[A-HW20] Advances and Emerging Methods in Tracer Hydrology

[AHW20-03] Simulation of an artificial tracer experiment: how can the explicit simulation of tracer transport inform model parameterization and predictions?

*Morgan Peel¹, Hugo Delottier¹, Schilling Oliver³, Blanc Théo¹, Matthias S Brennwald², Rolf Kipfer², Philip Brunner¹ (1.The Centre for Hydrogeology and Geothermics of University of Neuchatel, Switzerland , 2.Department Water Resources & Drinking Water of eawag, Switzerland, 3.Hydrogeology, Department of Environmental Sciences of Universisty of Basel, Switzerland)

Integrated surface-subsurface hydrological models (ISSHMs) are valuable tools for water resource management. The parametrization of such models commonly relies on calibration against hydraulic data, such as hydraulic heads and boundary fluxes. Model calibration against such data alone is associated with parameter non-uniqueness, and as a result to high uncertainties in model predictive capacity. The explicit simulation of tracers in mass-transport models, and the joint calibration of model parameters against hydraulic and tracer data, has the potential to reduce parameter non-uniqueness and therefore predictive uncertainties. Moreover, owing to the sensitivity of solute transport to geological features, tracer measurements may inform model parameterization in terms of the existence and location of preferential flow pathways (e.g. high-K paleochannels).
We present the results of assimilating measured tracer concentrations in the ISSHM HydroGeoSphere. A novel tracer experiment was carried out near a major groundwater pumping site (Bern, Switzerland), where helium (He) was continuously injected into an adjacent river (Emme river) for 35 days. Measured concentrations of dissolved He in both surface water (SW) and groundwater (GW) were included in the calibration dataset of the coupled SW-GW model, in which flow and tracer transport were explicitly simulated. An iterative ensemble smoother (PEST++IES, White [2018]) was employed to calibrate the model in a highly-parameterized framework, where each model element could be individually adjusted. An event-based fluvial model (ALLUVSIM, Pyrcz et al. [2009]) was used to ensure geological realism of the model priors, and to facilitate the emergence of connected structures in post-calibration parameterizations.
We illustrate how combining novel tracer techniques with the explicit simulation of solute transport in ISSHMs helps to constrain the locations of preferential flow and leads to improvements in model predictive ability. We focus on model outputs which are most relevant to water resource management in alluvial systems, such as GW residence time, SW/GW exchange fluxes, and SW-GW mixing ratios.

REFERENCES
Pyrcz, M. J., J. B. Boisvert, and C. V. Deutsch (2009), ALLUVSIM: A program for event-based stochastic modeling of fluvial depositional systems, Computers & Geosciences, 35(8), 1671-1685.
White, J. T. (2018), A model-independent iterative ensemble smoother for efficient history-matching and uncertainty quantification in very high dimensions, Environmental Modelling & Software, 109, 191-201.