Evaluation of GRACE filter tools from a hydrological perspective

S. Werth, A. Güntner, R. Schmidt, J. Kusche

Research output: Contribution to journalArticlepeer-review

88 Scopus citations


Approximately seven years of time-variable gravity data from the satellite mission Gravity Recovery and Climate Experiment (GRACE) are available to quantify present-day mass variations on and near the Earth's surface. Mass variations caused by the continental water cycle are the dominant signal component after subtracting contributions from atmosphere and oceans. This makes hydrology a primary area of application of GRACE data. To derive water storage variations at the scale of large river basins, appropriate filter techniques have to be applied to GRACE gravity fields given in a global spherical harmonic representation. A desirable filter technique minimises both GRACE data error and signal leakage across the border of the region of interest. This study evaluates the performance of six widely used filter methods (isotropic filters, anisotropic filters and decorrelation methods) and their parameter values to derive regionally averaged water mass variations from GRACE data. To this end, filtered time series from GRACE for 22 of the world's largest river basins were compared to continental water mass variations from a multimodel mean of three global hydrological models (WGHM, GLDAS and LaD). Filter-induced biases for seasonal amplitudes and phases of water storage variations, as well as satellite and leakage error budgets, were quantified for each river basin and explained in terms of storage variations in and around the basin. The optimum filter types and filter parameters were identified for each basin. The best results were provided by a decorrelation method that uses GRACE orbits for the filter design. Our ranking between all filter types and parameters depended on the geographical location, shape and signal characteristics of the specific river basin. Based on a multicriterial evaluation of satellite and leakage error, as well as an error assessment of the hydrological data, the filter selection and parameter optimisation results were shown to be reliable for 17 river basins. The results serve as a guideline for the optimal filtering of GRACE global spherical harmonic coefficients for hydrological applications.

Original languageEnglish (US)
Pages (from-to)1499-1515
Number of pages17
JournalGeophysical Journal International
Issue number3
StatePublished - Dec 2009
Externally publishedYes


  • Hydrology
  • Time series analysis
  • Time variable gravity

ASJC Scopus subject areas

  • Geophysics
  • Geochemistry and Petrology


Dive into the research topics of 'Evaluation of GRACE filter tools from a hydrological perspective'. Together they form a unique fingerprint.

Cite this