The safety case for a deep geological repository in Switzerland: assessment and modelling of the biosphere

1 Introduction

Radioactive waste disposed of in a deep geological repository contains long-lived radionuclides which, over geological timescales, may eventually reach aquifers and from there migrate with groundwater to eventually reach the surface environment (ICRP, 2007; ICRP, 2013). To assess the potential radiological consequences for future human populations, a stylised biosphere system is defined and modelled. This system describes the pathways through which radionuclides from the repository may be released to, and transported within, surface environments (the biosphere) and may lead to human exposure via locally produced food, ingestion of groundwater, inhalation, or external irradiation. Swiss legal guidelines require the quantification of the annual effective dose to a representative individual from the most affected population group under conservative but credible assumptions (ENSI, 2020). Within Nagra’s post-closure safety case for the General Licence Application, the biosphere assessment forms a key element of the radiological consequence analysis and is used to derive biosphere dose conversion factors (BDCFs) that convert radionuclide release rates from the geosphere to individual dose rates (Figure 1). The long timescales required for potential releases from a deep geological repository in the Opalinus Clay to reach the surface, and the slow variation of these releases with time, permit the decoupling of the biosphere model from the rest of the system and its models and justify the use of equilibrium BDCFs in calculations of radiological consequences.

Nagra’s biosphere modelling has been refined over several decades, starting with the Terrestrial-Aquatic Model of the Environment (TAME) in the 1990s (Klos et al., 1996). This has been successively iterated though the disposal feasibility study of “Project Opalinus Clay” (Nagra, 2002) and the multiple stages of the Swiss Federal Office of Energy’s Sectoral Plan for site selection (Brennwald and van Dorp, 2008; Nagra, 2008; Nagra, 2010; Nagra, 2014; Nagra, 2024f). The iterative process has benefitted from international collaborative efforts to develop and refine guidance on biosphere assessment, for example as part of the MODARIA programme of the IAEA ¹ (Nagra, 2024f), and via the BIOPROTA forum ² . These efforts have sought to build consensus, and thus confidence, in the methodology for biosphere assessment, and have led to the internationally accepted practices embodied in the IAEA’s enhanced BIOMASS methodology (Lindborg et al., 2022; IAEA, 2025). Indeed, the work to update and enhance the original BIOMASS guidance included extensive review of international biosphere assessments and established a common framework for biosphere assessment, whilst allowing appropriate site-specific adaption within the national contexts and legal requirements (IAEA, 2003; Lindborg et al., 2022; IAEA, 2025). Thus, a number of national disposal programmes have implemented comparable biosphere assessment approaches in site-specific safety cases, including Sweden (SKB, 2011; SKB, 2015), Finland (Posiva, 2012; Posiva, 2013), France (ANDRA, 2005; ANDRA, 2015), Canada (NWMO, 2017; NWMO, 2018), the United Kingdom (NDA, 2016; RWM, 2017), and the United States (U.S. DOE, 2008). Nagra’s approach is, therefore, consistent with international best practice, and with the national guidelines of ENSI (the Swiss Federal Nuclear Safety Inspectorate) (ENSI, 2018; ENSI, 2020). As such, Nagra’s biosphere assessment model is integrated within the wider assessment context, including the latest details of the present geological situation (summarised within this special issue in Wüst et al., 2026), and understanding from recent hydrogeological, morphological and climate evolution studies (see Nagra, 2024c for a summary). This scientific foundation provides the basis of a consistent and structured framework within which potential radionuclide releases can be evaluated with confidence over assessment periods of up to one million years.

