carry consumables for a 5-year mission. The present proposal for a joint NASA/CNES mission calls for a sequence of at least two satellites, with CNES proposing a three-satellite program. As currently configured, the TPFO mission closely follows the design for the T/P mission (see Table 1). The orbit is circular and prograde with an altitude of 1334 km, an inclination of 66.016 degrees, and a repeat period of 10 nodal days (9.92 solar days). The T/P ground track spacing at the equator is approximately 315 km, and the satellite completes 12.8 orbits each solar day. The sampling characteristics for this orbit provide maximum information on basin- and gyre-scale phenomena, while minimizing the aliasing of meso-scale variability and tides. The orbit altitude was chosen so that the effects of errors in the Earth's gravity field and in the atmospheric density model on precision orbit determination would be significantly reduced relative to those for the 800 km altitude at which previous altimetric spacecraft have flown. The higher altitude also significantly reduces the number of spacecraft maneuvers required to keep the orbit ground track within 1 km of the nominal value.
The TPFO spacecraft is scheduled to carry a two-frequency solid state altimeter, and a three-frequency microwave radiometer, as well as DORIS (Doppler Orbitography and Radiopositioning Integrated by Satellite) and laser retroreflector tracking systems. Although not currently in the baseline mission, there is an option to carry a GPS (Global Positioning System) receiver. As presently planned, the U.S. would provide the radiometer, the laser retroreflector, and the launch vehicle. CNES would provide the spacecraft, the altimeter, and DORIS. The U.S. would have overall project management responsibility; CNES would be responsible for precision orbit determination; and both the U.S. and France would have their own science data processing and distribution systems.
3.2 GFO-2 Mission Concept
GFO-2, the follow-on to GFO-1, is a Navy mission designed primarily to provide near-real-time measuring and monitoring of global mesoscale circulation, wave height, and ice extent for the operational Navy. This mission will provide the research elements of the Navy with data to help meet their requirements for monitoring and modeling the ocean and quantifying its role in global change. The planned launch date for GFO-1 is 1996. The advanced state of the design of GFO-2 would allow it to be launched as early as 1997 if the budget profile would permit. The present EOS budget profile is based on a launch for either GFO-2 or TPFO as EOS ALT-R in 1999 or 2000. The design life for each GFO mission is 8 years.
The GFO-2 mission is currently designed to duplicate the orbital characteristics of GFO. Consequently, the mission is designed around a circular, retrograde orbit with an altitude of 800 km, an inclination of 108 degrees, and a repeat period of 17 nodal days (17.05 solar days). The satellite completes 14.31 orbits per solar day, and the equatorial ground track separation is 165 km; hence, it has better spatial sampling characteristics than TPFO for mesoscale phenomena. During times of high solar activity, precision orbit determination for a spacecraft in this orbit is complicated by increased atmospheric drag. Furthermore, significantly more frequent maneuvers are required to maintain the ground track to within 1 km.
Instruments carried by the GFO-2 spacecraft include a two-frequency solid state altimeter, a two-frequency microwave radiometer, a 20-channel encrypted GPS receiver, and a laser retroreflector. The two-frequency radiometer will rely on the backscatter coefficient from the altimeter to correct the water vapor path delay for surface wind effects.
4. ACCURACY OF SEA SURFACE HEIGHT MEASUREMENT
To measure absolute sea surface height from a satellite, one must determine the geocentric position of the satellite orbit and measure the range from the orbit to the surface. Errors are introduced into these measurements from a variety of sources, as summarized here. A table at the end of this section recapitulates these errors for T/P, TPFO and GFO-2.
