Short-arc analysis of intersatellite tracking data in a gravity mapping mission

 A technique for the analysis of low–low intersatellite range-rate data in a gravity mapping mission is explored. The technique is based on standard tracking data analysis for orbit determination but uses a spherical coordinate representation of the 12 epoch state parameters describing the baseline between the two satellites. This representation of the state parameters is exploited to allow the intersatellite range-rate analysis to benefit from information provided by other tracking data types without large simultaneous multiple-data-type solutions. The technique appears especially valuable for estimating gravity from short arcs (e.g. less than 15 minutes) of data. Gravity recovery simulations which use short arcs are compared with those using arcs a day in length. For a high-inclination orbit, the short-arc analysis recovers low-order gravity coefficients remarkably well, although higher-order terms, especially sectorial terms, are less accurate. Simulations suggest that either long or short arcs of the Gravity Recovery and Climate Experiment (GRACE) data are likely to improve parts of the geopotential spectrum by orders of magnitude. Received: 26 June 2001 / Accepted: 21 January 2002     

Source characterisation by mixing long-running tsunami wave numerical simulations and historical observations within a metamodel-aided ABC setting

Uncertainty related to the source parameters of earthquake can largely impact the tsunami-induced wave characteristics, especially in the case of near-field tsunami source. The combination of numerical simulations and historical eyewitness accounts can be used to better constrain those uncertainties. In the present study, we propose a Bayesian procedure to infer (i.e. learn) the probability distribution of the source parameters of the earthquake. The strategy is based on the combination of: (1) kriging-based metamodelling techniques to overcome the high computation time cost of the numerical simulator; and (2) Approximate Bayesian Computation (ABC) procedure to perform the Bayesian inference. The procedure is applied to the Ligurian (North West of Italy) 1887 tsunami case, for which tsunami-induced sea surface elevations at the coast have been reported at four locations, namely Marseille, Imperia, Diano-Marina and Genoa. The kriging metamodels are trained using only 300 long-running numerical simulations that were performed using the FUNWAVE-TVD code. Contrary to recent inversion exercises that can benefit from current modern observation networks (like tide gauges, sea bottom pressure gauges, GPS-mounted buoys), the case of historical tsunami like Liguria is complicated by the imprecision and scarcity of the observations: this is accounted for by conducting the combined ABC-kriging procedure a large number of times (i.e. 1000); each time a new set of observations being randomly generated to account for this observational error. The combined analysis of the inference results and of the observation uncertainty reveals that: (1) the coseismic slip is the most important source parameter with a very peaky density distribution around low values ranging from 0.3 to 0.6 m; (2) The fault width has a peaky density distribution around low values ranging from 10 to 12 km; (3) The rake and azimuth distribution only slightly deviate from the uniform prior, hence indicating a low influence of those parameters; (4) The bi-modal distribution of the dip is also evidenced.     

Evaluation and validation of mascon recovery using GRACE KBRR data with independent mass flux estimates in the Mississippi Basin

The direct recovery of surface mass anomalies using GRACE KBRR data processed in regional solutions provides mass variation estimates with 10-day temporal resolution. The approach undertaken herein uses a tailored orbit estimation strategy based solely on the KBRR data and directly estimates mass anomalies from the GRACE data. We introduce a set of temporal and spatial correlation constraints to enable high resolution mass flux estimates. The Mississippi Basin, with its well understood surface hydrological modelling available from the Global Land Data Assimilation System (GLDAS), which uses advanced land surface modeling and data assimilation techniques, and a wealth of groundwater data, provides an opportunity to quantitatively compare GRACE estimates of the mass flux in the entire hydrological column with those available from independent and reliable sources. Evaluating GRACE’s performance is dependent on the accuracy ascribed to the hydrological information, which in and of itself is a complex challenge (Rodell in Hydrogeol J, doi:, 2007). Nevertheless, the Mississippi Basin is one of the few regions having a large hydrological signal that can support a meaningful GRACE comparison on the spatial scale resolved by GRACE. The isolation of the hydrological signal is dependent on the adequacy of the forward mass flux modeling for tides and atmospheric pressure variations. While these models have non-uniform global performance they are excellent in the Mississippi Basin. Through comparisons with the independent hydrology, we evaluate the effect on the solution of changing correlation times and distances in the constraints, altering the parameter recovery for areas external to the Mississippi Basin, and changing the relative strength of the constraints with respect to the KBRR data. The accuracy and stability of the mascon solutions are thereby assessed, especially with regard to the constraints used to stabilize the solution. We show that the mass anomalies, as represented by surface layer of water within regional cells have accuracy estimates of ±2–3 cm on par with the best hydrological estimates and consistent with our accuracy estimates for GRACE mass anomaly estimates. These solutions are shown to be very stable, especially for the recovery of semi-annual and longer period trends, where for example, the phase agreement for the dominant annual signal agrees at the 10-day level of resolution provided by GRACE. This validation confirms that mascons provide critical environmental data records for a wide range of applications including monitoring ground water mass changes.     

