Maike Schumacher

Calibration/Data Assimilation Approach for Integrating GRACE Data into the WaterGAP Global Hydrology Model (WGHM) Using an Ensemble Kalman Filter: First Results

We introduce a new ensemble-based Kalman filter approach to assimilate GRACE satellite gravity data into the WaterGAP Global Hydrology Model. The approach (1) enables the use of the spatial resolution provided by GRACE by including the satellite observations as a gridded data product, (2) accounts for the complex spatial GRACE error correlation pattern by rigorous error propagation from the monthly GRACE solutions, and (3) allows us to integrate model parameter calibration and data assimilation within a unified framework. We investigate the formal contribution of GRACE observations to the Kalman filter update by analysis of the Kalman gain matrix. We then present first model runs, calibrated via data assimilation, for two different experiments: the first one assimilates GRACE basin averages of total water storage and the second one introduces gridded GRACE data at

A systematic impact assessment of GRACE error correlation on data assimilation in hydrological models

5^\circ

Covariance analysis and sensitivity studies for GRACE assimilation into WGHM

5∘ resolution into the assimilation. We finally validate the assimilated model by running it in free mode (i.e., without adding any further GRACE information) for a period of 3 years following the assimilation phase and comparing the results to the GRACE observations available for this period.

Quantifying the impacts of ENSO and IOD on rain gauge and remotely sensed precipitation products over Australia

There are two spurious jumps in the atmospheric part of the Gravity Recovery and Climate Experiment-Atmosphere and Ocean De-aliasing level 1B (GRACE-AOD1B) products, which occurred in January-February of the years 2006 and 2010, as a result of the vertical level and horizontal resolution changes in the ECMWFop (European Centre for Medium-Range Weather Forecasts operational analysis). These jumps cause a systematic error in the estimation of mass changes from GRACE time-variable level 2 products, since GRACE-AOD1B mass variations are removed during the computation of GRACE level 2. In this short note, the potential impact of using an improved set of 6-hourly atmospheric de-aliasing products on the computations of linear trends as well as the amplitude of annual and semi-annual mass changes from GRACE is assessed. These improvements result from 1) employing a modified 3D integration approach (ITG3D), and 2) using long-term consistent atmospheric fields from the ECMWF reanalysis (ERA-Interim). The monthly averages of the new ITG3D-ERA-Interim de-aliasing products are then compared to the atmospheric part of GRACE-AOD1B, covering January 2003 to December 2010. These comparisons include the 33 world largest river basins along with Greenland and Antarctica ice sheets. The results indicate a considerable difference in total atmospheric mass derived from the two products over some of the mentioned regions. We suggest that future GRACE studies consider these through updating uncertainty budgets or by applying corrections to estimated trends and amplitudes/phases.

Exploring hydro-meteorological drought patterns over the Greater Horn of Africa (1979-2014) using remote sensing and reanalysis products

Recently, ensemble Kalman filters (EnKF) have found increasing application for merging hydrological models with total water storage anomaly (TWSA) fields from the Gravity Recovery And Climate Experiment (GRACE) satellite mission. Previous studies have disregarded the effect of spatially correlated errors of GRACE TWSA products in their investigations. Here, for the first time, we systematically assess the impact of the GRACE error correlation structure on EnKF data assimilation into a hydrological model, i.e. on estimated compartmental and total water storages and model parameter values. Our investigations include (1) assimilating gridded GRACE-derived TWSA into the WaterGAP Global Hydrology Model and, simultaneously, calibrating its parameters; (2) introducing GRACE observations on different spatial scales; (3) modelling observation errors as either spatially white or correlated in the assimilation procedure, and (4) replacing the standard EnKF algorithm by the square root analysis scheme or, alternatively, the singular evolutive interpolated Kalman filter. Results of a synthetic experiment designed for the Mississippi River Basin indicate that the hydrological parameters are sensitive to TWSA assimilation if spatial resolution of the observation data is sufficiently high. We find a significant influence of spatial error correlation on the adjusted water states and model parameters for all implemented filter variants, in particular for subbasins with a large discrepancy between observed and initially simulated TWSA and for north–south elongated sub-basins. Considering these correlated errors, however, does not generally improve results: while some metrics indicate that it is helpful to consider the full GRACE error covariance matrix, it appears to have an adverse effect on others. We conclude that considering the characteristics of GRACE error correlation is at least as important as the selection of the spatial discretisation of TWSA observations, while the choice of the filter method might rather be based on the computational simplicity and efficiency.

VLBI intensive sessions revisited

Climate extremes such as droughts and intense rainfall events are expected to strongly influence global/regional water resources in addition to the growing demands for freshwater. This study examines the impacts of precipitation extremes and human water usage on total water storage (TWS) over the Ganges-Brahmaputra-Meghna (GBM) River Basin in South Asia. Monthly TWS changes derived from the Gravity Recovery And Climate Experiment (GRACE) (2002–2014) and soil moisture from three reanalyses (1979–2014) are used to estimate new extreme indices. These indices are applied in conjunction with standardized precipitation indices (SPI) to explore the impacts of precipitation extremes on TWS in the region. The results indicate that although long-term precipitation do not indicate any significant trends over the two subbasins (Ganges and Brahmaputra-Meghna), there is significant decline in rainfall (9.0 ± 4.0 mm/decade) over the Brahmaputra-Meghna River Basin from 1998 to 2014. Both river basins exhibit a rapid decline of TWS from 2002 to 2014 (Ganges: 12.2 ± 3.4 km3/yr and Brahmaputra-Meghna: 9.1 ± 2.7 km3/yr). While the Ganges River Basin has been regaining TWS (5.4 ± 2.2 km3/yr) from 2010 onward, the Brahmaputra-Meghna River Basin exhibits a further decline (13.0 ± 3.2 km3/yr) in TWS from 2011 onward. The impact of human water consumption on TWS appears to be considerably higher in Ganges compared to Brahmaputra-Meghna, where it is mainly concentrated over Bangladesh. The interannual water storage dynamics are found to be strongly associated with meteorological forcing data such as precipitation. In particular, extreme drought conditions, such as those of 2006 and 2009, had profound negative impacts on the TWS, where groundwater resources are already being unsustainably exploited.

Exploring the influence of precipitation extremes and human water use on total water storage (TWS) changes in the Ganges-Brahmaputra-Meghna River Basin: IMPACTS OF PRECIPITATION EXTREMES ON TWS IN THE GBM BASIN

An ensemble Kalman filter approach for improving the WaterGAP Global Hydrology Model (WGHM) has been developed, which assimilates Gravity Recovery And Climate Experiment (GRACE) data and calibrates the model parameters, simultaneously. The method uses the model-derived states and satellite measurements and their error information to determine updated water storage states. However, due to the fact that hydrological models do not provide any error information, an empirical covariance matrix needs to be calculated. In this paper, therefore, we analyse the combined state and parameter covariance matrix of WGHM. We found that high correlations of up to 0.75 exist between calibration parameters and storage compartments, and that these allow for an efficient calibration. In addition, a sensitivity analysis is performed to identify those parameters that the water compartments are most sensitive to. The performed analysis is important, since GRACE cannot observe the model parameters directly. We found that those parameters, which the water storage is most sensitive to, differ not only regionally, but also with respect to the water compartments. Not unexpected, some climate input multipliers implemented in our model version have an overall strong influence. We also found that the degree of sensitivity changes temporally, e.g. between 0 (in summer) and 0.5 (in winter) for the snow storage.