4.1 Altimeter Performance
At the level of orbit accuracy being attained for TOPEX/Poseidon, all aspects of the measurement and its corrections warrant close scrutiny. For repeating ground-track measurements, the noise characteristics are probably the least critical, since truly random noise is unlikely to influence the sea level determination as long as the noise is small and does not mask significant biases. The effects of altimeter bias drift and the tracker response to changing sea-surface effects must be understood and reduced by calibration. The TPFO design will be based on the Poseidon altimeter flown on the T/P spacecraft; hence, its characteristics can be evaluated. The GFO-2 altimeter will be based on the GFO-1 altimeter which will not be flown until 1996. The precision of the TOPEX altimeter is estimated to be about 1.7 cm, while the Poseidon altimeter noise is approximately 2.0 cm. The current estimate for the noise of the GFO-2 altimeter is 2.8 cm (for K-band 1-second averages, EOS ALT-R / GFO Convergence Study, Ball Aerospace, October 28, 1994). For the EOS ALT-R objectives of measuring large-scale ocean circulation, the measurement noise for both TPFO and GFO-2 can easily be reduced to less than 1 cm by suitable along-track averaging of the data. Based on their design specifications, the two altimeter instruments thus appear likely to have comparable performance.
We conclude that the instrument characteristics of the radar altimeters for GFO-2 and TPFO are comparable, and the slightly higher noise level associated with the GFO-2 instrument will not likely limit its application to EOS science objectives.
4.2 Calibration
The stability of the biases in the range measurement is a critical concern. Short-term variations in the bias will look like short-period ocean surface signals, while longer term drifts will corrupt the determination of global sea level changes. Furthermore, the bias and the bias drift must be known if altimeter measurements collected by different instruments over several decades are to be combined for global change studies. The internal calibrations must be precisely monitored, while external absolute height calibrations are necessary for interpretation of absolute mean sea level changes. The accuracy with which the orbit can be determined is a fundamental part of the calibration process, which can be aided significantly by an overflight of a Satellite Laser Ranging System. Both satellites carry laser retroreflectors for calibration. The calibration activity is included in the current TPFO mission but not in the GFO-2 mission.
If the GFO-2 option is selected, a calibration activity will need to be added.
4.3 Ionospheric Correction
At radar altimeter frequencies, the ionosphere and atmosphere delay the apparent arrival time of the altimeter pulse. Because the ionospheric delay is frequency dependent, the time-delay effect can be estimated by measuring at two separate frequencies. The estimated precision of the ionospheric correction for TOPEX is about 0.5 cm. Although they are lower power solid-state designs, both the TPFO and GFO-2 are dual-frequency altimeters and should exhibit ionosphere correction precision better than 1 cm.
4.4 Water Vapor Correction
The wet-troposphere range correction derived from the three-frequency microwave radiometer in the TPFO option has been demonstrated from on-orbit T/P data to have an accuracy of 1.1 cm. Estimates derived from a two-frequency radiometer such as that on GFO-2 would require independent information on the wind speed to account for the effects of sea surface roughness and foam on the microwave brightness temperatures. This could be obtained from the altimeter backscatter if accurately calibrated. Pre-launch analysis and simulation studies by Ball Aerospace have concluded that such estimates should achieve an accuracy of 1.4 cm, only slightly less accurate than the 1.1 cm achieved by T/P, and very close to the EOS ALT-R specifications. A study using the T/P altimeter and radiometer data should be conducted to validate the two-frequency approach. The working group understands that such studies are underway.
We conclude that the performance of the water vapor radiometers for GFO-2 and for TPFO are comparable, but the two-frequency GFO-2 concept entails somewhat greater risk because the method has not yet been validated from on-orbit data.
