The availability, sustainability, and quality of water resources are threatened by many sources, among which the changing climate plays a critical part . A significant sign of climate change is global warming, which has been evidenced by the analysis of long-term air temperature records . Not only the Earth's surface temperature shows a constant warming trend, but the sea surface temperature has also increased . There has been adequate proof that the massive greenhouse gas emissions since the 18th century accelerate the global warming process . Consequently, it is urgently needed to estimate the change of meteorological variables (e.g., precipitation and temperature) in the future global warming scenarios. General circulation models (GCMs) combined with different emission scenarios or representative concentration pathways (RCPs) have been widely employed for climate impact study .

Climate change may significantly alter the pattern of terrestrial hydrological processes, influence the spatial and temporal behavior of shallow water storages, and manipulate the degree and frequency of extreme events such as floods and droughts . Hydrological processes and states (e.g., evapotranspiration, soil moisture, and potential recharge) are tightly coupled with the climate variables (e.g., precipitation, humidity, and atmosphere temperature). The impact of climate change on the terrestrial water cycle is uncertain. Climate model projections show a good consistency in future global averaged trends but may disagree on the magnitude of regional-scale variables, particularly for precipitation projection . Many studies estimate the control and uncertainty of climate change on hydrological states and fluxes . The frequency and intensity of extreme events (e.g., soil moisture droughts and heat waves) may be exacerbated owing to anthropogenic warming . The global water scarcity is likely to be exacerbated due to the potential decline in freshwater resources under the 2 ∘C global warming level .

As the single biggest source of the world's fresh water supply, groundwater plays a critical role in the sustainability of the terrestrial ecosystem and the environmental consequences of climate variability. Globally, groundwater makes up 35 % of the total freshwater withdrawals, constituting approximately 36 %, 27 % and 42 % of water consumption for households, manufacturing, and agriculture, respectively . Although the general knowledge of scale-dependent hydraulic properties of the subsurface hydrologic systems is still quite limited, they prove to be increasingly influenced by anthropogenic factors . The worldwide groundwater system can be affected by climate variability directly by a change in recharge or indirectly by a change in groundwater abstraction . Furthermore, these effects may be adjusted through anthropogenic activities such as land-use change. Many recent studies devoted to evaluating the impact of climate change on groundwater availability . These studies often use coupled climate–land-surface–subsurface models to investigate the potential response of groundwater storages to the outer forcings under different climate scenarios. Compared with the near land-surface fluxes/storages (e.g., soil moisture and evapotranspiration), the groundwater reservoir is less vulnerable to extreme events . The slow response of groundwater to climate variability can be explained by the highly dynamic surface-water–groundwater interaction, the existence of a variably thick unsaturated zone, and the big volume of groundwater storage. The quantification of uncertainty in future water resource projections and travel times (decades to centuries) of the regional groundwater system is critically important for regional water sustainability.

Due to the diverse patterns of the terrestrial water cycle in regions under different climate conditions, climate change will have diverse impacts on the groundwater recharge change. , for instance, found that in Tanzania, a 15 % decline in precipitation, without any change in air temperature, will result in a 40 %–50 % decline of groundwater recharge, indicating a potentially amplified change of recharge compared to that of precipitation. While some studies found an increasing trend of recharge in some regions , others indicate that climate change will likely lead to decreased recharge rates . The changes of recharge, regardless of the sign of change, will significantly influence the groundwater levels and may lead to ecological problems such as the vanishing of wetlands . The modification of groundwater recharge will control the flow paths and travel times of pollutants, which is critical to the sustainability of the regional groundwater system. Moreover, the modification of groundwater recharge can change the age distribution for water in both the vadose zone and the saturated zone, as well as significantly change the composite age distribution .

Groundwater travel time distribution (TTD) is a robust description of the storage and transport dynamics within aquifers under various external forcings. It has many implications for hydrogeological and environmental studies. For instance, significant time lags of the streamflow response to external forcings have been observed by multiple studies . Besides, the legacy pollutants in groundwater reservoirs can have a great impact on the total pollutant loads for agricultural catchments . Groundwater TTD, as a lumped description of the heterogeneous aquifers, sheds light on the assessment of groundwater responses to non-point source contamination subjected to a changing climate and land use .