Estimated SLR station position and network frame sensitivity to time-varying gravity

This paper evaluates the sensitivity of ITRF2008-based satellite laser ranging (SLR) station positions estimated weekly using LAGEOS-1/2 data from 1993 to 2012 to non-tidal time-varying gravity (TVG). Two primary methods for modeling TVG from degree-2 are employed. The operational approach applies an annual GRACE-derived field, and IERS recommended linear rates for five coefficients. The experimental approach uses low-order/degree $4\times 4$ coefficients estimated weekly from SLR and DORIS processing of up to 11 satellites (tvg4x4). This study shows that the LAGEOS-1/2 orbits and the weekly station solutions are sensitive to more detailed modeling of TVG than prescribed in the current IERS standards. Over 1993–2012 tvg4x4 improves SLR residuals by 18 % and shows 10 % RMS improvement in station stability. Tests suggest that the improved stability of the tvg4x4 POD solution frame may help clarify geophysical signals present in the estimated station position time series. The signals include linear and seasonal station motion, and motion of the TRF origin, particularly in Z. The effect on both POD and the station solutions becomes increasingly evident starting in 2006. Over 2008–2012, the tvg4x4 series improves SLR residuals by 29 %. Use of the GRGS RL02 $50\times 50$ series shows similar improvement in POD. Using tvg4x4, secular changes in the TRF origin Z component double over the last decade and although not conclusive, it is consistent with increased geocenter rate expected due to continental ice melt. The test results indicate that accurate modeling of TVG is necessary for improvement of station position estimation using SLR data.     

Orbit determination of the SELENE satellites using multi-satellite data types and evaluation of SELENE gravity field models

The SELENE mission, consisting of three separate satellites that use different terrestrial-based tracking systems, presents a unique opportunity to evaluate the contribution of these tracking systems to orbit determination precision. The tracking data consist of four-way Doppler between the main orbiter and one of the two sub-satellites while the former is over the far side, and of same-beam differential VLBI tracking between the two sub-satellites. Laser altimeter data are also used for orbit determination. The contribution to orbit precision of these different data types is investigated through orbit overlap analysis. It is shown that using four-way and VLBI data improves orbit consistency for all satellites involved by reducing peak values in orbit overlap differences that exist when only standard two-way Doppler and range data are used. Including laser altimeter data improves the orbit precision of the SELENE main satellite further, resulting in very smooth total orbit errors at an average level of 18 m. The multi-satellite data have also resulted in improved lunar gravity field models, which are assessed through orbit overlap analysis using Lunar Prospector tracking data. Improvements over a pre-SELENE model are shown to be mostly in the along-track and cross-track directions. Orbit overlap differences are at a level between 13 and 21 m with the SELENE models, depending on whether 1-day data overlaps or 1-day predictions are used.     

Spatial quick-clay predictions using multi-criteria evaluation in SW Sweden

The transformation of marine and glaciomarine clay deposits into high sensitive and quick clays is largely dependent on the influence of local and regional geologic history and the resulting stratigraphy. The general conditions that facilitate quick-clay development are well known from numerous laboratory investigations during the last century, but their local and regional in-field variation is less understood. In this study, the geographic distribution of quick clay in SW Sweden is predicted using a multicriteria evaluation model that incorporates both qualitative information (established theory and expert judgment concerning the influences on both quick-clay development and the stratigraphic and geomorphologic distribution of sediment types) and observational data (maps of surficial deposits, geotechnical records and digital elevation data). This information duality cannot be avoided if knowledge from different disciplines is utilized. Considering this, model transparency is important for improvements and for characterizing its reliability for risk analysis. The model was constructed stepwise by an initial parameterization with subsequent hierarchical structuring, weighting and standardization of criteria, before running the full analysis. Comparisons between regional model results and geotechnically documented localities have yielded promising results concerning the model’s ability to predict general trends. However, the large natural and site-specific variability of clay sensitivities is not always captured by the model. These deviations are examined and suggestions are given for minimizing their effect. Applications of model methodology and results are briefly discussed.     

