Résultat de la recherche
Noble gases as tracers of surface water – groundwater interactions: insights from novel field and modelling approaches
Understanding the interactions between surface water (SW) and groundwater (GW) in alluvial aquifer systems is of critical importance for the sustainable management of water resources. Advances in realtime and continuous measurement of a range of hydrological tracers provide new opportunities for the characterization of SW-GW dynamics at unprecedented spatiotemporal resolutions. Amongst several promising tracer methods, noble gases are particularly well-adapted to the study of SW-GW interactions, and provide an integrated signal on the flow paths and travel times of water. Capitalizing on the insights offered by novel measurement technologies requires tracer interpretation methods that appropriately capture tracer transport processes in dynamic environmental conditions. However, recourse to highly-simplified tracer interpretation methods, conceptually detached from the complexity of natural systems, is still widespread. In such cases, the potentially rich information content of tracer measurements may be underutilized. This thesis aims at investigating how established and emerging noble gas tracer methods can be optimally used - and when they should be avoided - for the study of SW-GW interactions in alluvial aquifer systems. To this end, a range of novel laboratory, field, and modelling approaches are employed to systematically assess the applicability, limitations, and potential of some gas tracer methods toward informing SW-GW exchange processes. The first gas tracer method examined is the 222Rn apparent age model, which is widely used to estimate the ages of very young GW (days to weeks). High-resolution measurements of the spatial distribution of 222Rn emanation rates in an alluvial aquifer reveal significant spatial heterogeneity in 222Rn production. The explicit simulation of 222Rn in synthetic mass-transport models shows that this level of heterogeneity, combined with mixing of GW, can result in strongly biased estimates of GW age, effectively limiting the applicability of the 222Rn apparent age method to a limited range of field conditions. Although temporal changes in 222Rn concentrations may reveal insights into GW age dynamics, the information content of 222Rn measurements may be best extracted through the integration of 222Rn observations in the calibration process of physics-based flow and transport models. Indeed, the second part of this thesis is devoted to exploring how the explicit simulation of tracer concentrations and the assimilation of untransformed tracer data in highly parameterized, physicsbased models may inform model parameters and ultimately predictions of management interest. Within this framework, the joint assimilation of hydraulic and noble gas data (222Rn and helium), acquired over the course of a novel tracer injection experiment, is shown to strongly inform model parameters and reduce predictive uncertainty of several important water management quantities, such as GW age, SW-GW mixing ratios, and SW infiltration fluxes, far beyond what is achieved with “traditional” hydraulic data alone. These results build upon mounting evidence as to the benefits of explicitly simulating and assimilating diverse observation types with physically-based flow and transport models, avoiding the layer of conceptual simplification and potential bias accrued with simplified tracer interpretation models. Finally, the successful combination of novel gas injection methods, developed over the course of this project, and the assimilation of high-resolution gas tracer measurements in an explicit tracer simulation framework strongly support further developments of (noble) gas tracer methods and tracer-numerical model synergies.
A Framework for Untangling Transient Groundwater Mixing and Travel Times
Understanding the mixing between surface water and groundwater as well as groundwater travel times in vulnerable aquifers is crucial to sustaining a safe water supply. Age dating tracers used to infer apparent travel times typically refer to the entire groundwater sample. A groundwater sample, however, consists of a mixture of waters with a distribution of travel times. Age dating tracers only reflect the proportion of the water that is under the dating range of the used tracer, thus their interpretation is typically biased. Additionally, end-member mixing models are subject to various sources of uncertainties, which are typically neglected. In this study, we introduce a new framework that untangles groundwater mixing ratios and travel times using a novel combination of in-situ noble gas analyses. We applied this approach during a groundwater pumping test carried out in a pre-alpine Swiss valley. First, we calculated transient mixing ratios between recently infiltrated river water and regional groundwater present in a wellfield, using helium-4 concentrations combined with a Bayesian end-member mixing model. Having identified the groundwater fraction of recently infiltrated river water (Frw) consequently allowed us to infer the travel times from the river to the wellfield, estimated based on radon-222 activities of Frw. Furthermore, we compared tracer-based estimates of Frw with results from a calibrated numerical model. We demonstrate (i) that partitioning of major water sources enables a meaningful interpretation of an age dating tracer of the water fraction of interest and (ii) that the streambed has a major control on the estimated travel times.