4.5 Orbit Accuracy
The ASG notes that various orbit accuracy requirements have been presented in previously published documents. These range from the 13 cm rms prescribed for the TOPEX mission to the 1 cm rms requirement discussed in Section 2.1. The GFO-2 evaluation study was conducted with a 5 cm requirement for radial orbit error, which was prescribed as the EOS ALT-R requirement at the time the study contract was initiated. The EOS ALT-R requirements in Appendix A call for a radial orbit accuracy of 3 cm rms, with a geographically correlated component of no more than 1 cm rms. The current T/P value of 2.8 cm rms with 1.6 cm rms geographically correlated error is better than the 3.0 cm rms value specified for EOS ALT-R in Appendix A. Further, an accuracy better than 2.0 cm rms is projected by T/P mission end. Table 2 indicates the current TOPEX radial orbit error budget along with the anticipated error budgets for T/P at mission end, GFO-2, and EOS ALT-R.*
Generally speaking, orbit errors from both the gravity field and forces such as atmospheric drag are greater at the 800 km altitude of the GEOSAT orbit than at the 1300 km T/P altitude. For GFO-2, the current state-of-the-art satellite gravity and atmospheric density models predict an rms orbit error of 8 cm rms or greater for a satellite in the GEOSAT orbit using a purely dynamic approach. The proposed method for reducing the effect on orbit accuracy of the satellite dynamic models is to use a reduced dynamics approach, which entails Kalman filtering of the residuals from the best dynamic solution, using data from a receiver capable of tracking up to 10 GPS satellites. Gravity model tuning with data taken on GFO during a period of low solar activity will be required for gravity and drag improvement. Preliminary analyses suggest that if these model improvement activities are pursued, an orbit accuracy on the order of 3 to 4 cm rms, as shown in Table 2, may be possible for GFO-2.
Table 2. Several orbit error budgets, in centimeters.
By contrast, the current T/P radial orbit accuracy is 2.8 cm rms with 1.9 cm rms projected by mission end (see Table 2.) The orbit accuracy obtained by the dynamic and by the reduced dynamic techniques are essentially identical, when a gravity model that includes GPS data is used for the dynamic solution. Furthermore, the accuracy achieved on T/P with the reduced dynamic technique is closely tied to the accuracy of the gravity model. Thus, the T/P results do not provide a definitive test of the degree of improvement that the reduced dynamic technique can provide, but they do confirm the need for the gravity model improvement effort described in the previous paragraph for a satellite in the GEOSAT 800 km orbit.
Radial orbit errors are particularly problematic for ocean circulation studies because of the predominance of the once-per-revolution signal. This signal has a wavelength in excess of 40,000 km and, as Figure 1 illustrates, such a sinusoid with 5 cm rms, for example, can give rise to a bias or slope of 7 cm across an ocean basin. Because the geographically correlated component of the orbit error cannot be reduced by averaging, it is critically important that the maximum accuracy be achieved.
Figure 1. Representative radial orbit errors versus ground
track distance for several levels of rms orbit accuracy. From
Table 2, we see that 5 cm corresponds to the GFO-2 anticipated
error, and 2.8 cm to the T/P performance.
The ASG recommends that the radial orbit accuracy requirement for the EOS ALT-R be restated so as not to exceed 3 cm rms, with a goal of the T/P accuracy at mission end.
Usually there is a cost factor associated with more-accurate measurements. However, in this case, orbits more accurate than the 3.0 cm rms specification are being obtained routinely and should be possible for EOS ALT-R with little additional financial impact.
We note the contribution that the GPS receiver has made to both the T/P gravity and surface model improvement and to the orbit accuracy.
If the TPFO spacecraft is selected, it should include a GPS receiver.
Based on the studies to date, there is reason to believe that the rms orbit altitude error will be at least one centimeter more in the lower GEOSAT orbit than in the higher T/P orbit during periods of low solar activity and could be as much as 2 to 3 cm rms more during solar maxima.
We believe that the difference in the orbit accuracies for the T/P 10-day orbit and the GEOSAT 17-day orbit may have a significant impact on the overall scientific yield of the mission and is a basis for preferring the T/P 10-day orbit.
With an effort patterned after that of the T/P orbit model improvement effort, we believe it may be possible to obtain orbit accuracies for the lower 17-day orbit approaching those being obtained for T/P, but this assumption involves some risk.
4.6 Overall Measurement Error
In Section 4 we have reviewed errors in the overall height measurement from these sources:
The scientific yield will depend on the extent to which all of these errors can be minimized. Summaries of several error budgets for the missions considered by the ASG are given in Table 3. There remains the subtraction of the tidal signal to allow the study of other oceanographic phenomena; this topic is discussed in Section 5.