Although there are plenty of studies that have focused on assessing the impact of future climate change on groundwater recharge , groundwater budget , and surface-water–groundwater exchange , there is an absence of a systematic evaluation of both the groundwater quantity and TTDs under different warming levels that incorporates the uncertainties in both climate projections and hydrological parameterizations. In this study, we analyze the response of groundwater (quantity and TTDs) to the 1.5, 2, and 3 ∘C global warming levels (above the preindustrial levels) in a central German basin (Nägelstedt) using a coupled hydrological model, mHM-OGS mesoscale hydrologic model; OpenGeoSys;. The key questions we aim to answer are the following. (1) How can the flow and transport conditions of a regional groundwater system in future decades differ from the historical period under different warming levels? (2) Can we quantify the degree of different uncertainty sources (e.g., uncertainties in climate projections and groundwater models) and their influences on the resulting groundwater simulations? To answer these questions, we pay particular attention to the assessment of the long-term effect of climate change on the regional groundwater systems considering the buffering effect of groundwater aquifers.

This paper is organized into several sections: Sect.  describes the basic topographical, geometrical, and geological properties of the study area. Section  introduces the methodology and materials for this study. Section  shows the setup and validation for the mHM and OGS models. The simulation results are presented in Sect. . A comprehensive discussion on the simulation results is displayed in Sect. , and the main conclusions are drawn at the end of this section.

As a subbasin of Thuringian basin, Nägelstedt subbasin is located in central Germany, and it has an area of about 850 km2 (Fig. ). It is a headwater catchment of the Unstrut River. The Unstrut River is a typical, meandering lowland river with only a moderate flow velocity under natural conditions. The mountains surrounding Nägelstedt subbasin drain almost simultaneously into the Unstrut during heavy precipitation events, which in the past led to regular, prolonged flooding of large parts of the floodplains. The topographic elevations of this catchment range from 164 m at the southeastern lowland to 516 m in the Hainich mountainous region. This region is classified as a Cfb climate region based on the Köppen–Geiger classification, where Cfb stands for a warm temperate, fully humid, and warm summer climate . It shows a leeward decreasing trend of areal precipitation and rising mean air temperature from the eastern Hainich ridge to the Unstrut Valley . In the larger Thuringian basin, groundwater has been intensively extracted for domestic, industrial, and agricultural uses. About 70 % of the freshwater requirement for Thuringia is satisfied by groundwater .

The extremely fertile soils in the meadows (wet black soil and loess) make Thuringian basin one of the best agricultural basins in Germany. Approximately 88 % of the total land use of this region is regarded as arable land . At the same time, the proportion of woodland and grassland has fallen sharply, leading to an extreme reduction in biodiversity in these areas .

The stratigraphy in this area is characterized by a succession of carbonate–siliciclastic alternations. The main aquifer system consists of several sedimentary rocks, including the Middle Keuper (km), the Lower Keuper (ku), the Upper Muschelkalk (mo), the Middle Muschelkalk (mm), and the Lower Muschelkalk (mu) . The Middle Keuper consists of a marly series with gypsum and dolomite, whereas the Lower Keuper is constituted of grey clays and dolomitic limestone. The Upper Muschelkalk (Hauptmuschelkalk) is mainly made up of shelly limestone, marl, and dolostone. The Middle Muschelkalk consists mainly of evaporites (gypsum, anhydrite, and halite); meanwhile, the Lower Muschelkalk consists of limestone and marls . Karstification occurs in the Muschelkalk formation but has proved to be limited or concentrated in specific zones in this area .

Nägelstedt subbasin is chosen as the study area for the following reasons: (1) it is a typical agricultural basin where potential non-point source contamination may threaten the sustainability and resilience of groundwater, and (2) the critical zone (CZ) in Nägelstedt subbasin has been comprehensively investigated using infrastructure platform from the Collaborative Research Center AquaDiva .

To investigate the impact of different climate change scenarios, we modified the modeling framework from EDgE (End-to-end Demonstrator for improved decision-making in the water sector in Europe) and HOKLIM (High-resolution Climate Indicators for 1.5 Degree Global Warming) projects through coupling it to a three-dimensional subsurface model . Specifically, we use temperature and precipitation derived from five GCMs under three different RCPs to force the mesoscale Hydrologic Model (mHM), aiming to derive the land-surface fluxes and states under different future warming scenarios. The projected recharges from mHM calculations are fed to the groundwater model OpenGeoSys (OGS) for the assessment of groundwater quantity and TTDs.