This paper provides an overview of the biosphere system for Northern Switzerland (Figure 1), including the boundary conditions and exposure pathways, and illustrates how radionuclide behaviour in sediments, soil, groundwater and surface water are incorporated into a narrative description and translated into conceptual and mathematical models. The schematic overview in Figure 1 illustrates how potential radionuclide fluxes from the repository and geosphere models (bottom left box, not part of the biosphere assessment) enter the biosphere model, where transport through environmental compartments (local aquifer, soils, surface water and atmosphere) is represented using the conceptual and mathematical framework described in this article (Figure 1 top box, and summarised in Sections 3, 4). Exposure pathway modelling links these environmental concentrations to human intake and irradiation pathways using habits and dose coefficients, allowing the calculation of biosphere dose conversion factors (BDCFs) for a reference and alternative biospheres (Figure 1 bottom middle box, and summarised in Section 5). As shown in the bottom half of Figure 1, these BDCFs are then used in the radiological dose consequence analysis to derive annual individual dose rates (Figure 1 bottom right box, not part of biopshere assessment) directly from the radionuclide fluxes calculated in the repository and geosphere modelling (see Li, 2026). These calculated dose rates can then be compared with the regulatory dose constraint of 0.1 mSv/h for the assessment period (ENSI, 2020).

According to the Swiss regulator, assumptions made for the biosphere modelling must reflect credible environmental conditions and human behaviour, including the possibility of a self-sufficient population living under future (including alternative) climate conditions (ENSI, 2020). The biosphere model must therefore incorporate both current conditions in Northern Switzerland (reference biosphere) and a range of alternative conditions that account for long-term geomorphological and climatic changes over a period of up to one million years. A present-day valley in northern Switzerland characterised by a temperate climate is used to represent reference biosphere conditions and is evaluated along with three bounding alternatives: a warmer, drier climate, a cold climate case with permafrost, and a drained farmland system with shallow groundwater. Understanding of such alternative biosphere conditions is drawn from consideration of climate and landscape evolution of the past and projections of future conditions.

Consistent with international best practice, the stylised approach that is applied aims to be realistic in the representation of radionuclide behaviour in the biosphere, whilst using pessimistic assumptions regarding potential human exposure pathways (Lindborg et al., 2022; IAEA, 2025). This ensures that the potential significance of any radionuclide releases to the biosphere are not underestimated whilst simultaneously avoiding an overly pessimistic assessment. Combining this approach with the use of alternative biosphere cases, provides a robust envelope for uncertainties in long-term climate, geomorphology and hydrological conditions that may impact potential future radiological exposure.

Deterministic and probabilistic sensitivity analyses were also performed to further address the impact of uncertainties in site characteristics, lifestyles and parameters relevant to mass transport. Collectively, these aimed to strengthen system understanding and build confidence in the robustness and comprehensiveness of the modelling approach described in this paper.

2 Establishment of a biosphere for Northern Switzerland (narrative description)

2.1 Landscape characterisation, future evolution, and implications for radionuclide release

The current landscape of Northern Switzerland is characterised by a low-relief cuesta morphology, with gently inclined layers of more and less erodible strata. Broad fluvial valleys with extensive terrace systems evolved because of fluvial and glacial processes over the last several hundred thousand years. Valleys typically contain glaciofluvial sediments over bedrock that form flat valley floors with incised rivers and lakes; both hillslopes and valley bottoms are generally well drained (Figure 2). Locally, limestone bedrock crops out in areas such as the Randen near Schaffhausen and the Tabular Jura towards the west may show karstic phenomena, with shallow aquifers and rapid drainage to deeper systems. These areas represent variants within the general valley landscape distinguished by scale and flow characteristics and were integrated into Nagra’s model during the last assessment (Haldimann and Schatzmann, 2014).

Geomorphological process rates in northern Switzerland are generally low during interglacial periods and catchment denudation rates are lowest in low-relief areas, including the Randen valleys (0.05–0.1 mm/a), where chemical weathering dominates (Ott et al., 2025). On the valley flanks, rates may double reflecting more active hillslope processes, with erosion and weathering contributing equally. The highest rates (0.3–0.5 mm/a) occur in an isolated case (Wutach gorge), where rapid physical erosion reflects transient adjustment following a post-glacial river capture (Ott et al., 2025). This pattern is consistent with the preservation of fluvioglacial deposits. For example, while 1-2 million-year-old terrace deposits occur mainly as isolated relics, the younger Hochterrasse (<∼190 ka) and Niederterrasse (last glacial) deposits still form continuous gravel bodies with flat morphological surfaces. Such preservation implies landscape rejuvenation timescales on the order of a few hundred thousand years, further supporting the inference of generally low denudation and incision rates (Nagra, 2024b).

