GRACE in the Murray-Darling Basin: Using Gravity Measurements to Improve Hydrologic Prediction in Catchment-Scale Models
Kevin M. Ellett, Jeffrey P. Walker, Rodger B. Grayson, Adam Smith and Matthew Rodell
Hydrologic processes occurring throughout the earth’s surface lead to temporal changes in the distribution of mass which subsequently cause subtle changes in the earth’s gravity field. The GRACE mission (Gravity Recovery And Climate Experiment) of NASA and the German Aerospace Centre will provide global data sets of changes in earth’s gravity field at unprecedented accuracy over the next several years. Thus observations by the twin GRACE satellites have the potential to provide precise measurements of changes in terrestrial water storage occurring over large regions at monthly to annual time scales. By coupling these measurements with numerical modelling we hope to reduce uncertainty in the prediction of hydrologic phenomena in ungauged basins. In this paper we present a methodology designed to address two fundamental questions regarding the applicability of GRACE: are the soil moisture and ground water components of terrestrial water storage change detectable in a gravity signal and can such large-scale measurements of gravity changes be used to improve our understanding and simulation of catchment-scale hydrological processes? The methodology involves three major components: (1) ground-based monitoring of gravity and terrestrial water storage changes; (2) development of a modelling framework for the downscaling/disaggregation and assimilation of gravity data into catchment-scale models; and (3) incorporating other remote-sensed observations into an alternative downscaling/disaggregation scheme. Ground based monitoring is conducted at 40 sites throughout the Murrumbidgee catchment which span the full range of climate, soils, vegetation, topography and land-use (i.e. primary controls on soil moisture distribution) observed throughout the catchment and the larger Murray-Darling Basin (MDB). Monitoring sites were carefully selected to provide insight on sub-grid scale variability (nested sub-catchment and grid-based designs) and to further examine the CASMM concept of Grayson and Western [1998]. The initial hydrologic model employs a catchment-based land surface scheme over the whole of the MDB and comparison of water balance between model simulations and highly accurate monthly GRACE measurements (approximately 4 mm of water uncertainty) will provide insight into model performance and parameterization of dominant processes from a Top-Down perspective. GRACE data will then be spatially disaggregated based on multiple regression analysis of the 40 monitoring sites and a Kalman filter scheme will be used to assimilate the gravity measurements into the hydrologic model. Comparison of model simulated soil moisture (with and without gravity data assimilation) with measured time-series from the 40 Murrumbidgee sites will allow us to quantify the reduction in uncertainty (if any) of predicted soil moisture. An alternative downscaling/disaggregation scheme based on AMSR satellite observations of near surface soil moisture patterns at higher spatial (25km) and temporal (daily) resolution will also be tested. Preliminary results from 18 monitoring sites installed in 2001 suggest that changes in root-zone soil moisture represent the dominant fraction of terrestrial water storage changes occurring in the Murrumbidgee and the magnitude of such changes (as high as 130 mm monthly) should produce a statistically significant signal in both GRACE and ground-based observations of gravity.