We designed two parallel numerical experiments to investigate the effect of uncertainties in both the climate and groundwater models on the groundwater resources. For the evaluation of climate uncertainty, 35 GCM/RCP pairs are used, whereas one parameter set (related to hydrogeological features) is used for the groundwater model. In parallel, to assess the parameter uncertainty in the groundwater model, one sole climate realization is used, whereas many realizations of hydraulic-conductivity fields constrained by the observations are used for the groundwater model. Specifically, this climate realization is the ensemble average of all 35 members of GCM/RCP combinations.

In this section, we present the ensemble of simulated changes in groundwater recharges, levels, and TTDs. For the sake of clarity, we use the plus sign to represent simulated values of increases and the minus sign to represent decreases.

In this study, we systematically explore the response of a regional groundwater flow system to different global warming scenarios by means of the sequentially coupled land-surface model (mHM) and a groundwater model (OGS). The results of ensemble simulations manifest that groundwater recharge is likely to increase moderately under all three warming levels in a central German river basin. However, the ensemble simulations do not all agree on the sign of projected changes in groundwater recharge. This is consistent with a previous finding that low flows are expected to increase slightly in this region under future climate scenarios, considering that baseflow is the main component of low flow and recharge feeds the baseflow . A similar increasing trend in climate-induced recharge rates has been suggested for other regions such as a northern European catchment , the high plains of the US , an upper Colorado catchment , and a Snake River basin .

The simulated changes in groundwater levels also manifest similar increases under all three warming levels but show a strong spatial variability depending on the local topography and elevation. These changes can be critical to surface-water–groundwater interaction because the increase or decrease in groundwater table would modify the dynamics of groundwater discharge into streams . In areas with a complex topography and dense drainage network, a rising groundwater level may activate shallow groundwater flow paths and intensify shallow local flow pathways . This way, the mixing behavior of groundwater storage can also change because the activation of shallow flow paths will lead to a stronger systematic preference for discharging young water . Moreover, changes of groundwater levels will impact the land-surface processes such as evapotranspiration, soil moisture, and overland flow .

The influence of climate change on the catchment-scale groundwater TTDs is of critical importance to the sustainability of the groundwater system. Simulated MTTs suggest a moderate decline for the three warming levels, which is not surprising, since the travel time of water parcels is directly controlled by the recharge rate. A critical finding of this study is the nonlinear relationship between the change in MTT and that in the groundwater level, which is mainly attributed to the spatial variability in diffuse recharge change. With weighted-average MTTs being at a centurial timescale, climate-induced variations can significantly affect the long-term sustainability of the regional groundwater system . The projected decrease in groundwater MTTs will remarkably shorten the life span of non-point source pollutants (e.g., nutrient and pesticide) in groundwater aquifers and may introduce substantial changes in the spatiotemporal distributions of pollutant concentrations within the aquifer system. Given that the nutrient budget of the connected surface water body is linked with the groundwater system, the water quality of the surface water body in this region (e.g., the Unstrut River) will respond accordingly in the future, although with a long delay . This observation is in line with many recent studies, wherein they highlight the importance of legacy nutrients in catchments as a reason for long-term catchment response .

One essential topic when assessing future climate impact is to quantify the uncertainties in the projected changes. This study provides original insights into the uncertainty propagation from the outer forcing (associated with climate models) and the internal hydraulic properties (associated with groundwater models) to the groundwater travel times. Among simulations corresponding to different GCM/RCP combinations, simulated changes in hydrologic variables (e.g., recharge, groundwater level, and mean travel time) vary not only in their absolute values but also in sign (positive or negative) because of the large variations in different climate projections. The contribution of a climate model induced to the predictive uncertainty is also found to be greater than that of hydraulic parameters in the groundwater model. Within the current modeling framework, predictive uncertainty and error may also be introduced by other sources, such as the internal variability in climate projection using different initial states, internal parameter uncertainty in the mHM model, and the downscaling algorithm. Enhancements in climate projection and downscaling algorithms can effectively reduce the variability in the projected impacts of global warming on the regional groundwater system. Nevertheless, the dominant source of uncertainty is highly likely due to the climate projections of varying GCMs and RCPs . Except for the above-mentioned climate projection uncertainty, other uncertainty sources have not been assessed in this study. This fact indicates that the range of predictive uncertainty quantified in this study is only a conservative estimate.