In the present hydrological system, deep groundwater discharges mainly to topographic lows, especially valley bottoms and major river valleys such as the Rhine (Figure 2). Beside observations, palaeo-hydrogeological modelling studies (Nagra, 2024d; Gmünder et al., 2014) confirm that discharge pathways from deep aquifers (Malm, Hauptrogenstein, Keuper, Muschelkalk) preferentially emerge in these low-lying gravel-filled valley areas.

Over timescales of one million years, erosion will continue to shape the landscape, primarily through fluvial incision and glacial modification. Erosion rates are expected to be lower than in past glacial cycles, with no major drainage reorganisations and smaller future glaciers, as well as increasing resistance of deeper lithologies below the Molasse (Nagra, 2024b). This will lead to a gradual lowering of the river network, reduced repository overburden thickness, and a progressive shallowing of deep aquifers. As a result, discharge points for deep groundwater may shift closer to the repository area, and flow systems may become more active with shorter discharge pathways in aquifers such as the Malm (Nagra, 2024c).

These changes are relevant for radionuclide migration because they determine where, and how, deep groundwater potentially carrying radionuclides may discharge to the biosphere. However, radiological consequence analysis conservatively assumes instantaneous radionuclide release from the Malm/Keuper/Muschelkalk aquifers to the biosphere, independent of actual transport path lengths and associated dilution and dispersion. Consequently, potential shortening of discharge pathways would not affect calculated radiological impacts. Furthermore, future landscape conditions are expected to evolve towards configurations comparable to existing limestone valley systems (e.g. Randen and Tabular Jura), which are already included in biosphere characterisation and parameter ranges (e.g. Haldimann and Schatzmann, 2014). In summary, the well characterised morphological and hydrological settings of Northern Switzerland provide a robust analogue for future conditions relevant to radionuclide release scenarios. Anticipated landscape changes over the assessment period are gradual and within the range of present-day variability, and their hydrological consequences are included in existing biosphere models and sensitivity analyses.

2.2 Paleoclimate and future climate scenarios

The Quaternary climate history of Northern Switzerland closely followed global climate cycles, driven largely by Milankovitch orbital variations (Jouzel et al., 2007). As a result, over the past 800,000 years, the region experienced repeated glacial-interglacial cycles roughly every 100,000 years (Nagra, 2024a), with interglacial periods typically lasting several thousand years. Over the next million years, climate oscillations are also expected (Nagra, 2024a), but cyclicity intervals may vary and the current warm interglacial period may be extended as a result of anthropogenic global warming, thus delaying the onset of the next glaciation (Talento and Ganopolski, 2021; Kaufhold et al., 2025; Nagra, 2024a; Fischer et al., 2021). Current interglacial conditions provide a natural reference and have been used in Nagra’s past assessments to define the reference biosphere for radiological consequence analysis (Nagra, 2014; Nagra, 2010; Brennwald and van Dorp, 2008; Klos et al., 1996). Present-day climate in Northern Switzerland lies at the transition between temperate and continental zones (Cfb/Dfb in the Köppen-Geiger classification; Beck et al., 2018; Figure 3), with year-round precipitation, warm summers and cold winters featuring periodic snowfall and thaw.

Paleoclimate records and climate simulations foresee temperatures above current levels, and a warmer, drier climate is, therefore, considered a bounding alternative for biosphere assessment, in line with previous assessments (Brennwald and van Dorp, 2008; Nagra, 2010; Nagra, 2014). Such a scenario would feature lower precipitation in spring and summer, and higher average temperatures year-round, as currently experienced by parts of Spain (Csa/Csb in the Köppen-Geiger classification; Beck et al., 2018; Figure 3).

