Performance assessment and Benchmarking of a conceptually coupled groundwater - surface-water model

In continuation to the previously presented methodological approach to estimate vadose zone boundary fluxes titled “A novel conceptualization to estimate unsaturated zone mass-fluxes and integrate pre-existing surface- and ground- water models” at the EGU GA 2022, this study explores the performance of the outlined implementation and benchmarks the model.

 

To recap, the previous study presented a conceptual numerical scheme that aimed to adequately estimate the in- and out- fluxes of the Unsaturated Zone (UZ) with the primary aim of coupling existing groundwater (GW) and surface-water (SW) models. It was expected that such a numerical scheme would provide a viable alternative to solving the computationally expensive Richard’s model for cases where description of fluxes within the UZ and the spatial description of the soil moisture were not in the interest of the modeller. Examples of such cases would be efforts to model the hydro(geo)logical effects of various climate-scenarios, efforts to estimate GW recharge dynamically, and efforts to design integrated watershed management design structures and systems, among others.

 

The model numerical scheme has been implemented in Fortran for computational efficiency and a Python wrapper has been developed for the same for ease of use. The model itself is spatially agnostic and is solved for each model element discreetly, in the UZ. The global simulation period is split into local simulation periods between which the three models (GW, SW, and UZ) exchange information via a coupling scheme. At the beginning of each local simulation period, GW and SW states are read from the respective models (here MODFLOW 6 and Delft 3D - Flexible Mesh), the solution for the UZ is determined by the GWSWEX model given the precipitation and evapotranspiration rates, and the calculated discharges are then prescribed to the respective models. The internal time-step size for the local simulation period is dynamically determined based on the precipitation intensity. The coupling scheme harnesses the Basic Model Interface (developed by CSDMS) offered by both MODFLOW 6 and Delft 3D - Flexible Mesh to rapidly exchange information during the model run without having to restart the models. Support for multiple soil-type layers for the UZ is currently under development.

 

This study aims to assess and establish the capacity to simulate the fluxes of the UZ as desired by benchmarking it for the Tilted-V theoretical catchment setup and comparing its results to the physically based ParFlow model.

In addition to this, the study also aims to assess the reduction in computational resources achieved by employing a conceptual numerical scheme for solving the UZ fluxes.

 

It is expected that the findings of such a study shall point out any necessary improvements, bias-corrections, or considerations to be made in the model development before it may be applied to real-world applications.

The authors also hope that the study fosters discussions to unify the polarising modelling approaches as outlined in Markus Hrachowitz er al., 2017