International audienceUsing lumped models and a transfer function model, this paper deals with the interpretation of exceptionally long (up to 50 years (y)) and precise tritium chronicles characterising the rainfall, recharge (efficient rainfall) and outflow from various types of glacial aquifers from the French Alps (Evian-Thonon area). The efficient rainfall tritium chronicle was computed from tritium measurements performed for 11 years (1969-1979) in a lysimeter. The evapotranspiration induces a mean 15% drop of the annual tritium signal. The three superficial glacial aquifers (two fluvio-glacial kame terraces and a lateral till) provide similar results: a best fit with an exponential flow model (EM) (playing the major role) combined in parallel with a piston flow model (PFM), and a rather short mean transit time (T 5-7 y). The deepest mineral aquifer (Evian) can only be fitted with the in a series combination of a highly dispersive model (DM; T 68 y; DP = 0.5) and a piston flow model (T 2.5 y) or, better, by the in a series combination of an EM (T 8 y) modelling the subsurface aquifer and a DM (T 60 y; DP = 0.75) and the same piston flow model (T 2.5 y) modelling the deep mineral aquifer, this latest combination of models providing the following parameters: T 70 y and median transit time 45.5 y. It is also to be noted that a very small part of the recharge; about 1.3%, avoids both the EM and the DM, and directly enters the PFM (at the Northern limit of the Gavot Plateau). These models are very sensitive regarding the T (±1 y, 0.25 y for the PFM), less so with DP. These results will prompt hydrologists to (re)work historical data to determine if important hydrologic information is available. The interest and limits of such a modelling, also for other constituents than tritium, along with the future for tritium as a tracer are discussed and it also provides new insights on the structure and functioning of alpine paleo glacial hydrosystems
Diffuse nitrogen (N) pollution from agriculture in groundwater and surface water is a major challenge in terms of meeting drinking water targets in many parts of Europe. A bottom-up approach involving local stakeholders may be more effective than national- or European-level approaches for addressing local drinking water issues. Common understanding of the causal relationship between agricultural pressure and water quality state, e.g., nitrate pollution among the stakeholders, is necessary to define realistic goals of drinking water protection plans and to motivate the stakeholders; however, it is often challenging to obtain. Therefore, to link agricultural pressure and water quality state, we analyzed lag times between soil surface N surplus and groundwater chemistry using a cross correlation analysis method of three case study sites with groundwater-based drinking water abstraction: Tunø and Aalborg-Drastrup in Denmark and La Voulzie in France. At these sites, various mitigation measures have been implemented since the 1980s at local to national scales, resulting in a decrease of soil surface N surplus, with long-term monitoring data also being available to reveal the water quality responses. The lag times continuously increased with an increasing distance from the N source in Tunø (from 0 to 20 years between 1.2 and 24 m below the land surface; mbls) and La Voulzie (from 8 to 24 years along downstream), while in Aalborg-Drastrup, the lag times showed a greater variability with depth—for instance, 23-year lag time at 9–17 mbls and 4-year lag time at 21–23 mbls. These spatial patterns were interpreted, finding that in Tunø and La Voulzie, matrix flow is the dominant pathway of nitrate, whereas in Aalborg-Drastrup, both matrix and fracture flows are important pathways. The lag times estimated in this study were comparable to groundwater ages measured by chlorofluorocarbons (CFCs); however, they may provide different information to the stakeholders. The lag time may indicate a wait time for detecting the effects of an implemented protection plan while groundwater age, which is the mean residence time of a water body that is a mixture of significantly different ages, may be useful for planning the time scale of water protection programs. We conclude that the lag time may be a useful indicator to reveal the hydrogeological links between the agricultural pressure and water quality state, which is fundamental for a successful implementation of drinking water protection plans.
Interpre´tation des fluctuations pie´zome´triques et des pre´curseurs associe´s au se´isme de magnitude 7,4 du 29 novembre 2007 sur l'ı ˆle de la Martinique (Petites Antilles
The development of groundwater levels (GWL) simulations, based on deep learning (DL) models, is gaining traction due to their success in a wide range of hydrological applications. GWL Simulations allow generating reconstructions to be used for exploring past temporal variability of groundwater resources or provide means to generate projections under climate change on decadal scales. Owing to the diversity of large-scale and local scale forcing factors involved in explaining GWL variability, machine learning or even deep learning approaches reveal relevant tools to simulate GWL. In addition, such methods do not require too much-extended knowledge of physical variables in the links between climate variables and GWL. In this paper, we investigated the capacities of three deep learning models (Long Short-Term Memory (LSTM), Gated Recurrent Unit (GRU) and Bidirectional LSTM (BLSTM)) to reproduce GWL variations over time. Among the three deep learning models, GRU performed relatively better in most cases. Another aspect was to evaluate the input data’s impact and usefulness of wavelet pre-processing considering its limitations and best practices. Two different input datasets are compared to each other, one considering Effective Precipitation only, the other considering Precipitation and Temperature. Maximum Overlap Discrete Wavelet Transform (MODWT) preprocessing was used to decompose the input variables to explore the impact of wavelet transform in improving the simulations on several types of GWL time series by unravelling “hidden” though useful information in input data. Results show that the preprocessing (MODWT) helps the models generate better simulations. This improvement is higher with raw climate data (precipitation & temperature) as compared to when effective precipitation was used as input. Finally, the Shapley Additive exPlanations (SHAP) approach was used to interpret the impact of input variables on the model simulations. Analysis of SHAP values indicated that the sources of the information content preferentially learned by the models to achieve best simulations. For instance, it was clear that simulation of inertial and mixed GWL required the models to learn from low-frequency variability presented in the input data.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.
hi@scite.ai
10624 S. Eastern Ave., Ste. A-614
Henderson, NV 89052, USA
Copyright © 2024 scite LLC. All rights reserved.
Made with 💙 for researchers
Part of the Research Solutions Family.