It is well established that groundwater is an important buffer to hydrologic systems; stabilizing water supplies across spatial scales and long time frames. However, groundwater surface water interactions are non-linear and can vary greatly based on climate and hydrogeologic setting.  This challenge is exacerbated in changing systems where shifting land cover, extreme droughts and floods can significantly change groundwater storage, discharge and recharge dynamics. Observations of groundwater levels and the hydrogeologic properties that govern flow are sparse both in space and time.  As a result, we rely heavily on physically based numerical models to help us understand this critical component of the hydrologic cycle. There are an increasing number of national to global scale groundwater models that take a variety of numerical approaches and simplify the system to varying degrees.  One of the challenges we face is that the integrated models best suited to capture changing dynamics, are also by far the most computationally expensive. This creates a trade-off between the physical complexity we can represent and the size of the ensembles we can explore (another critical dimension in highly uncertain systems). 

            Here we will explore the potential for machine learning emulators to help accelerate solutions while maintaining physically rigorous solutions in changing systems. First, we present progress in the development of the next generation high resolution (1km2) ParFlow model of the contiguous US (ParFlow-CONUS).  Next, we explore a range of machine learning architectures that have been developed to emulate the national model. Here we focus on solutions that can emulate the full 3D subsurface system as this allows for the most flexibility in hydrologic applications.  Specifically, we explore 3D convolutional neural networks, recursive neural networks and LSTM approaches.  For every approach we evaluate the fidelity with which the machine learning model can emulate the physics-based model. We focus specifically on performance for extreme hydrologic conditions both when the ML emulator is provided training data on these cases and when the ML model is applied to out of sample scenarios.  One of the key strengths of physically based hydrologic models is their ability to represent scenarios that we haven’t seen in the past. This can be a large challenge for purely data driven ML approaches which can provide erroneous results on conditions that fall outside historical behavior.  

How to cite: Condon, L., Bennett, A., Tran, H., Horowitz, B., Leonarduzzi, E., Melchior, P., and Maxwell, R.: Accelerating large scale groundwater simulation with machine learning: modeling approaches and science implications for changing systems, EGU General Assembly 2022, Vienna, Austria, 23–27 May 2022, EGU22-8739, https://doi.org/10.5194/egusphere-egu22-8739, 2022.

The increasing population numbers and demand for food has greatly increased the dependence of irrigated crops on groundwater resources. This has resulted in a steep rise of groundwater withdrawal for irrigation around the globe, with a decline of groundwater levels and the potential economic depletion of aquifers as a result. In this presentation we revisit the classic problem of determining economically optimal groundwater withdrawal rates for irrigation. The novelty compared to previous mathematical analyses is the inclusion of non-linear groundwater-surface water interaction that allows for including the impact of capture and the application of this framework at the global scale.

We base our analysis on a recently published analytical framework of groundwater-surface water interaction subject to groundwater pumping (Bierkens et al., 2021). This framework distinguishes between two regimes: 1. a physically stable withdrawal regime, for which groundwater withdrawal q is smaller than a critical withdrawal rate q_crit. Here, groundwater level decline reaches an equilibrium and all groundwater withdrawal eventually comes out of capture; 2. a physically non-stable regime (q > q_crit) where groundwater withdrawal is larger than maximum capture and leads to persistent groundwater level decline. Using a simple hydroeconomic model based on competition of resources, we derive an equation for the optimal withdrawal rate under the stable regime. Similarly, we use the hydroeconomic model to derive economically optimal withdrawal and depletion trajectories for the non-stable regime assuming either full competition or optimal control (intertemporal efficiency). The expressions derived for optimal depletion trajectories under the non-stable regime are a generalization of the work of Gisser and Sánchez (1980), by including (non-linear) groundwater-surface water interaction.

We apply the hydroeconomic framework at the global scale, limited to regions with significant groundwater use for irrigation. For the regions with stable groundwater withdrawal (q<q_crit) we determine the optimal withdrawal rate q_opt and check whether it is attainable in the stable regime (q_opt < q_crit). We also assess the economic gain that can be achieved when the current withdrawal is set equal to q_opt. For regions with non-stable groundwater withdrawal (q>q_crit) we estimate the final groundwater level decline and associated net present value (NPV) of accumulated profits over time, and compare these between competition and optimal control. This allows us to assess, at first order, globally where the so-called Gisser-Sánchez effect olds, in that competition and optimal control lead to similar depletion rates and economic value. Finally, we use the hydroeconomic framework to assess for regions with non-stable groundwater withdrawal (q>q_crit) whether it is more profitable (in the long run) to pursue controlled depletion or reduce withdrawal rates to the stable regime.