5. SAMPLING ISSUES
The selection of the orbit configuration has enormous impact on the scientific utility of the data. Besides its effect on orbit accuracy as discussed above, it affects how tides alias into lower frequency signals, temporal and spatial resolution, completeness of global coverage, and orbit crossover geometry.
5.1 Tidal Aliasing
To compute ocean circulation from sea surface topography, one must subtract the tidal component from the observed sea surface height. If a perfect model of ocean tides were available, we could do the subtraction with no error, and tidal aliasing would not be an issue in the selection of altimeter orbit configuration. Until recently, global tide models have been accurate to only 5 or 10 cm which is an unacceptably large error for altimetric studies of large-scale and mesoscale variability. With the availability of T/P data, a large number of new tide models are being developed. The accuracies of these new models have not yet been fully quantified, but several of the models appear to be accurate to 3 or 4 cm when compared with global mid-ocean tide gauge data. With further model refinements and availability of additional T/P data, an accuracy of 2 cm appears achievable in the near future for some of these models. Ultimately, an accuracy of 1 cm may be possible, but present planning should assume an uncertainty of 2 cm. This would represent one of the larger error sources for the EOS ALT-R altimeter. The selection of the orbit configuration should, therefore, be made with careful consideration to tidal aliasing.
So that tidal errors are not misinterpreted as narrow-band signals of other origin known to exist in the ocean, an ideal altimeter orbit configuration would avoid aliasing any of the major tidal constituents into the mean (zero frequency), or into annual or semiannual frequencies or the interannual frequency band associated with short-term climate variability such as the El Ni˜no/Southern Oscillation phenomenon. Of particular concern for the objectives of EOS ALT-R is aliasing near the zero frequency, which would corrupt estimates of the mean and low-frequency, large-scale circulation. It is an unfortunate fact that aliasing of at least one of the tidal constituents into the semiannual frequency is a fundamental limitation of all practical altimeter orbit configurations. The orbit configuration should be chosen so that none of the constituents with the largest expected errors (e.g., the semi-diurnal constituents M2, S2 or N2) alias into the semiannual frequency.
The TPFO option is based on the T/P 10-day repeat orbit, which was adopted for T/P because of its highly desirable tidal aliasing properties. As shown in Table 4, five of the six dominant tidal constituents alias into short periods (46-89 days) that are easily distinguished from the mean, annual and semiannual signals. The K1 constituent aliases into the semiannual band with very long zonal wavelength. Errors in this tidal constituent could complicate interpretation of zonally coherent semiannual variability associated with the seasonal cycle. This is especially true at high latitudes, where the K1 tide generally has largest amplitude. Note that the K1 tidal alias is nearly identical in both the T/P and GEOSAT orbit configurations.
The GFO-2 option is based on the GEOSAT 17-day repeat orbit, which has highly undesirable tidal aliasing characteristics. As shown in Table 4, the P1 constituent aliases into the zero frequency, S2 aliases into the semiannual band, and M2 aliases into the annual band. Distinguishing GFO aliases of the S2 and M2 tides from semiannual and annual frequencies requires minimum record lengths of 6.3 years and 6.6 years, respectively. To do this with statistical reliability, these record lengths should be doubled. To make matters worse in the case of the GFO alias of the M2 tide, the wavelength and westward propagation of the alias are difficult to distinguish from baroclinic Rossby waves in a broad latitudinal band centered near 30 degrees latitude. The amplitudes of annual Rossby waves are typically only a few centimeters, only slightly larger than the anticipated 2 cm accuracy of tide models. As these waves are the dynamical mechanism by which the ocean adjusts to annual atmospheric forcing, it is important that they not be confused with tidal aliases. The 17-day orbit is, therefore, much less desirable than the 10-day orbit configuration from the point of view of tidal aliasing.
We, therefore, conclude that the T/P 10-day orbit meets the EOS altimeter tidal aliasing requirements and that the GEOSAT 17-day orbit does not.