Past biosphere assessments by Nagra have also included cold climate biospheres (Nagra, 2010; Nagra, 2014), which are retained here as bounding alternatives. Such climates could occur during transitions between interglacial and glacial periods, with conditions similar to those in northern Scandinavia or Canada (Dfc in the Köppen-Geiger classification; Beck et al., 2018; Figure 3). These are characterised by coldest-month averages below 0 °C, only 1–3 months above 10 °C, and no distinct dry season. Most precipitation falls as snow, leading to significant snowpack, and some rivers may experience freezing in winter. Spring thaw triggers an active hydrogeological phase with high surface and groundwater flow. Such conditions remain compatible with human habitation and local food production, whereas, under full glacial conditions, humans are unlikely to be present in northern Switzerland (e.g., due to ice cover), and the presence of ice in the lowlands would restrict groundwater recharge.

3 Biosphere characterisation

Nagra’s biosphere model for the General Licence Application is based on conditions and landscape evolution in Northern Switzerland and integrates the current understanding of regional climate, topography, lithostratigraphy, and hydrogeology, while considering human lifestyles largely reliant on local environment resources and land-use. The reference biosphere represents current conditions in northern Switzerland, because temperate, interglacial conditions allow for maximum use of the local environment. As such, it is used as an analogue for potential future situations in which radionuclides released from a deep geological repository could enter the near-surface environment under climatic conditions comparable to those of today. Thus, the focus is on a population group living in this local region who would be primarily affected by the potential release of radionuclides in the long term.

Uncertainty in biosphere conditions at the time of radionuclide releases in the distant future are bounded by the alternative biosphere cases that address different climatic and morphological contexts. In the radiological consequence analyses performed in support of the safety assessment (Nagra, 2024g), the BDCFs associated with each biosphere case are applied throughout the whole assessment period of one million years for high-level waste. This approach is deemed reasonable given the uncertainty in biosphere evolution and the long time scales involved in the transport of radionuclides to the near surface. As such, the application of alternative biospheres in separate assessment cases acts to envelope this uncertainty.

The narrative descriptions of these biospheres and their conceptualisations for modelling are provided in the following sections. As part of the wider post-closure safety case, separate biospheres for hypothetical “what-if?” cases addressing excavation of the repository by erosion processes (Nagra, 2024h) and future human action safety scenarios (Nagra, 2024i) were also developed, but are not described here.

3.1 Reference biosphere description and conceptual model

Regional hydrogeological modelling (Nagra, 2024d) indicates that, in the siting regions considered in the third and final stage of the repository siting programme, deep groundwater generally discharges into larger river valleys, and primarily into the Rhine River in the north of the country. Accordingly, the reference biosphere represents a valley section where deep groundwater potentially transporting radionuclides may enter a local near-surface aquifer within the unconsolidated Quaternary glaciofluvial deposits (Figure 2). The reference biosphere covers an area of a few square kilometers, the extent of which is informed by the spatial extent of this shallow aquifer and by the area required to sustain a self-sufficient community (Figure 4). Here, the valley floor is covered by agricultural soils and sediments that interact with the aquifers. Valley sides supply uncontaminated groundwater and sediments to the valley floor but do not overlie the shallow aquifer and are, therefore, not directly exposed to potential contamination. In the reference biosphere, a temperate climate with warm summers and cold winters including periods of snow cover and spring melt is assumed.

Surface water is represented by a river comparable in scale to the Rhine River, while lakes are conservatively disregarded due to the substantial dilution that they would provide. Groundwater flow in the shallow aquifer is dominated by along-valley transport. The water table fluctuates seasonally and is hydraulically connected to the river, allowing discharge of groundwater to surface water. In addition, the agricultural soils have low organic carbon contents, are slightly acidic, well-drained and are thus suitable for a wide range of agricultural uses. This allows a broad range of exposure pathways to be represented, including crops, pasture and livestock, along with fish consumption from the river. Non-agricultural plants and animals are included insofar as they influence radionuclide transport and distribution. The underlying deeper unsaturated soil horizon transitions into variably saturated sediments of the aquifer.

A self-sufficient rural community is assumed to have a population density of ∼100 people/km ² , resulting in a few hundred inhabitants within the biosphere area. To ensure a pessimistic assessment, maximal use of local resources is assumed, including food production and abstraction of groundwater from the shallow aquifer for drinking water and irrigation.

3.2 Alternative biosphere descriptions and conceptual models

Uncertainty in biosphere evolution over long timescales is addressed through the consideration of alternative biosphere conditions. In all cases, radionuclides are assumed to discharge in groundwater to a shallow aquifer in a valley setting in northern Switzerland, and each alternative biosphere is defined by a distinct conceptual and mathematical model (Nagra, 2024f). These differences give rise to individual sets of biosphere dose conversion factors used in the assessment of radiological consequences. The alternative biospheres include both climate (warmer–drier and cold) and morphological (drained farmland) differences to explore associated uncertainties relevant to radionuclide transport and exposure.

The warmer–drier climate case represents conditions analogous to Mediterranean-type climates (Csa/Csb in Figure 3) with higher temperatures and reduced precipitation (i.e., lower groundwater and surface-water flow rates) resulting in reduced dilution of radionuclides. These conditions demand greater irrigation from the shallow aquifer and its associated radionuclide source term. The range of processes and potential exposure pathways is identical to the reference biosphere. Differences between the cases are therefore captured primarily through parameterisation (i.e., hydrogeological and water-use parameters).

The cold climate case represents subarctic or periglacial conditions (Dfc climate, Figure 3), comparable to parts of northern Scandinavia and Canada. These conditions include long, cold winters, a short growing season, substantial snowfall, and spring thaw events associated with high surface-water and groundwater flows. Human habitation, livestock herding, and limited food production are assumed to remain feasible. The geographical setting remains a valley bottom with discharge to a shallow aquifer composed of glaciofluvial sediments. A key conceptual difference from the reference biosphere is the presence of permafrost between the aquifer and surface soils that prevents direct exchange, though gradual weathering and mass erosion over long periods may bring radionuclides to the surface soils. Soils are thin, seasonally frozen, and relatively rich in organic matter due to slow decomposition rates. Agricultural activity is limited to small-scale agriculture during the summer months, potentially supported by greenhouses, while the wider valley is used for grazing and foraging. Surface waters are used for drinking and irrigation. Exposure pathways, therefore, emphasise surface water mediated pathways, ingestion of animal products, limited crop production, wild foods, fish, and external exposure during seasonal agricultural activities. Given the low productivity, a significant proportion of the energy/food requirements would need to be derived from outside the domain of the biosphere model.

The drained farmland case represents a morphological variant in which the groundwater table naturally lies close to the soil surface, as observed in parts of the valley bottoms in present-day northern Switzerland. To enable agriculture, the land is assumed to have been drained via ditches that lower the water table into the deep soil, whilst maintaining generally wetter conditions than in the reference biosphere. The overall biosphere structure remains the same as in the reference case, but hydrological connectivity differs. The variably saturated zone of the reference biosphere is fully saturated and incorporated into the shallow aquifer. Higher soil moisture and historical wetland conditions result in elevated organic matter content in the topsoil, where water is transferred from soils to surface water via drainage systems, and periodic flooding allows river sediments to be deposited on land. Agriculture does not require irrigation, though production is restricted to green vegetables, root vegetables, and pasture, with cereals and fruit assumed to be produced elsewhere. Human drinking water is obtained from groundwater, which is readily accessible due to the shallow water table, whereas livestock consume surface water.

4 Abstraction and modelling

The narrative descriptions are translated into a conceptual model that considers all the features, events and processes (FEPs) that influence radionuclide transport and exposure pathways within the biosphere. Following this, mathematical models (a compartment transport model and an exposure pathway model) are developed and supported with data to determine the BDCFs for quantitative dose assessment. In a first step, the principal features of the biosphere are defined (i.e., soils, unconsolidated sediments, aquifer type, river, atmosphere, crops, livestock and humans) and key hydrological processes and mass transfers identified (illustrated with arrows in Figure 4). These determine how radionuclides entering the system via groundwater discharge from depth are redistributed through hydrological, geochemical, geomorphological and biological processes, and how they may leave the region that is modelled via outflow of water and sediments. An interaction-matrix approach (IAEA, 2025) was used to ensure explicit representation of all the relevant events and processes operating between the biosphere features. The resulting conceptual model was also reviewed against Nagra’s FEP database (Nagra, 2024e) to ensure consistency and build confidence that all important FEPs were appropriately represented.

