Posiva publishes Working Reports and Posiva Reports. From the year 2006 nearly all the reports have been published on our webpage and they can be found in the databank. In the databank you can also find our Annual Reviews and some other publications as well. You can also find print-quality pictures and useful links in the databank.

Recent publications

POSIVA Report 2016-18



Long-term safety of a planned geological repository for spent nuclear fuel in Forsmark, Sweden and Olkiluoto, Finland Phase 2: impact of ice sheet dynamics, climate forcing and multi-variate sensitivity analysis on maximum ice sheet thickness


Aurélien Quiquet, Florence Colleoni, Simona Masina



Page count:





This study is the second and final phase of a study aiming at simulating the impact of a past extensive glaciation over Forsmark, southeastern Sweden, and Olkiluoto, southwestern Finland. To this end, the Late Saalian glaciation (~ 192–135 kyrs BP), which is known to be the most extensive glaciation that occurred over Eurasia during the last 400 kyrs, was used as case study. In fact, some recent reconstructions suggest that during the peak of this glaciation, the area of the Eurasian ice sheet was at least twice as large as during the Last Glacial Maximum (LGM, ~ 21 kyrs BP). In the first phase of the study, we used a coupled Atmosphere-Ocean-Sea-Ice-Land model, set up and integrated with constant-in-time forcing Late Saalian conditions to simulate the Late Saalian glacial maximum climate at ~ 140 kyrs BP (i.e. the MIS 6 period). A set of two climate simulations were performed since the extent of the Laurentide ice sheet during this glaciation is uncertain: one including the LGM Laurentide ice sheet topography (B140_Topo1) and one including a smaller Laurentide ice sheet topography (B140_Topo2) (Colleoni et al. 2014b, 2016a, b). In both simulations, the prescribed Eurasian ice sheet was based on a previous ice sheet model simulation for the Late Saalian glacial maximum. Secondly, the simulated climates were, in the first phase of the study, used as initial condition to carry out two sets of 70 univariate simulations with an ice sheet model (Colleoni et al. 2014b, 2016b, Wekerle et al. 2016). In the complete set of sensitivity experiments, the ice sheet thickness in the Forsmark region ranged from 2991 (2650) metres to 3472 (3195) metres when using the B140_Topo1 (B140_Topo2) climate forcing. The corresponding bedrock depression increased from 707 (632) metres to 822 (760) metres. In the Olkiluoto region, the ice thicknesses ranged from 3109 (2791) metres to 3551 (3138) metres when using B140_Topo1 (B140_Topo2) climate forcing, which induced a bedrock depression ranging from −735 (−663) to −846 (−790) metres over the site.

In the second phase of this study, described in the present report, remaining issues identified in the previous study are investigated: i) the impact of different ice dynamics by means of ice sheet modeling, ii) the effect of different surface mass balance parametrisations, iii) the sensitivity of the ice sheet to different climate forcing, and iv) the impact of multi-variate sensitivity experiments on the Late Saalian Eurasian ice sheet topography. In addition, a second ice sheet model, with a different treatment of the ice temperature, but lacking activation of fast flow areas, is used to assess the impact of thermodynamics on the results presented in this study. The results show that the activation of fast-flowing areas in the ice sheet models is the process that most affects the ice thickness. Simulating the LGM Eurasian ice sheet or the last deglaciation (using six different simulated LGM climates from the PMIP3 project) does not help to further constrain those basal ice sheet processes for the Late Saalian. All together, the impact of the Surface Mass Balance (SMB) parametrisations does not induce large variations in ice thickness compared to the reference simulations over the two sites. Conversely, the use of various reconstructed numerical Late Saalian climates, based on six AOGCMs from the PMIP3 project, has a larger influence on the simulated Late Saalian than the SMB parametrisations.

Finally, multi-variate sensitivity experiments are carried out with the Latin Hypercube Sampling method (LHS). Parameters related to SMB and to ice sheet dynamics are sampled evenly, and two sets of 100 multi-variate experiments are carried out. The resulting averaged ice volumes are of about 58 m

SLE using B140_Topo1 and about 56 m SLE when using B140_Topo2. This is slightly more than the averaged peak Saalian ice volumes obtained during the first phase of the study (by about 2 to 3 m SLE). The averaged ice thickness at Forsmark, from all simulations, ranges from 3128 to 3460 m. This leads to an averaged bedrock depression of −736 to −822 m. Over Olkiluoto, as already observed in the results of the first phase of the study, the averaged ice thickness is slightly larger. The thickness ranges from 3318 to 3585 m, associated with an averaged bedrock depression of −784 to −856 m. If we consider that the ensemble mean has more meaning than the members presenting the extreme values, the ice thickness obtained in response to changes in the climate forcing is the largest of this study. The maximum simulated steady-state ice thickness over both sites reaches around 4000 metres. If taking into consideration the global ice volume estimate for this time period, even though geological evidence supports a Laurentide ice sheet smaller than during the Last Glacial Maximum, this maximum ice thickness over Forsmark and Olkiluoto is likely to be an overestimation.


Olkiluoto, climate modelling, Saale glaciation, ice-sheet, modelling


POSIVA 2016-18_web (pdf) (22.4 MB)


Share article: