Considering the erosion assessment aimed at ensuring uncompromised barrier properties during the period under consideration, the following conclusions are drawn:

  • The overall knowledge about the capability, timescales, form and rate of erosion processes, and the erosion history of Northern Switzerland have been improved considerably by detailed field and laboratory work during recent years. The effects of recurring climate-driven erosion processes superposed on the low-strain domain of Northern Switzerland have been identified and analysed with respect to landscape evolution. Nevertheless, age determination and the limited availability of dated Quaternary geomorphic markers and sediments (e.g. Decken­schotter, infills of glacially overdeepened valleys) remain a challenge. Consequently, the age uncertainties of Quaternary landforms and deposits are accounted for in a systematic assessment of future erosion.

  • Future erosion processes (glacial and non-glacial) are expected to be generally comparable to processes that have repeatedly influenced the area of the siting regions during the Quaternary, but they are inferred to operate at lower rates. One main argument for slower future process rates is the formation of smaller glaciers and a delay in glacial inception because of anthropogenic CO2 emissions. In addition, analyses of the fluvial topography show that the catchment of the Rhine River will grow at the expense of the Danube catchment, thus no major pulse of drainage reorganisation will cause deep incision as the major wave of retrogressive erosion of the Rhine has already passed by the area of the three siting regions. In addition, less erodible rocks in the stratigraphic sequence below the Molasse strata are predicted to slow down erosion further.

  • The assessment of future erosion was done using a robust, structured hybrid-probabilistic approach, based on available field data, rate determinations, laboratory tests and modelling, to evaluate the behaviour of the erosional system in Northern Switzerland for different future scenarios.

  • A residual cover thickness of 200 m as required for maintaining the self-sealing properties of the Opalinus Clay can be shown for the most likely scenarios in all the siting regions for the period under consideration. Using the 5 – 95% range as reference, the JO site shows more limitations for such a 200 m criterion at the location of the provisional disposal area. How­ever, suitable conditions in JO are available in the immediate surroundings of this provisional disposal area, such that JO can also reliably provide sufficient overburden thickness to secure the barrier function over the period under consideration. Nevertheless, because of the com­parably small distance between the repository depth and the local erosion base, JO is dependent on the preservation of the local topography and is thus more sensitive to changes in the fluvial network compared to the other sites. The NL site is a well-suited location for the development of a repository because it is protected against future erosion from all of the principal erosion processes capable of causing deep incision by the large emplacement depth.

  • Fluvial incision, driven mainly by rock uplift, dominates the modelled overburden thickness assessment on a timescale of one million years. Although deep glacial erosion can greatly increase the degree of incision, it is less likely at the sites. This is because of the low pro­bability of glacial overdeepenings occurring at the location of the provisional disposal area, the more resistant rocks below the Molasse strata, and the longevity of elevated atmospheric CO2 concentrations that are predicted to delay future foreland glaciations. Because of conti­nuous erosional downcutting and subsequent decrease of the accumulation area of future glaciers, foreland glaciations might be smaller even with similar climate forcing.

  • Excavation of the repository by erosion within the next one million years can be considered extremely unlikely in all the provisional disposal areas but is nearly an order of magnitude lower in NL.