The compartment model of the biosphere is based on the identified FEPs (Figure 5a), where each important feature is represented by a single compartment, permitting the relevant water and mass transfers to be tracked throughout the system. These transport processes are expressed using differential equations, with the calculations performed using the SwiBAC model (Nagra, 2024f), which employs AMBER as the underlying computational code (Quintessa Ltd., 2025). Parameterisation of the model was reviewed and updated for the General Licence Application, including consideration of previous studies, published literature and databases (e.g., flow rates, precipitation rates, and sorption coefficients). The model allows time-dependent radionuclide concentrations to be calculated, which provide the basis for the exposure pathway model (Figure 5b).

A similar procedure is followed for identification and representation of potential exposure routes in the exposure pathway model (Figure 5b). Consistent with the principle of conservatism, the model considers a representative individual from the self-sufficient community who makes maximal use of the local food and water and resides in the region of highest radionuclide concentrations throughout the entire year. The exposure pathways considered address contaminated food and water intake, inhalation of airborne radionuclides and exposure via external irradiation. The exposure pathways are parameterised consistent with a self-sufficient lifestyle, drawing on generalised habit data. Compartment concentrations (for external irradiation) and the mass of each radionuclide reaching the exposed individual (for ingestion and inhalation) are converted to dose through the application of dose coefficients. For ingestion and inhalation, recommended values are available in the Swiss Radiological Protection Ordinance (ROP, 2017), which are based on those of the International Commission on Radiological Protection (ICRP, 2012; ICRP, 1996). Dose coefficients for external irradiation are drawn from US EPA guidance (EPA, 2019). Equilibrium results for a constant unit input of each radionuclide reaching the local aquifer in groundwater from depth provide the basis for the BDCFs. Radionuclides with half-lives greater than ∼60 days are explicitly represented in the biosphere modelling, along with their decay chains. This includes the contributions from short-lived progeny, especially those whose half-lives are short compared to biosphere transport processes. Thus, they can be assumed to be present in secular equilibrium with longer-lived parent consistent with any associated branching ratios.

Whilst most of the radionuclides modelled are radioisotopes of trace elements in the biosphere and can be represented using the same modelling approach, C-14 is an exception due to its fundamental role in biogeochemical processes. Thus, its transport in the biosphere and plants requires a bespoke modelling approach, which is documented in Nagra (2024j), Walke and Thorne (2023).

5 Assessment results

Calculated BDCFs for a selection of representative radionuclides in the reference biosphere are shown in Figure 6. The reference values vary over several orders of magnitude, reflecting the differing half-lives, mobilities and radiosensitivities. Additionally, the time taken for each radionuclide to reach equilibrium varies considerably, e.g., 1.3 years for Po-210 and 450,000 years for Pu-244, primarily related to their propensity to sorb (Nagra, 2024f). Thus, use of equilibrium values throughout the duration of the assessment period contributes to the overall conservative approach.

The contribution of each exposure pathway to the reference BDCFs for the same selection of radionuclides is shown in Figure 7. Whilst most pathways are represented across the selected radionuclides to some degree, a strong radionuclide dependency is evident, which reflects the differences in sorption/retention, bioavailability and relative radiotoxicity of each radionuclide. Milk ingestion, for example, represents a dominant pathway, particularly for Cl-36, Ca-41 and I-129, where it accounts for 80%–90% of the total BDCF. For Ag-108m, the large contribution of external irradiation from soil relates to its relatively high external dose coefficient. Beef consumption is particularly important for Se-79 exposure, whilst grain consumption is well represented across most radionuclides. Well water used for drinking tends to be important for C-14 and many of the actinide series.

