At this time, about four years of monthly global gravity field solutions from the satellite gravity mission GRACE (Gravity Recovery And Climate Experiment) are available for the scientific public. Transformed into time series of global maps of surface mass anomalies, the data can be used to quantify mass redistributions close to the Earth surface and to contribute significantly to an improved modeling of the underlying processes. Due to the current processing strategies applied by the Science Data Systems for the GRACE gravity recovery, mass variations caused by the continental water cycle dominate the GRACE data series. This makes hydrological model improvement a first application. However, because of correlated errors of the GRACE-only gravity models visible as a striping fragmentation in spatial maps of the GRACE derived gravity data, appropriate filter techniques have to be applied before any analysis. Questions repeatedly posed in this context are: (1) Which filter has sufficient de-striping (i.e. decorrelation) properties? (2) What is the resulting impact on the signal amplitude (i.e. damping) and phase lags? (3) Which filter is optimal for which scale, location or shape of the area to be analyzed? (4) Which filter is preserving local signal properties, thus optimal for which source of mass variations? This study focuses on the third and fourth question by means of a comparative analysis of several filter types from a hydrological perspective, i.e., for surface mass variations by continental water storage changes at the scale of river basins. To this end six published external methods for the derivation of regionally averaged water mass variations from GRACE gravity data were evaluated. These comprise isotropic and anisotropic filters respectively decorrelation methods that use specific constraints on the signal-noise properties of the GRACE and the to-be detected process signal. To evaluate the filter properties, we compute time series of water mass variations from GRACE GFZ-RL04 data and from the WaterGAP Global Hydrology Model (WGHM) for 22 of the world's largest river basin systematically by varying the relevant parameter settings of each filter tool. Applying a criterion of correspondence between modeled and measured time series we obtain a comprehensive assessment of the individual methods. Dependencies on the geographic location, basin shape and signal characteristics relevant for hydrological applications are analyzed and results are confirmed by alternative global hydrological models.
AGU Fall Meeting Abstracts
- Pub Date:
- December 2007
- 1217 Time variable gravity (7223;
- 1800 HYDROLOGY