5.2 Spatial and Temporal Resolution
The "best" orbit configuration in terms of spatial and temporal resolution of sea level variations depends on the specific oceanographic application of interest. For example, 10-day repeat T/P data from the tropical Pacific have resolved instability waves with approximately 20-day periods. These short-period variations could not be unambiguously detected in 17-day repeat GEOSAT data. On the other hand, the GEOSAT orbit provides a better spatial description of mesoscale features such as the meandering Gulf Stream and other intense ocean currents. Similarly, the GEOSAT orbit can better resolve large eddies such as those formed at the Agulhas Retroflection in the southeastern Atlantic. Perhaps most importantly, the GEOSAT orbit provides more nearly global coverage of the statistics of eddy variability. For studies of the mean and slowly varying large-scale circulation (the primary science objectives of EOS ALT-R), the 10-day repeat is preferable because it provides better temporal resolution of mesoscale variability at each measurement location than does the 17-day repeat. Mesoscale variability can, therefore, be more effectively removed by low-pass temporal filtering at each measurement location, thus reducing aliasing in the larger scales of interest.
From the perspective of the primary scientific objective of EOS ALT-R, which is "to determine the general circulation of the ocean and its variability with sufficient accuracy to allow a quantitative assessment of the ocean's role in the Earth's climatic, hydrological and biogeochemical systems," the temporal and spatial sampling provided by the T/P 10-day orbit configuration is preferable to the 17-day GEOSAT orbit.
5.3 Maximum Latitude
The 72-degree maximum latitude of the GEOSAT orbit is clearly preferable to the 66-degree maximum latitude of the T/P orbit for a number of applications. The higher latitudinal extent of the GEOSAT orbit increases coverage from 93% to 97% of the world's oceans. Most importantly, the GEOSAT orbit provides complete coverage of the Antarctic Circumpolar Current in the southern hemisphere; part of this current is not observed by the T/P orbit.
For latitudinal coverage, the GEOSAT orbit is preferable.
5.4 Orbit Crossovers
A related issue is the angle between ascending and descending ground tracks at crossover locations. An orthogonal crossover resolves both components of the surface geostrophic velocity. Estimates of the two components of geostrophic velocity can also be obtained from non-orthogonal crossovers, but the errors in the geometrical transformation increase as the crossing angle decreases. The T/P orbit optimizes the crossover angle for middle latitude currents such as the Gulf Stream, the Kuroshio, and the Agulhas Return Current. The GEOSAT orbit optimizes the crossover angle for higher latitude currents such as the Antarctic Circumpolar Current. The issue of orbit crossovers provides no clear basis for preferring either of these options.
5.5 Data Continuity
We noted in Section 2.1 that the detection of trends in mean sea level is a major goal of altimetric missions. It is a commonplace of the statistics of trend detection that gaps and changes in records greatly increase the uncertainty of the results, and they should be avoided to the greatest extent possible. When different measurement systems are used, the most desirable approach is a significant temporal overlap in coverage by the two systems, the duration of the overlap being chosen on the basis of the signals to be resolved and the background variability.
The ASG considered different aspects of the launch schedules of TPFO, GFO-2, and the lifetime of T/P. The scenarios are based on general planning rather than specific project schedules. For our purposes, there are two types of measurement discontinuities: in observation system type (e.g., orbit configuration and measurement accuracy), and in temporal coverage.
GFO-2 would cause a discontinuity in the measurement system. An early launch of GFO-2 in 1997 or 1998 would allow overlap with T/P's expected continuation into late 1997. If budget constraints cause the launch to occur in 1999 or later, then the same temporal gap as expected with TPFO would appear, but there would also be a change in system type, and all of the system features would have changed significantly: ground-track, orbit accuracy, tidal aliasing, etc. If GFO-2 could be guaranteed to launch at least one year before the demise of T/P, the issue is less clear, depending upon the unstudied problems of connecting the GFO configuration to the T/P configuration without significant systematic error.