On regularized time varying gravity field models based on grace data and their comparison with hydrological models

Determination of spherical harmonic coefficients of the Earth’s gravity field is often an ill-posed problem and leads to solving an ill-conditioned system of equations. Inversion of such a system is critical, as small errors of data will yield large variations in the result. Regularization is a method to solve such an unstable system of equations. In this study, direct methods of Tikhonov, truncated and damped singular value decomposition and iterative methods of ν, algebraic reconstruction technique, range restricted generalized minimum residual and conjugate gradient are used to solve the normal equations constructed based on range rate data of the gravity field and climate experiment (GRACE) for specific periods. Numerical studies show that the Tikhonov regularization and damped singular value decomposition methods for which the regularization parameter is estimated using quasioptimal criterion deliver the smoothest solutions. Each regularized solution is compared to the global land data assimilation system (GLDAS) hydrological model. The Tikhonov regularization with L-curve delivers a solution with high correlation with this model and a relatively small standard deviation over oceans. Among iterative methods, conjugate gradient is the most suited one for the same reasons and it has the shortest computation time.     

The 1-Centimeter Orbit: Jason-1 Precision Orbit Determination Using GPS, SLR, DORIS, and Altimeter Data

The Jason-1 radar altimeter satellite, launched on December 7, 2001 is the follow on to the highly successful TOPEX/Poseidon (T/P) mission and will continue the time series of centimeter level ocean topography measurements. Orbit error is a major component in the overall error budget of all altimeter satellite missions. Jason-1 is no exception and has set a 1-cm radial orbit accuracy goal, which represents a factor of two improvement over what is currently being achieved for T/P. The challenge to precision orbit determination (POD) is both achieving the 1-cm radial orbit accuracy and evaluating the performance of the 1-cm orbit. There is reason to hope such an improvement is possible. The early years of T/P showed that GPS tracking data collected by an on-board receiver holds great promise for precise orbit determination. In the years following the T/P launch there have been several enhancements to GPS, improving its POD capability. In addition, Jason-1 carries aboard an enhanced GPS receiver and significantly improved SLR and DORIS tracking systems along with the altimeter itself. In this article we demonstrate the 1-cm radial orbit accuracy goal has been achieved using GPS data alone in a reduced dynamic solution. It is also shown that adding SLR data to the GPS-based solutions improves the orbits even further. In order to assess the performance of these orbits it is necessary to process all of the available tracking data (GPS, SLR, DORIS, and altimeter crossover differences) as either dependent or independent of the orbit solutions. It was also necessary to compute orbit solutions using various combinations of the four available tracking data in order to independently assess the orbit performance. Towards this end, we have greatly improved orbits determined solely from SLR+DORIS data by applying the reduced dynamic solution strategy. In addition, we have computed reduced dynamic orbits based on SLR, DORIS, and crossover data that are a significant improvement over the SLR- and DORIS-based dynamic solutions. These solutions provide the best performing orbits for independent validation of the GPS-based reduced dynamic orbits. The application of the 1-cm orbit will significantly improve the resolution of the altimeter measurement, making possible further strides in radar altimeter remote sensing.     

Improved GRACE science results after adjustment of geometric biases in the Level-1B K-band ranging data

The GRACE (Gravity Recovery and Climate Experiment) satellite mission relies on the inter-satellite K-band microwave ranging (KBR) observations. We investigate systematic errors that are present in the Level-1B KBR data, namely in the geometric correction. This correction converts the original ranging observation (between the two KBR antennas phase centers) into an observation between the two satellites’ centers of mass. It is computed from data on the precise alignment between both satellites, that is, between the lines joining the center of mass and the antenna phase center of either satellite. The Level-1B data used to determine this alignment exhibit constant biases as large as 1–2 mrad in terms of pitch and yaw alignment angles. These biases induce non-constant errors in the Level-1B geometric correction. While the precise origin of the biases remains to be identified, we are able to estimate and reduce them in a re-calibration approach. This significantly improves time-variable gravity field solutions based on the CNES/GRGS processing strategy. Empirical assessments indicate that the systematic KBR data errors have previously induced gravity field errors on the level of 6–11 times the so-called GRACE baseline error level. The zonal coefficients (from degree 14) are particularly affected. The re-calibration reduces their rms errors by about 50%. As examples for geophysical inferences, the improvement enhances agreement between mass variations observed by GRACE and in-situ ocean bottom pressure observations. The improvement also importantly affects estimates of inter-annual mass variations of the Antarctic ice sheet.