Monthly mean water storage variations by the combination of GRACE and a regional hydrological model: Application to the Zambezi River

Abstract

The lack of hydrological data and the low quality of global hydrological models' output caused by incapability of proper calibration is the main problem when using hydrological models to study the closure of the water balance at river basin scales. Monthly GRACE gravity field solutions can be used to infer water storage variations at river basin scale. However, the estimates are erroneous due to measurement errors, aliasing effects, and limited spatial resolution, which makes it difficult to verify and improve the hydrological models. Currently, a combination of GRACE gravity data and regional hydrological models appears to be an integral part of future investigations for the estimation of the monthly mean water storage variations over river basins as large as 105 - 106 km2. This paper presents the results of the analysis of 17 monthly GRACE gravity models and of a comparison with the output of the LEW regional hydrological model (Lumped Elementary Watershed) for the estimation of monthly water storage variations over the upper Zambezi River catchment (Southern Africa). The comparison of GRACE and LEW shows significant differences in amplitude, which cannot be attributed unambiguously to an improper calibration of the LEW model, but may also be attributed to the contribution of storage variations in surrounding areas, errors in the GRACE monthly gravity solutions, and low-pass filter errors. The contribution of the surrounding areas to the GRACE monthly mean water storage variations over the upper Zambezi has a maximum amplitude of about 20 mm/month. Applying this correction to the GRACE estimates reduces the maximum amplitudes of the GRACE monthly mean water storage variations to about 20 - 30 mm/month. At the same time the amplitude of the unfiltered water storage variation of the upper Zambezi from LEW is about 60 - 80 mm/month. When applying the same filter to the LEW output as used to smooth the GRACE monthly estimates, the LEW amplitudes reduce dramatically to about 6 - 8 mm/month. This is significantly smaller than the 20 - 30 mm/month obtained from GRACE. The total effect of mass variations in atmosphere and oceans on the GRACE monthly mean water storage variations vary between -10 and +18 mm/month. Ocean effects are of the same order of magnitude as atmospheric effects. Therefore, it is necessary to quantify uncertainties in the de-aliasing procedure currently applied to GRACE data prior to the estimation of gravity field parameters. © Springer-Verlag Berlin Heidelberg 2007.