If GFO-2 is selected, there must be an overlap of about a year to provide a strong relationship between the T/P data record and the new GFO-2 record.
If TPFO is chosen, there would almost certainly be a gap of a year or more, depending upon the actual lifetime of T/P and the launch date of EOS ALT-R. But there would be a very large degree of overall system stability: ground-track coverage, orbit accuracies, tidal model validity, etc.
If a year's overlap is not possible, then continuation of the T/P orbit for EOS ALT-R is preferable to the change in mission configuration associated with the GEOSAT orbit.
6. INTER-AGENCY ISSUES
6.1 Navy Needs
In our review of the GFO-2 and TPFO missions, we were given a summary of the Navy's applications of altimeter data, which can be categorized as: research on ocean circulation and global change, and real-time distribution of orbital data and derived products to ships for operations. The Navy operational requirements are classified, and therefore were not presented to us. The applications in the first category were explained in some detail, and we were advised that, based on experiments using GEOSAT, T/P, and ERS-1 data, either mission would satisfy Navy research and data assimilation needs. The only exception is that the Navy would like 20 Hz data for better spatial resolution in the coastal regions, and this requirement is not part of the current TPFO specifications, although it could be accommodated. The operational requirements include passing to ships various forms of data:
The Navy's need for real-time data would require a modification of the EOS ALT-R mission requirements, which currently provides only for real-time wave heights, not real-time sea surface height.
The GFO-2 mission would provide data over more of the ocean, and its higher maximum latitude and better spatial resolution would give better ice-edge detection. This latter point does not seem particularly compelling, because the ice edge can be observed better using currently available passive microwave data. However, the Navy does require sea surface height information at high northern latitudes such as the Iceland and Norwegian seas. We had difficulty in evaluating the impact of the choice of mission on the Navy, because Navy needs appear to be changing in response to the end of the Cold War with research emphasis shifting from the mesoscale circulation in the deep ocean to the circulation in marginal seas and coastal regions. Neither altimetric mission is particularly useful for these regions because spatial resolution is poor, and because global tidal models needed to infer currents from the altimeter data are not accurate in shallow water. For observing arctic seas and for the older Navy requirement of observing mesoscale variability and eddies, the GEOSAT 17-day orbit configuration is better, although the ground-track resolution is still coarse compared with meso-scale eddies, and the 17-day temporal sampling does not resolve all of the time scales of mesoscale variability.
6.2 Security
The Department of Defense requires that data from its satellites be encrypted for transmission, and that there be restrictions on data availability in time of war. NASA advocates a free and open data policy to encourage other countries to make their data available for weather prediction and climate assessment. Encrypting the data transmission would require substantial modifications to the NASA receiving sites and would jeopardize NASA's credibility in the international community as a proponent of an open data policy. Furthermore, the altimetric data are not as sensitive as most meteorological data, which are not currently encrypted. Further, we note that the Navy currently makes use of data from the ERS-1 and TOPEX/Poseidon satellites, which are freely available, and would undoubtedly make use of any future altimetric data, regardless of the source.
Presently ERS-1 data provide an ocean-wide ground track with a minimum ground track spacing of 8 km. The high-resolution sea surface implied by these data obviates the need for classification of non-real-time data.
We do not see that the encryption of non-real time altimeter data is crucial for the Navy mission. Further we believe that the practice of classifying data whose operational utility has passed is not warranted and should be discontinued.
6.3 Future Navy/APL, NASA/JPL, and CNES Roles
Independent GEOSAT and TOPEX mission teams have co-existed since the early 1980s, and there has been significant overlap and cooperation between the Navy/APL and NASA/JPL engineers who designed, built, and flew these altimeters and satellite tracking systems. Similarly, CNES has successfully demonstrated its capability through development of the Poseidon altimeter and DORIS tracking system. A merger of these missions into one EOS altimeter series thus will have serious consequences in terms of the existing teams. It seems likely that selection of the TPFO option would bring an end to the Navy/APL involvement in development of advanced altimeter systems, and even the NASA/JPL role would be diminished. Selection of the GFO-2 option would exclude JPL and CNES engineers, although CNES could consider an independent French mission. Because of the close coupling of mission engineers and scientists, there are related but less severe consequences to the various altimeter science teams.
