Lateral terrestrial water fluxes in the LSM of WRF‐Hydro: Benefits of a 2D groundwater representation

• Published in: Hydrological processes Vol. 36; no. 3
• Main Authors: Rummler, Thomas, Wagner, Andreas, Arnault, Joël, Kunstmann, Harald
• Format: Journal Article
• Language: English
• Published: Hoboken, USA John Wiley & Sons, Inc 01-03-2022; Wiley Subscription Services, Inc
• Subjects: Aquifers; Atmosphere; Atmospheric models; Boundary conditions; Catchment area; Catchments; Complexity; Drainage; Drainage basins; Dry periods; Evapotranspiration; Fluxes; free drainage; Groundwater; Groundwater table; Hydrologic cycle; Hydrological cycle; Hydrology; land‐surface model; Modelling; Moisture content; Moisture effects; Performance measurement; Representations; River basins; Simulation; Soil; Soil moisture; Soil moisture content; Stream discharge; Stream flow; Terrestrial environments; Water availability; Water content; Water table; Weather forecasting; WRF‐hydro
• ISSN: 0885-6087; 1099-1085
• DOI: 10.1002/hyp.14510

The interactions between the atmosphere and the land surface are characterized by complex, non‐linear processes on varying time scales. The Noah‐MP is a medium complexity land‐surface model (LSM), which was recently selected as the new default LSM for the hydrologically enhanced Weather Research and Forecasting modelling system (WRF‐Hydro). Compared to its predecessor, several parameterizations were considerably improved and new ones added, inter alia more sophisticated groundwater descriptions, which aim to replace the traditional free‐drainage lower boundary condition. This study investigates the benefits that can be obtained from a two‐dimensional groundwater representation within the WRF‐Hydro modelling system by performing two offline simulations for the upper Danube river basin. In comparison to the free‐drainage reference simulation, the lateral routing of groundwater and the two‐way interaction with the water table greatly enhances small scale variability in simulated fields of soil moisture content and evapotranspiration (ET). The representation of upward fluxes from the aquifer helps to maintain higher soil moisture contents and thus ET during prolonged dry periods. These differences are rather small though (<2%) and explained by the fact that the study region is considered to be limited by radiative energy and not water availability. The most striking difference however is the performance gap in simulating streamflow. WRF‐Hydro with 2d groundwater scheme clearly outperforms the reference simulation in terms of performance metrics. A comparison with hourly streamflow observations for the water year of 2016 yields average Kling‐Gupta efficiencies of 0.79 versus 0.57 for the reference. Given that both model configurations were not calibrated beforehand, we conclude that the two‐dimensional groundwater option is especially beneficial for applications in poorly or even ungauged catchments. Furthermore, the inclusion of a so far missing compartment of the water cycle in the WRF‐Hydro modelling system allows for a more holistic representation of interactions between atmosphere land surface and subsurface, which will be advantageous in feedback studies with the fully coupled WRF‐Hydro.
The representation of lateral groundwater fluxes and two‐way interactions between water table and soil column within the WRF‐Hydro modeling system lead to a considerable higher spatial heterogeneity in terms of simulated soil moisture content and evapotranspiration. The evaluation of simulated hourly streamflow at four different locations in the study region revealed that the simulation with groundwater representation is far more capable in capturing the dynamics, variability and total amounts of observed streamflow.