How to cite: Bierkens, M., van Beek, R. L. P. H., and Wanders, N.: Revisiting optimal groundwater withdrawal under irrigation: including groundwater-surface water interaction and global analyses, EGU General Assembly 2022, Vienna, Austria, 23–27 May 2022, EGU22-989, https://doi.org/10.5194/egusphere-egu22-989, 2022.

Growing water demands in the Mediterranean region have increased groundwater exploitation, imposing urgent and efficient groundwater management. Sustainable management requires a proper understanding of groundwater status and accurate estimates of groundwater levels with less uncertainty. In this context, large-scale modelling has been shown to assess groundwater resources under changing conditions, especially in regions known for data scarcity. This study aims to quantify the steady-state groundwater levels at continental scales using a model ensemble and in-situ groundwater observations. To test the models' applicability and validity, we utilize one of the most monitored groundwater systems in the Mediterranean region, the Iberian Peninsula. Outputs of three global gradient-based groundwater models (Reinecke et al. (2019), de Graaf et al. (2017), and Fan et al. (2013)) were compared to observations from long-term groundwater monitoring network. The model ensemble showed reasonable performance in replicating the groundwater levels for shallow groundwater, but performance deteriorated with increased elevation.

In this study, we argue that we can develop continental scale groundwater maps for groundwater assessment by combining model results with in-situ data. Historical groundwater levels were used to test, train and validate the different combination methods. Here we present the outcomes and discuss the accuracy of the final product. We see this study as a benchmark approach of using multi-model ensemble and observations to deliver better groundwater steady-state conditions as a baseline for groundwater users and managers in the Mediterranean region.

This work was supported by the German Federal Ministry of Education and Research (BMBF, Germany, Grant 01DH19015) under the Project Sustain-COAST, co-funded by EU PRIMA 2018 programmes.

How to cite: Ben-Salem, N., Reinecke, R., P. Karatzas, G., Rode, M., and Jomaa, S.: Combining a global groundwater model ensemble with in-situ data for groundwater assessment in the Mediterranean region , EGU General Assembly 2022, Vienna, Austria, 23–27 May 2022, EGU22-9196, https://doi.org/10.5194/egusphere-egu22-9196, 2022.

Groundwater is the primary drinking water supply of billions of people worldwide. While groundwater is under pressure globally due to extensive water abstractions, proximity to coasts amplifies these pressures due to potential sea water intrusion that can endanger groundwater quality. It is unclear how climate change (changing potential groundwater recharge), as well as rising sea levels, will alter coastal groundwater dynamics, i.e., submarine groundwater discharge and seawater intrusion.

Various factors impact coastal groundwater dynamics, including groundwater recharge & extraction, hydraulic gradients, permeabilities, water densities, and oceanic activity (e.g., tidal pumping and wave setup). It is currently unclear how much these different factors control submarine fluxes along global coastlines. We developed perceptual models of coastal groundwater fluxes based on a literature review of regional and global models. Here we present our perceptual model and discuss it in the context of currently available global data, uncertainties, climate change, and whether it can be implemented with an existing Open Source global groundwater modeling framework (G³M-f).

How to cite: Kretschmer, D., Reinecke, R., Moosdorf, N., Michael, H., and Wagener, T.: Understanding coastal groundwater processes in a changing climate: A perceptual model of global-scale coastal groundwater dynamics, EGU General Assembly 2022, Vienna, Austria, 23–27 May 2022, EGU22-7057, https://doi.org/10.5194/egusphere-egu22-7057, 2022.

The Global Gravity-based Groundwater Product (G3P) aims at developing a satellite-based groundwater storage (GW) data set as a new product for the EU Copernicus Climate Change Service. As the world’s largest distributed freshwater storage, GW is a key resource for mankind, industrial, and agricultural demands. In Copernicus, there is no service available yet to deliver data on this fundamental resource, nor is there any other data source worldwide that operationally provides information on changing groundwater resources in a consistent way, observation-based, and with global coverage. Therefore, G3P develops an operational global groundwater service as a cross-cutting extension of the existing Copernicus portfolio. G3P capitalizes from the unique capability of GRACE and GRACE-FO satellite gravimetry as the only remote sensing technology to monitor subsurface mass variations, and from other satellite-based water storage products to provide a data set of groundwater storage change for large areas with global coverage. G3P is obtained by using a mass balance approach, i.e., by subtracting satellite-based water storage compartments (WSCs) such as snow water equivalent, root-zone soil moisture, glacier mass, and surface water storage from GRACE/GRACE-FO monthly terrestrial water storage anomalies (TWSA). For a consistent subtraction of all individual WSCs from GRACE-TWSA, the individual WSCs are filtered in a similar way as GRACE-TWSA, where optimal filter types were derived by analyses of spatial correlation patterns. G3P groundwater variations are provided for almost two decades (from 2002 to the present), with the monthly resolution, and at a 0.5-degree spatial resolution globally. In this contribution, we also illustrate preliminary results of the G3P data set and of its uncertainties, as well as its evaluation by independent groundwater data.