The ASG recognizes that a key element in the success of T/P has been the completely "open" management of the mission through all of its elements including hardware, data handling, and calibration/validation. Science Team scrutiny of the end-to-end products has been a major contributor to the accuracies and precision of the data, and the ease with which data have flowed through, and been handled by, the wider community. In particular, the knowledgeable altimetric community is an international one, and the T/P mission results are better than they otherwise would have been, owing to the work of scientists and engineers from many countries. Whatever the configuration of future altimeter missions, it is essential that continuous civilian access to all mission components be assured.
7. FINDINGS AND RECOMMENDATIONS
Having reviewed the two proposed altimeter missions that might serve as EOS ALT-R, we summarize our conclusions in response to the four charges listed in Section 1.1. Several of the secondary recommendations in Sections 4 and 5 that relate to Charge No. 2 are not repeated here.
CHARGE NO. 1.
"Clarify the requirements both of the global change research community and of the Navy for future altimeter missions."
CHARGE NO. 2.
"Given the current mission definitions for GFO-2 and TPFO, state which mission is most suitable to meet the needs of the global change research community."
CHARGE NO. 3.
"State whether the best mission for global change research appears capable of meeting the Navy's operational requirements."
CHARGE NO. 4.
"State what compromises are advisable to reach a common set of altimeter requirements for the Navy and NASA needs."
ACRONYMS
ASG Altimeter Study Group APL Applied Physics Laboratory, Johns Hopkins University CNES Centre National d'Etudes Spatiales DORIS Doppler Orbitography and Radio-positioning Integrated by Satellite EOS Earth Observing System EOS ALT-R EOS Radar Altimeter ERS-1 ESA Remote Sensing Satellite No. 1 GEOSAT Geodetic Satellite GFO GEOSAT Follow-On (satellite series GFO-1, GFO-2) GPS Global Positioning System GRAVSAT Gravity Satellite JPL Jet Propulsion Laboratory LRA Laser Retroreflector Array NASA National Aeronautics and Space Administration SLR Satellite Laser Ranging TOPEX Ocean Topography Experiment T/P TOPEX/Poseidon TPFO TOPEX/Poseidon Follow-On WOCE World Ocean Circulation Experiment
APPENDIX A
For EOS Satellite Radar Altimetry for Oceanography
Requirements
(These are the requirements for EOS ALT-R as of September 9,
1994. The requirements given to the GFO project are an
earlier and different version of TPFO requirements.)
The scientific objectives of EOS Radar Altimetry are:
Determine the general circulation of the ocean and its
variability with sufficient accuracy to allow a quantitative
assessment of the ocean's role in the Earth's climatic,
hydrological, and biogeochemical systems.
Observe global sea level changes; improve the knowledge of
ocean tides; observe ocean wave height; observe ocean
surface wind speed; observe inland water level changes and
land topography, wherever possible; improve the knowledge of
the marine gravity field and the geophysical processes in
the oceanic lithosphere and mantle; and observe changes in
the continental ice sheet wherever possible without
compromising primary oceanography objectives.
EOS Radar Altimeter Science Requirements:
Radar Ranging:
Orbit:
Data:
APPENDIX B
Historical Perspective on Orbit Accuracy
The objective for TOPEX/Poseidon is to observe the general
circulation of the ocean. The T/P measurement objective of a
sea surface height error of no more than 13.2 cm rms was
limited primarily by the capability to compute an accurate
orbit: 13.0 cm rms of this value was due to orbit error. At
the start of the T/P mission planning, the rms radial orbit
accuracy for Seasat (the preceding altimeter mission) was
around l50 cm, so the 13 cm radial T/P orbit accuracy
represented an order of magnitude improvement and was viewed
as an extremely challenging objective.
