Groundwater Data Assimilation for Better Water Resources Management [China]
Sustainable water resources management is a challenge in the arid inland watersheds of northern China. Population growth, intensification of agriculture, and modified water availability due to climate change have led to devastating ecological water shortages and to high groundwater production costs where the groundwater level has been lowered due to overconsumption. Robustness of an oasis with regard to water shortages can be improved considerably by optimizing joint use of surface and groundwater resources for the entire catchment. Thereby the groundwater storage is recharged with surface water during high flow periods and tapped when needed during low flow periods.
hydrosolutions ltd. is developing a real-time groundwater flow model for the Heihe River mid-reach for the optimization of the joint use of groundwater and surface water for irrigation. Contrary to conventional numerical groundwater models, real-time models consider model uncertainties and update the model with observations as they are measured, i.e. in real-time. The latter procedure is called data assimilation.
Data assimilation is a standard procedure in control engineering and meteorological forecasting. However, it is still a rather novel tool for hydrological modeling applications.
Data assimilation routines are highly flexible and can be coupled to any kind of model. In the case of the real-time groundwater model in the Heihe mid-reach, we coupled a conventional MODFLOW model to our own implementation of an Ensemble Kalman Filter in MATLAB.
The real-time model produces the best possible interpolation of model states (groundwater levels in the present case) at any given time since it combines a distributed model with uncertainty information with actual measurements. It further provides the user with information about the uncertainty of the current model state and of forecasts of the future model which is crucial for robust decision making.