6 Management of uncertainty in biosphere modelling

There is inherent uncertainty in the modelling of radionuclide transport in the biosphere related to both its conceptualisation and parameterisation, but particularly related to long term evolution of the biosphere. Due to the complexity of the system and the long timescales involved, a deterministic approach was adopted to explore uncertainty associated with the characteristics of the biosphere in the long term. Here, the sensitivity of the results to differences in climate and morphology has been investigated through alternative biosphere cases (Section 3.2). BDCFs for the warmer-drier, cold climate and drained farmland cases are compared in Figure 8a with respect to the reference biosphere case for the radionuclides considered above.

The BDCFs for the warmer-drier climate case are on average three times higher than in the reference case (considering the whole set of modelled radionuclides), and a maximum of an order of magnitude higher for a few radionuclides. These results reflect the effects of lower water flow rates in the shallow aquifer (i.e., less dilution), and increased use of well water for irrigation. Hence, relatively mobile radionuclides, such as Cl-36 and Ca-41, are most impacted by such conditions compared to radionuclides that are readily retained on solids.

BDCFs for the cold climate case are typically between one and four orders of magnitude lower than in the reference case, in spite of notably conservative assumptions related to local food consumption under these conditions. The most significant changes are associated with radionuclides for which ingestion of crops and animal produce are important, such as Cl-36 (see Figure 9). These differences reflect the reduced irrigation of crops, including the absence of pasture irrigation, and the lower contribution of local food to diet.

For the drained farmland case, most BDCFs are lower than the reference case by 10%–75%, though even lower examples can be found in the subset of the radionuclides displayed in Figure 8a. This effect is primarily related to the lack of irrigation with water from the potentially contaminated shallow aquifer, but also due to a reduction in locally grown food.

In addition to the various alternative biosphere cases, deterministic sensitivity analyses were also performed to address the potential impact of uncertainties related to site characteristics and human lifestyles. The key differences in the conceptualisations and the resulting impact on the BDCFs are discussed here, and details of the explored parameters can be found in Nagra (2024f).

A smaller biosphere was used to explore sensitivity to the size of the modelled area (0.1 km² compared with ∼2.3 km² in the reference case). As shown in Figure 8b, BDCFs for the selected radionuclides are typically one to two orders of magnitude higher than the reference case. However, it should be noted that the BDCFs continue to assume that all the potentially contaminated deep groundwater discharges to this significantly smaller valley, thus resulting in a very pessimistic assessment. In contrast, a larger biosphere area (3 km²), with larger water bodies and associated water fluxes, results in BDCFs that are typically a factor of up to six lower than the reference case. The differences in BDCFs between these two cases provide insights into the sensitivity to biosphere area and hydrology.

In a further case, the radionuclides in deeper groundwater were discharged directly into the river, as opposed to being released to the local aquifer; this case explored the effect of dilution in surface water and a reduction in retention and radioactive decay processes in the aquifer. In this case, all water was assumed to be drawn from the river rather than the local aquifer. BDCFs for most radionuclides are one to five orders of magnitude lower than in the reference case, which is explained by the increased dilution. Potential exposure via fish consumption becomes a more significant pathway for all radionuclides in this case.

Finally, a case considering more realistic lifestyle assumptions was also formulated to examine the degree of pessimism introduced in the reference case by conservatively assuming that consumed food is locally produced. In this case, 50% of the dairy and poultry consumed by the group and 20% of their beef, grain, fruit and vegetables were taken from the local area (as opposed to 100% in the reference case). In addition, the group was assumed to spend two thirds of their time in the local area (as opposed to 100% of their time in the reference case). These changes result in BDCFs that are typically one to five times lower than in the reference case, with radionuclides that are strongly linked to beef and grain consumption (such as Se-79) being the most impacted by these changes. This clearly demonstrates the pessimism introduced through the assumption of self-sufficiency in the local population.

