Skip to main content

Retrieval of global surface mass variations from space measurements

Final Report Summary - GRAVIMASS (Retrieval of global surface mass variations from space measurements)

GRAVIMASS Final report (10.2008-9.2011)


The aim of the GRAVIMASS project was to improve the retrieval and the interpretation of global mass variations from the Gravity And Recovery Climate Experiment (GRACE) mission.

The GRACE mission, launched in April 2002, consists in two similar satellites co-orbiting on a quasi-polar orbit at an altitude about 450 km. The main improvement, compared to the previous satellites dedicated to the study of the earth's gravity field, has been the precise measurements of the inter-satellite distance through a K-band range-rate (KBRR) link, in addition to accelerometers located at the center-of-mass of each spacecraft, allowing the estimating of the non-conserving forces (atmospheric drag, solar pressure radiation, etc.).

The first two years of the project hosted at the Planetary Geodynamics Laboratory at NASA Goddard space flight center has been mostly dedicated to the improvement of the GRACE processing, especially in testing various forward models needed to correct for high-frequency time-variable gravity variations (see Ray et al., 2009), and in improving the inversion algorithm. Several regional and global solutions have been produced in order to investigate continental water storage variations, as well as oceanic circulation.

During the third year, hosted at ecole et observatoire des sciences de la terre at the university of Strasbourg, we started the valorisation of the different results. Several articles are currently been finalised.
- Improvement of GRACE solutions

The planetary geodynamics laboratory at NASA/GSFC has developed a unique technique to recover mass variations from the GRACE inter-satellite measurements using a localised approach called mascons (mass concentration). These solutions have been shown to be able to recover time-variable gravity field at higher spatial (about 200 km) and temporal (10 days) than classical spherical harmonic solutions.

Compared to the other centers, NASA/GSFC has also tested different forward models of atmospheric, oceanic and hydrological contributions prior to the inversion of GRACE KBRR data. These corrections are needed to correct for known gravity perturbations, which could otherwise be aliased into the 10-day or monthly solutions. One of my contributions has to be to estimate these effects and show their impact at the measurement level (KBRR), but also in the solutions.

During my stay, another significant improvement has been the development of global high-resolution (two degrees and ten days) mascons (see Rowlands et al., 2010 and Sabaka et al., 2010 for more details). I significantly contributed in developing the grid, as well as the validation of the solutions by comparing lake volume variations retrieved by GRACE to estimates derived for altimetry, and over the oceans (see below).

These efforts will be continued, through currently two NASA-funded projects as a collaborator:
- Improved time-variable gravity from GRACE: advanced methods, algorithms and error characterisation (PI: Scott B. Luthcke).
- GRACE and tides (PI: Richard D. Ray).
- Analysis of continental water storage for Africa, Australia and Asia from high-resolution regional mascon solutions

In addition to the global mascons, we have continued in solving regional mascon solutions. Their smaller size allows investigating different geometry, different strategies for the constraints applied for the inversion of KBRR data. I have specifically investigated continental water storage variations over Africa, Australia and Asia.

The GRACE mascon solutions produced for these three areas have been compared to various existing state-of-the-art global and regional hydrology models, in-situ measurements of soil-moisture and groundwater measurements when available (South-East Australia) as well as ocean general circulation models, for validation purpose. Our main goal here was to quantify precisely the improvement in spatial resolution compared to the classical spherical harmonic solutions. We proved that our two-degree solutions are able to capture small wavelength features (typically 200 to 300 km) that cannot be access by any other solutions.

One of the goals of the project was to simultaneously invert GRACE continental water storage (W), with various model of precipitation (E), precipitation minus evapotranspiration (P-E) from atmospheric model and runoff R:
- We decimated various precipitation datasets (TRMM, CMORPH, AMSRE, PERSIANN, CMAP) to two-degree and ten day samples, as well as P-E estimates from different operational (ECMWF) and reanalysis (JRA-25, NCEP, ERA-interim). We found that there are large discrepancies between these different fields, even for the annual cycle (see Boy et al., 2011). For longer-term variations, such an inversion seems currently impossible. We are currently investigating ways to add constraints to our inversion for the annual cycle. Our results will be validated by comparing annual runoff estimates to seasonal river flow measurements, although they are only available for only a few basins.

