Modeled water – vegetation dynamics under revision using GRACE-based data assimilation
- 1Institute of Geodesy and Geoinformation, University of Bonn, Bonn, Germany
- 2Remote Sensing Research Group (RSRG), Department of Geography, University of Bonn, Bonn, Germany
- 3Center of Remote Sensing of Land Surfaces, University of Bonn, Bonn, Germany
- 4Leibniz Centre for Agricultural Landscape Research, Müncheberg, Germany
Water is a major source for growing crops and to ensure freshwater, thus it is essential to prevent the population from water shortages in agriculture and water supply. To globally observe changes in surface water and vegetation from space, remote-sensing satellites enabled a great opportunity in the last decades. But, especially in semi-arid and arid regions observing subsurface water gains a high importance as well. In-situ data and global hydrological models can provide subsurface information, however, the in-situ data are limited to an irregular temporal and spatial resolution that might not cover each climate regime and models do not yet perfectly represent the reality because of structural and forcing uncertainties. So far, the satellite mission GRACE (Gravity Recovery and Climate Experiment) and its successor GRACE-FO (FollowOn) are the only missions that observe the vertical sum of all water storages and thus observe surface and subsurface water, but they are limited to a coarser spatial resolution of about 300 km and can not distinguish between different water storages. To overcome these limitations, we combine GRACE observations with a global hydrological model (WaterGAP 2.2d) via data assimilation to make the model more realistic while spatially downscaling and vertically disaggregating the GRACE data into the different water compartments.
In a case study for South Africa, we use observation-based surface water, soil moisture and groundwater (via assimilation) together with the remote sensed vegetation indices Leaf Area Index and Actual Evapotranspiration (via Moderate Resolution Imaging Spectroradiometer) to extract signatures and subsignals of the water propagation in the water cycle in the period from 2003 to 2016. The observed (via assimilation and remote-sensing) signatures are then compared to modeled signatures and subsignals. Two main processes are analyzed: First, the precipitation-storage dynamics and second, the storage-vegetation dynamics. Thus, we assess the propagation of water that is beginning as precipitation, recharges water storages and finally contributes to vegetation growth. Our study shows an overestimation of the amount of precipitation in the model that refills the water storages and also an overestimation of the amount of water stored that contributes to vegetation growth. Furthermore, we identify differences in the duration of the precipitation-storage-vegetation process. For example, we find that in general the annual peak of modeled groundwater lags the annual precipitation peak by 3 months, while the observations identify a 4-month lag. We believe that this study highlights the importance of assimilating GRACE into hydrological models and that modelers can use this information in future to improve model structures and relevant model processes.
How to cite: Gerdener, H., Kusche, J., Schulze, K., Ghazaryan, G., and Dubovyk, O.: Modeled water – vegetation dynamics under revision using GRACE-based data assimilation, EGU General Assembly 2022, Vienna, Austria, 23–27 May 2022, EGU22-6981, https://doi.org/10.5194/egusphere-egu22-6981, 2022.