Probabilistic sensitivity analyses were employed to explore sensitivity to parameters that were not explored via the alternative biospheres and the deterministic sensitivity cases. These analyses aimed to identify important processes and parameters to increase system understanding and build confidence in the modelling, rather than to provide precise uncertainty bounds on BDCFs. To this end, parameters were varied over plausible ranges and sampled 10,000 times based around the reference biosphere case. The most sensitive parameters are discussed below, while details are given in Nagra (2024f). The 5th and 95th percentiles for the resulting BDCFs are shown in Figure 6 and typically span a few orders of magnitude, illustrating sensitivity to the varied parameters. To identify the main controllers of this variability and the importance of particular processes and exposure pathways, a correlation analysis was performed (Figure 9). The results illustrate the importance of sorption and transfer factors to food. For example, the negative correlation between aquifer sorption coefficients and the BDCFs of several radionuclides is indicative of how greater retention in sediments of the local aquifer reduces the potential for exposure via drinking water or irrigation. Given the importance of sorption on biosphere materials, an updated sorption database for substrates and geochemical conditions specific to northern Switzerland was developed to improve the selection and justification of the parameter values used in support of the General Licence Application (Metcalfe et al., 2023).

Transfer factors to animal produce are also shown to be important for many radionuclides, especially those for milk. This is particularly true for radionuclides that are not strongly sorbing, such as Cl-36 and I-129, whereby their relative mobility in ground and surface water enhances their availability for animal and crop uptake pathways. Given the sensitivity of the results to this parameter, and the general importance of the milk consumption pathway (Figure 7), an updated database for cows’ milk transfer factors was employed in the modelling documented here (Howard et al., 2017). This further supports the exploration of the broader uncertainty in the milk exposure pathway that is addressed in part through the alternative biospheres and the deterministic sensitivity case that employs more realistic values for dairy consumption.

In this context, the overall assessment philosophy of employing pessimistic assumptions about potential exposure pathways, whilst retaining realistic representation of radionuclide behaviour, is highlighted; this approach helps to ensure that potential radiological doses to humans are not underestimated, whilst avoiding an overly pessimistic assessment.

7 Conclusion

Nagra’s biosphere assessment for the post-closure safety case of the General Licence Application for a deep geological repository is underpinned by a robust, evidence-based understanding of the long-term geomorphological evolution of Northern Switzerland. Consistent with international best-practice, knowledge of site-specific climate evolution, landscape development, hydrogeology and land use have been integrated to provide a scientifically robust representation of both a reference biosphere and a set of bounding alternative biospheres. These stylised yet credible systems are informed by a comprehensive set of FEPs that influence radionuclide migration (from groundwater discharge in valley bottoms to soil-plant interactions and food-chain transfers) and provide the basis for realistic radionuclide transport models and evaluation of a broad range of pessimistic exposure pathways.

The range of biospheres considered, and their corresponding sets of BDCFs, allow radionuclide release rates into the deep aquifers to be converted to potential human dose rates under a wide range of plausible future conditions. In tandem with the deterministic and probabilistic sensitivity analyses, these provide a framework that envelopes uncertainty in climatic, geomorphic and human behaviour, ensuring that potential exposures are not underestimated. Importantly, the modelling shows that the reference biosphere provides a generally pessimistic basis for dose assessment, with alternative climates or morphologies displaying higher or lower BDCFs depending on the hydrological and agricultural context. In summary, the comprehensive approach to biosphere assessment contributes toward robust system understanding and provides a sound basis for evaluating the long-term safety of the deep geological repository system.

Generative AI statement

The author(s) declared that generative AI was not used in the creation of this manuscript.

Any alternative text (alt text) provided alongside figures in this article has been generated by Frontiers with the support of artificial intelligence and reasonable efforts have been made to ensure accuracy, including review by the authors wherever possible. If you identify any issues, please contact us.

Publisher’s note

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

Footnotes

Edited by: Paul Smith, Safety Assessment Management GmbH, Switzerland

Reviewed by: Axel Liebscher, Federal Company for Radioactive Waste Disposal, Germany

Lauri Parviainen, Posiva, Finland

Footnotes

https://www-ns.iaea.org/projects/modaria/modaria2.asp

https://www.bioprota.org/