We also investigate the contribution of surface water to seasonal mass variations in the Amazon basin, by using a routing model, forced by the runoff provided by different hydrology model (Han et al., 2010).
- Retrieval of lake and reservoir water level variations from laser altimetry

In order to validate the GRACE solutions, we have proposed to compare lake and reservoir volume variations retrieved from GRACE high-resolution mascon solutions to estimates derived from altimetry. As lake mass variations is only one component of continental water storage variations, all other effects have to be removed. In most cases, global hydrology models do not take into account surface waters (rivers, lakes, etc.) but also groundwater. As we used GLDAS/Noah as one of the forward model in some of our solutions, we can easily derived lake mass variations from our high-resolution (global or regional) mascon solutions, if we neglect groundwater contributions (valid hypothesis in most places).

Thanks to several decades of radar altimetry missions, water level variations of major lakes are remotely recorded with a precision of a few centimeters at sub-monthly samples. However the large size of the radar footprint on the earth's surface (a few kilometers) can be a significant issue for small bodies such as rivers and narrow reservoirs. Because of its smaller footprint (about 70 m of diameter) and its ability to record the entire waveform, we choose to analyse ICESat (Ice, Cloud and land Elevation Satellite) laser altimeter data for the 2003-09 period, and retrieve water level changes for the major African lakes, as well as the Three Gorges Dam (TGD) reservoir along the Yangtze river in China. We will extend in a close future to other bodies all over the world. Compared to radar waveforms, it is simpler to discriminate ICESat returns depending on the surface characteristics: water, land or mixed.

The forward modeling of the GLDAS/Noah model, prior to the inversion of the KBRR data, allows a retrieval of lake and reservoir volume variations in agreement with the radar altimetry measurements, for the places where the model is valid. This is not the case for example for lake Victoria, where the precipitation data used to force this model is not well calibrated due to the lack of ground rain gauge observation.

Although these research activities differ the topic of the GRAVIMASS project, I also collaborated with the Planetary Geodynamics Laboratory at NASA/GSFC in validating global digital elevation model (GDEM), derived from SRTM (shuttle radar topography mission), ASTER (advanced spaceborne thermal emission and reflection radiometer) or GMTED (the new multi-resolution DEM from USGS) (Carabajal et al., 2011). These two themes of research will be continued with the upcoming ICESat-2 mission; a Science Definition Team proposal has been submitted to NASA for funding:
- Defining valuable ICESat-2 geophysical products, measurements calibration and validation strategies and simulation studies (PI: Claudia C. Carabajal).
- Oceanic mass redistribution from GRACE global mascons

Time-variable gravity field can also be used to assess global mass redistribution within the oceans, in addition to the land hydrology. In collaboration with the MERCATOR-OCEAN group in Toulouse, France, I have compared GRACE derived ocean bottom pressure to various operational and reanalysis models on a global scale, but also on the Mediterranean Sea. As state-of-the-art ocean general circulation models assimilate most of available oceanographic datasets (sea-surface height and temperature, profiles of temperature and salinity from ARGO floats, etc.), their validation requires other measurements. High-resolution mascon solutions allow the retrieval of ocean mass variations at high temporal spatial and temporal resolutions, but not on real-time, we propose to use our global GRACE solution to validate the first version of the MERCATOR-OCEAN global reanalysis (GLORYS1V1) in addition to existing ocean bottom pressure measurements. Because GLORYS is an eddy-resolving model (with a spatial resolution of 25 km), this model is able to capture small wavelength circulation features, unlike GRACE with resolution of about 200 km, and then the overall higher correlation with the ocean bottom pressure measurements. These research activities will continue when other reanalysis models become available.
- Precise orbit determination and loading service

I also collaborated with the Planetary geodynamics laboratory research activities in improving the orbit determination of radar altimeters (especially the CNES/NASA Jason mission) by adding atmospheric, oceanic and hydrological contributions, and then better recovery of water level variations (see Lemoine et al., 2010).