The ASG notes that the original TOPEX Science Working Group
(SWG) Report, written in 1980, on which the TOPEX mission
was based, specified a radial orbit accuracy of 5.0 cm rms
as the minimal orbit accuracy for the complete range of
oceanographic topics of interest. That report noted that,
for some applications, even this accuracy level was not
totally adequate and proposed to average repeating
measurements over 6 months to achieve a precision
approaching 2 cm rms along each repeating ground track.
The 5 cm rms orbit accuracy was predicated on the assumption
that a gravity mapping mission, GRAVSAT, would be flown to
eliminate the gravity model error, which was the major error
source in the orbit computation. When this mission was not
selected by NASA, the TOPEX project committed to a looser
orbit accuracy level of 13 cm rms, with the gravity model
error contributing 10 cm rms of this total, and focused
science objectives on the primary WOCE goal of measuring the
basin-scale general ocean circulation, which has a maximum
amplitude of 150 cm. The primary objective would be obtained
by averaging multi-year data sets to eliminate smaller
amplitude phenomena with time variations on the order of a
few weeks to a year. The T/P objective became one of the
central topics of interest in current global change studies
and led to the T/P data set being regarded as a primary set
of precursor measurements for the EOS program.
In TOPEX SWG deliberations, it was recognized that better
measurement accuracy would shorten the time required to
determine the large-scale general circulation and there was
a conflict between the requirement for an ocean surface
measurement accuracy approaching 1 cm rms and a
technologically possible accuracy of about 13 cm rms to
which the T/P project could commit. Since the signal
associated with the large scale general ocean circulation
has maximum amplitude on the order of 150 cm, a single track
13 cm rms orbit accuracy was deemed acceptable, provided
that the stated requirement of a geographically correlated
orbit error no greater than 5 cm rms was achieved. This last
requirement was necessary if the previously discussed
averaging was to succeed.
Recognizing that the major limitation to achieving the 13 cm
rms radial orbit accuracy was the gravity model error, the
TOPEX project initiated an extensive effort to improve the
gravity model. The gravity model improvement effort spanned
8 years and reduced the radial rms orbit accuracy from 80 cm
rms using the GEM-10B model (the best model then available)
to 2.8 cm rms using the T/P-developed JGM-3 model. The
gravity model improvement effort also led to an improvement
in the marine geoid, although this correction is still one
of the major error sources in utilizing the altimeter data
to obtain absolute values of the general circulation. The
requirement for the gravity mapping mission, which existed
at the start of the T/P mission, is still present. The
discussion presented in this report argues for a radial
orbit accuracy of 3 cm rms or better.
In light of this experience, the TOPEX/Poseidon Science
Working Group, at its annual meeting in Toulouse in December
1993, recommended that efforts be made to reduce the radial
rms accuracy for the T/P mission from the existing value of
3.5 cm rms to as close to 1 cm rms as possible. Further, the
T/P SWG recommended that the follow-on TPFO mission orbit
accuracy be reduced to match the T/P performance. Following
this meeting, the TPFO project reduced the specified radial
rms accuracy to 5 cm rms.
Although the requirements for orbit accuracy better than 5
cm are consistent with the requirements of the original T/P
Working Group, the ASG recognizes that using recommendations
which were formulated in 1980 to assess the validity of
requirements for a mission in 1998 is inappropriate. The
requirements for the EOS ALT-R should be based on the
science measurement needs, as perceived at present, and the
technological and cost implications associated with
satisfying
these requirements. As the T/P data have indicated, the
ability
to observe changes in the ocean surface topography with
amplitudes of 5 cm rms or less opens up new and important
applications of satellite altimeter data ad substantiates
the requirement of a measurement accuracy as close as
possible
to 1 cm rms.
General:
Editor's Note: The Navy raised issues and
concerns with this report
when initially released in late December. The Altimeter
Sutdy Group subsequently
responded to these concerns and, after careful
consideration, had found no
reason to alter the original report.