Data Assimilation for Integrated Subsurface Flow Problems
M.Sc. Bastian Waldowski, Prof. Dr. Insa Neuweiler
In cooperation with: DFG research unit FOR2131
Numerical subsurface flow models are an important tool for gathering information on subsurface states, such as the root zone soil moisture or the depth to the groundwater table. However, their inputs are often highly uncertain, which also results in a high uncertainty of their outputs. Data assimilation (DA) tackles this uncertainty by estimating the most likely state of a system based on the numerical model forecast, given the observations gained by measurements.
In terrestrial modeling, the interconnectedness of different compartments makes it advisable to use an integrated approach. However, coupled surface/subsurface modeling at a large scale can be very computationally demanding. This is commonly compensated by a higher grid size (compared to a stand-alone model), which may reduce model accuracy significantly due to the smoothing of soil heterogeneities as well as topography. A coarser spatial grid generally leads to a decrease in curvature of the surface, which results in an underestimation of lateral fluxes.
This research project is concerned with the effects such biases have on data assimilation in the integrated surface/subsurface system. Strategies to compensate and correct these biases are analyzed and their impacts on the DA are evaluated. Further, we investigate different approaches for improving estimates in the integrated subsurface system with observations from both the groundwater and the vadose zone.
Letzte Änderung: 10.03.25