This past year, I also started the development of a loading service providing vertical and horizontal surface displacements caused by atmospheric, oceanic and hydrological loading effects. By the end of 2011, when fully validated, these products will become available, in near real-time, for the entire geodetic community, in order to improve the processing of precise geodetic measurements from VLBI, SLR, DORIS and GPS techniques. Later one, surface gravity variations at various locations (superconducting and eventually absolute gravity measurements) will be added to this service.
- Conclusion and perspectives

Thanks to an innovative localised inversion scheme (mascon), we have been able to resolve surface mass variations at higher spatial (two-degrees) and temporal (ten days) resolutions than classical spherical harmonic solutions. This improvement has been quantified by comparing volume variations of lakes and reservoirs inverted from GRACE K-Band range rate data to estimates derived from radar and laser altimetry, but also by comparing continual water storage to precise regional models (Africa) and in-situ measurements of soil-moisture and groundwater (South-Eastern Australia). Several papers describing our results in Africa, Asia and Australia are currently being finalised. All series will be updated since GRACE mission is still operational.

The inversion of the mass-balance equation still requires some further work, as some constraints need to be added, especially in order to reconcile inconsistencies between space measurements of precipitation (from radiometer and infrared sensors), and between water exchanges with the atmosphere estimated from operational and reanalysis weather models. Only seasonal variations could currently be inverted, as the discrepancies are too large at longer-term timescales.

The collaboration with the planetary geodynamics laboratory will continue, mainly on two principal research activities:
- Time-variable gravity from the GRACE mission and its follow-on. I particularly provide the atmospheric, oceanic and hydrological forward models.
- Processing of ICESat data for the retrieval of lake, reservoir and river water levels, and validation of global digital elevation models.

The development of the loading service (surface displacement and gravity) will hopefully help the geodetic community and improve the processing of VLBI, SLR, DORIS and GPS measurements.

Publications related to GRAVIMASS

Boy, J.-P. J. Hinderer and C. de Linage, Retrieval of large-scale hydrological signals in Africa from GRACE time-variable gravity field, Pure Appl. Geophys., 2011, in press.

Carabajal, C. C., D. J. Harding, J.-P. Boy, J. J. Danielson, D. B. Gesch and V. J. Suchdeo, Evaluation of the Global Multi-resolution terrain elevation data 2010 (GMTED2010) using ICESat geodetic control, SPIE Proceedings of the 'international symposium on LIDAR and radar mapping: technologies and applications- LIDAR & RADAR 2011', Nanjing, China, May 26th to 29th 2011.

Garcia-Garcia, D., B. F. Chao, J.-P. Boy, Water mass variations in the Mediterranean Sea from GRACE mission, J. Geophys. Res., 115, C12016, doi:10.1029/2009JC005928 2010.

Han, S.-C. I.-Y. Yeo, D. Alsdorf, P. Bates, J.-P. Boy, H. Kim, T. Oki, and M. Rodell, Movement of amazon surface water from time-variable satellite gravity measurements and implications for water cycle parameters in land surface models, Geochem. Geophys. Geosyst., 11, Q09007, doi:10.1029/2010GC003214 2010.

Lemoine et al., Towards development of a consistent orbit series for TOPEX, Jason-1, and Jason-2. Advances in Space Research. 46 (12), 1513-1540, 2010.

Ray, R. D., S. B. Luthcke and J.-P. Boy, Qualitative comparisons of global ocean tide models by analysis of intersatellite ranging data, J. Geophys. Res., 114, C09017, doi:10.1029/2009JC005362 2009.

Rowlands, D. D., S. B. Luthcke, J. J. McCarthy, S. M. Klosko, D. S. Chinn, F. G. Lemoine, J.-P. Boy and T. Sabaka, Global mass flux solutions from GRACE ; a comparison of parameter estimation strategies: mass concentrations versus Stokes coefficients, J. Geophys. Res., 115, B01403, doi:10.1029/2009JB006546 2010.

Sabaka, T. J., D. D. Rowlands, S. B. Luthcke and J.-P. Boy, Improving global mass-flux solutions from GRACE through forward modeling and continuous time-correlation, J. Geophys. Res., 115, B11403, doi:10.1029/2010JB007533 2010.