Process-based modeling of soil hydrology and its verification at the Biosphere 2 Landscape Evolution Observatory
Final Report Abstract
Water movement in soils is a key process for water and land management, for example to optimize irrigation in agriculture or to limit the discharge of pollutants into water resources. Soil water movement is notoriously difficult to model accurately, and quantitative predictions are challenging already at the plot scale. This becomes even more ambitious at larger scales of hillslopes to catchments relevant for water and land management. The challenges lie in the high nonlinearity of the processes in combination with the heterogeneity of soil hydraulic material properties, which are present at many scales. This is further complicated by typically limited measurement possibilities in the subsurface, that make the measurement of states difficult and of fluxes nearly impossible. This project focused on the representation of the heterogeneity in soil hydrologic models at the transition from the plot scale to the hillslope scale: • At the patch scale the project showed the accuracy of an effective description of soil hydraulic material properties based on few local water content measurement that cannot resolve the full heterogeneity. For the effective description reference material properties were estimated that are locally scaled to incorporate the local heterogeneity at measurement locations. The study showed that the estimated material properties deviate significantly from true material properties. However, accurate predictions of water content at measurement locations and horizontally averaged vertical fluxes are still possible with such an effective description, if an accurate description of evaporation can be achieved. • At the hillslope scale the project explored the impact of the structure of heterogeneity of soil hydraulic material properties on the hillslope discharge behavior. For this the measurement capabilities the Landscape Evolution Observatory (LEO) at Biosphere 2 were used. LEO consists of three artificial hillslopes with an extensive sensor network and allows the control of external forcing. This allowed to perform an experiment, in which through extended constant irrigation a gravity flow regime was reached, where the water content increases until the hydraulic conductivity matches the irrigation flux above. Based on over 400 water content sensors within the hillslopes this allowed to map the heterogeneity of the soil hydraulic conductivity. The experiment showed that it is crucial to include the observed structure in soil hydraulic modeling due to their impact on the hillslopes discharge behavior. However, it also showed that the observed heterogeneity has strong local variations that cannot be resolved, even at a research infrastructure like LEO, and needs to be represented through effective material properties. • Both approaches showed the need for effective soil hydraulic material properties that may differ from the true material properties. To compensate unrepresented heterogeneity these effective material properties may even depend on the specific hydraulic situation. On the example of a strongly simplified hillslope model, a representative elementary watershed (REW) model, the project demonstrated how data assimilation methods may be able to track such temporally changing properties. As part of the outreach to the broader public Biosphere 2 accompanied the gravity flow experiment conducted in this project on Instagram.
Publications
- (2019). Velocity field estimation on density-driven solute transport with a convolutional neural network. Water Resources Research, 55 (8)
Kreyenberg, P. J., Bauser, H. H., & Roth, K.
(See online at https://doi.org/10.1029/2019WR024833) - (2020). Challenges with effective representations of heterogeneity in soil hydrology based on local water content measurements. Vadose Zone Journal, 19 (1), e20040
Bauser, H. H., Riedel, L., Berg, D., Troch, P. A., & Roth, K.
(See online at https://doi.org/10.1002/vzj2.20040) - (2021). Hysteretic behavior of flow recession dynamics: Application of machine learning and learning from the machine
Kim, M., Bauser, H. H., Beven, K., and Troch, P. A.
(See online at https://doi.org/10.1002/essoar.10506592.1) - (2021). Technical Note: Sequential ensemble data assimilation in convergent and divergent systems. Hydrology and Earth System Sciences, 25 (6), 3319–3329
Bauser, H. H., Berg, D., & Roth, K.
(See online at https://doi.org/10.5194/hess-25-3319-2021)