A potential limitation of the current modeling framework lies in the one-way coupling approach that does not account for the feedback from groundwater level change to the near-surface processes. The change in the groundwater table can alter the partitioning of water balances, which further exerts a second-order impact on the groundwater level and travel times . A fully coupled system, based on a mixed form of the Richards equation to solve unsaturated and saturated zone flow simultaneously, is more realistic than the one-way coupled system. However, a fully coupled model consistently suffers from an expensive computational burden, which limits its applicability in a large-scale real-world case study. It also introduces extra parameters that are essentially unknown at the catchment scale. The current one-way coupling, although less accurate than the two-way coupling, is computationally more efficient – allowing us to understand the first-order control of climatic variability on groundwater characteristics (groundwater level and travel times). The applied Lagrangian particle tracking in the three-dimensional groundwater model is computationally very expensive. The total computational time performing a single model run is around 14 d using eight cores in a computer cluster facility. Moreover, we have successfully demonstrated the utility of this model for adequately capturing the observed behavior of groundwater levels across the study basin . Consequently, the one-way coupling method used here is a practical choice, allowing us to perform the large ensemble scenarios with reasonable computational resources (and time).

The second potential limitation of this study is the discrepancy between the fine-resolution groundwater model and coarse-resolution mHM simulations. mHM simulations in this study were established within the scope of the EDgE/HOKLIM project, which focuses on the impact of future climate scenarios on European water resources. All databases used for the mHM model setup are on a European scale and typically have coarse spatial resolutions (e.g., 5×5 km2). Although the MPR technique embedded in mHM facilitates the characterization of subgrid-scale features, it does not guarantee that all subgrid-scale features can be captured if the resolutions of input data are too coarse . We note that this is a common problem when utilizing coarse-resolution forcings to drive fine-resolution physically based models. In this respect, simulation results in this study can be considered as first-order approximations based on currently available databases. The conclusions drawn in this study can be tentative and, therefore, open to revision.

The steady-state nature of groundwater simulations used here is reasonable for the assessment of long-term climate impact on regional groundwater system because (1) it reduces the computational burden, (2) the temporal fluctuations under the future climate cannot be reasonably projected, and (3) high-frequency fluctuations in external forcings have minor influences on long-term TTDs . However, analyzing the likely future changes in the transient behavior of groundwater dynamics can be important for many cases where the temporal scale is small and the input forcings are highly dynamic. In recent years, the subject of the transient behavior of TTDs has become more and more prevalent in groundwater hydrology .

We note that the results in this study are only suitable for the Nägelstedt site in central Europe. In other regions of Europe, groundwater recharge change induced by global warming may have behaviors distinct from those shown in this study. For example, some studies indicate a decrease in groundwater quantity in Mediterranean regions due to the decrease in projected precipitation . Besides, baseflow is also expected to decrease, leading to a potential increase in drought in Mediterranean regions .

We only consider the direct impact (i.e., impacts exerted through changed precipitations) of climate change on the regional groundwater system. Interactions between the climate and groundwater are exacerbated by land-use change, which is mainly exerted by the intensification of irrigated agriculture. In South Australia and the southwestern US, the transition from natural catchments to rain-fed cropland significantly alters the groundwater storage through the increase in recharge . These indirect influences of global warming on groundwater systems have not been considered in this study. Such influences can be a dominant factor threatening the local groundwater system for many regions worldwide . Future investigations are needed to incorporate both the direct and indirect impacts of global changes on the sustainability of the regional groundwater system.

To summarize, climate change can significantly alter the quantity and travel time behavior of the regional groundwater system through the modification of recharge, especially at longer timescales. Ensemble simulations indicate remarkable uncertainties in projections of future regional groundwater quantity and travel times, which are introduced primarily by the driving climate projections and secondly by hydrologic/groundwater model parameterizations. In the study domain, moderate absolute changes in recharge rates, groundwater levels, and travel times that are nonlinearly related to the varying level of global warming are found. However, the variability of these changes increases with the warming levels that might also affect the cost of managing the groundwater system. Therefore, it is still advisable to restrain global warming to 1.5 ∘C and avoid global warming of 3 ∘C.