Assessing Ice-Sheet Variability for Post-Closure Safety of Deep Geological Repositories

In many regions considered for deep geological repositories (DGR) to contain nuclear waste, there will be repeated glaciations throughout periods pertinent to their long-term safety (up to 1 million years; Ma). Ice sheets can influence the containment of radionuclides through various mechanisms. For example, when the margin of an ice sheet is situated in close proximity to the DGR, groundwater flow may increase, potentially leading to enhanced erosion and corrosion of the technical barriers within the DGR. Furthermore, the temporal extent of glaciations at the DGR site impacts groundwater chemistry, such as salinity and oxygen content, as well as the magnitude of glacial isostatic adjustment and surface bedrock denudation over the ensuing 1 Ma. Consequently, evaluations of long-term DGR safety must account for uncertainties related to ice-sheet variability at the DGR site throughout the next 1 Ma, specifically addressing the frequency of glaciations (n_glac) and the total duration of ice-sheet coverage (t_glac). Additionally, assessments should consider the potential for ice-marginal stillstands, denoting temporary halts in the advancement and/or retreat of the ice-sheet margin over the DGR site.

The utilization of coupled ice sheet-climate models for constraining uncertainties in n_glac and t_glac over the next 1 Ma is not feasible due to the long timescales involved and substantial computational requirements. To assess future ice-sheet variability, we propose a simplified methodology that uses (i) reconstructions of historical ice sheets, (ii) records of past global ice-volume fluctuations, and (iii) simulations of future global ice-volume changes. These simulations are conducted using a simple multi-step climate model, which is driven by changes in insolation and radiative forcing due to atmospheric greenhouse gases.

Utilizing the proposed methodology on the Swedish site chosen for nuclear waste disposal (Forsmark) suggests that the onset of the first glaciation at the site is projected not to take place within the coming 100,000 years (100 ka), irrespective of human-induced greenhouse-gas emissions. Following the initial glacial event at Forsmark, the frequency and duration of subsequent glaciations will likely be similar to those observed in the late Quaternary (last 800 ka). Taking into consideration identified model and scenario uncertainties, the total glaciation duration (t_glac) at Forsmark may either decrease by a factor of five or increase by a factor of two in comparison to the average conditions of the late Quaternary. In contrast to t_glac, the number of glaciations (n_glac) at Forsmark is found to be largely insensitive to the evaluated uncertainties.

The potential for ice-margin stillstands within the next 1 Ma is assessed through theoretical considerations complemented by simulations with ice-sheet model simulations. Initial findings indicate that stillstands will be brief, lasting less than 1000 years under nearly all examined scenarios. This observation aligns with historical records of stillstands during the last deglaciation. The occurrence of stillstands modestly exceeding 1000 years can only occur if a glacial maximum is promptly followed by a millennial-scale cooling event.