Coupled modelling of permafrost and groundwater : a case study approach

This report investigates the sensitivity of simulated permafrost thickness and dynamics to a variety of climatic, geological and hydrogeological conditions for two geological environments, basement under sedimentary cover and a low permeability succession of Mesozoic shales and siltstones (Case 1 and Case 2 respectively). A combination of one dimensional heat conduction modelling, including the effects of freeze-thaw, and two dimensional heat conduction-advection modelling, including freeze thaw, has been undertaken to simulate permafrost development in these two contrasting geological environments. This enables an assessment of the sensitivities to a range of possible geological parameters, advective heat flow, and the effect of glaciation with and without the influence of glacial loading. In this report, permafrost is defined as the sub-surface in which ice is present even in very small amounts, i.e. ice content is greater than 0%, and in the model, this is at the zero degree isotherm. The maximum permafrost thickness is strongly dependent on the mean annual surface temperature, the presence of ice that will insulate the system and the duration of the cold phase. By scaling the minimum temperature of 57 Pliocene-Pleistocene globally distributed benthic δ18O records to temperatures of 14°C, 18°C and 25 °C below the present day mean annual temperature, the maximum permafrost thickness for Case 1 is simulated to reach 171 m, 248 m, and 475 m, and for Case 2 80 m, 138 m, and 238 m respectively. The difference in permafrost thickness between the two Cases is attributed to the variation in subsurface rock properties. Deeper permafrost depths than for Case 1 and 2 can be expected where the thermal conductivity is higher than for Case 1 and 2. A sensitivity study of the geological parameters has shown that there is a strong, non-linear, relationship between thermal conductivity, latent heat and geothermal heat flow for a series of temperatures representative of the glacial cycles of the past one million years. This is in contrast to a steady state temperature profile, where permafrost thickness relates linearly to thermal conductivity, heat flow and ground surface temperature. Thickest permafrost under unchanged climatic conditions is to be expected where there is a low heat flow, a high thermal conductivity and a low porosity, such as for example in the north of Scotland. The results of the modelling show that when the temperature regime is dominated by heat conduction, such as for the low permeability Case 2, a heat conduction only model is sufficient to estimate the thickness and distribution of permafrost. However, when heat advection is likely to be important, such as in Case 1, the coupling of permafrost and groundwater flow is necessary to simulate the permafrost distribution during freeze and thaw, or during shallow permafrost events. This particularly holds true when permafrost is modelled to be relatively permeable, where modelling suggests that heat advection of cold water at recharge points (interfluves) results in cooling and thicker permafrost compared to discharge points where discharge of warmer water results in thinner permafrost. However, these variabilities in local permafrost thickness are of minor importance for the question of freezing of the repository. However, when assessing the broader influences of permafrost on a geological environment, local variations in permafrost extent of thickness can have consequences on the biosphere. Glaciation influences the thermal regime of the ground surface. If the glacier bed is undergoing pressure melting, as found in the ablation zone, a reduction in permafrost depth can be expected. If the glacier bed is cold based, as often found in the accumulation zone or at ice divides where strong vertical advection of cold ice has a cooling effect, then the maximum permafrost thickness can be expected to be similar to the scenario without glaciation. It may even increase if the temperatures at the glacier bed are colder than the ground surface temperatures, which may occur. when the temperature in the area where the ice is forming is colder than that prevailing downstream. Recharge and discharge decrease considerably during periods when permafrost is present. In the case of a model with an open model boundary to one side, representing the coast for example, and a high topographic gradient (Case 1), a large drop in hydraulic heads is observed beneath the permafrost. This results in lower groundwater flows at depth compared to unfrozen conditions. Where a modelled area is closed on all sides (Case 2), a decrease in flow at depth is also observed, however the hydraulic heads do not decrease to the same extent as the hydraulic gradient is less than for Case 1. During permafrost thaw, hydraulic heads rise, resulting in an uptake of groundwater into elastic storage from recharge over the top boundary of the model domain. When taliks underneath surface water bodies develop, the groundwater flow system remains more active than during continuous permafrost. Recharge and discharge are focused on the lakes and a regional groundwater flow system connecting the lakes can develop. Heat advection remains more important during thick permafrost when through taliks remain open. In the model, during periods of glaciation, hydraulic heads increase by ~1500 m at depth for Case 1 and Case 2 when ice loading is applied. When ice-sheet loading is not accounted for, the hydraulic head signal in low permeability layers is dampened. During glacial advance, groundwater recharge increases by up to two orders of magnitude, and during glacial retreat discharge increases. During ice advance, groundwater flow is in a downward direction but during ice retreat it is in an upward direction. Depending on the flow direction of the glacier, groundwater flow directions can be reversed during a glaciation. Modelling the Anglian Glaciation (middle Pleistocene glaciation, equivalent to the Elsterian or Mindel glaciation in Europe and the Alps, most extensive glaciation in the British Isles, MIS 12), the hydraulic head and groundwater flow magnitude are affected by the glaciation for tens of thousands of years, whereas after the Devensian glaciation (late Pleistocene glaciation, equivalent to the Weichselian/Vistulian or Würm glaciation in Europe and the Alps, MIS 5d to 2), the signal remains for thousands of years. High hydraulic heads that may be present during glaciation are likely to modify the groundwater flow around a GDF. The modelling presented here based on two settings and typical thermal and hydraulic properties for the rocks present, demonstrates that the depth of permafrost could extend up to a depth of 300m below the surface and, depending on specific characteristics (large thermal conductivity and low porosity) and an exceptionally long cold period, could extend to greater depths. Permafrost to these depths may affect the engineering properties of some rock types and could lead to the development of new fracture pathways in more brittle formations. Permafrost could also affect some of the engineered components of a GDF in similar ways, such as the properties of clay materials.