This study has received funding from the European Union’s Horizon 2020 research and innovation programme for G3P (Global Gravity-based Groundwater Product) under grant agreement nº 870353.

How to cite: Sharifi, E. and Güntner, A. and the G3P team: The Global Gravity-based Groundwater Product (G3P): first results , EGU General Assembly 2022, Vienna, Austria, 23–27 May 2022, EGU22-1659, https://doi.org/10.5194/egusphere-egu22-1659, 2022.

How much precipitation recharges groundwaters varies enormously across Earth's surface, but recharge rates are uncertain because field observations are sparse and modeled global estimates remain largely unvalidated. Here we show that annual recharge is predictable as a simple function of climatic aridity — the ratio of long-term potential evapotranspiration to precipitation — using a global synthesis of measured recharge of 5237 sites across six continents. We use this relationship to estimate long-term recharge globally outside of permafrost regions. Our estimates double previous global hydrological model estimates and are more consistent with empirical field observations. These revised higher estimates of global groundwater recharge imply that groundwater contributes more actively to evapotranspiration and streamflow than previously represented in global water cycle depictions or global hydrological and Earth system models. In addition, we quantify the sensitivity of groundwater recharge to changes in aridity using the empirical relationship between groundwater recharge rates and climatic aridity. This analysis indicates that recharge is most sensitive to climate aridity in mesic regions, where changes in the replenishment of aquifers will be amplified relative to projected changes in precipitation. Global hydrological models seem to underestimate changes in recharge with climate aridity. Thus, the impacts of climatic changes on the replenishment of Earth's largest liquid freshwater stores may be larger than previously anticipated.

How to cite: Berghuijs, W., Allen, S., Jasechko, S., Moeck, C., Luijendijk, E., and van der Velde, Y.: A more active role for groundwater in the land water cycle, EGU General Assembly 2022, Vienna, Austria, 23–27 May 2022, EGU22-1009, https://doi.org/10.5194/egusphere-egu22-1009, 2022.

Groundwater recharge quantification is essential for sustainable groundwater resources management, but typically limited to local and regional scale estimates. A high-resolution (1 km x 1 km) dataset consisting of long-term average actual evapotranspiration, effective precipitation, a groundwater recharge coefficient, and the resulting groundwater recharge map has been created for all of Europe using a variety of pan-European datasets and seven national gridded recharge estimates. As an initial step, the approach developed for continental scale mapping consists of a merged estimate of actual evapotranspiration originating from satellite data and the vegetation controlled Budyko approach to subsequently estimate effective precipitation.  Secondly, a machine learning model based on the Random Forest regressor was developed for mapping groundwater recharge coefficients, using a range of covariates related to geology, soil, topography and climate. A common feature of the approach is the validation and training against effective precipitation, recharge coefficients and groundwater recharge from seven national gridded datasets covering the UK, Ireland, Finland, Denmark, the Netherlands, France and Spain, representing a wide range of climatic and hydrogeological conditions across Europe.  The groundwater recharge map provides harmonised high-resolution estimates across Europe and locally relevant estimates for areas where this information is otherwise not available, while being consistent with the existing national gridded estimates. The Pan-European groundwater recharge pattern compares well with results from the global hydrological model PCR-GLOBWB 2. At country scale, the results were compared to a German recharge map showing great similarity. The full dataset of long-term average actual evapotranspiration, effective precipitation, recharge coefficients and groundwater recharge is available through the EuroGeoSurveys’ open access European Geological Data Infrastructure (EGDI).

How to cite: Stisen, S., Martinsen, G., Bessiere, H., Caballero, Y., Koch, J., Juan Collados-Lara, A., Mansour, M., Sallasmaa, O., Pulido Velázquez, D., Hunter Williams, N., and Jan Zaadnoordijk, W.: Pan-European high-resolution groundwater recharge mapping – combining satellite data and national survey data using machine learning, EGU General Assembly 2022, Vienna, Austria, 23–27 May 2022, EGU22-4158, https://doi.org/10.5194/egusphere-egu22-4158, 2022.