Remaining uncertainties

5.9.1 Deterministic model for rock domains

Despite the significant increase in borehole data in the local model volume after model stage 2.1, only very minor changes have taken place in the rock domain model in this volume compared with that in model stage 2.1. Furthermore, the predictions provided by model stage 2.2 for the intersection of rock domains RFM029 and RFM045 in borehole KFM08D have been verified satisfactorily during model stage 2.3. In addition, few changes in the regional rock domain model have occurred since SDM version 1.2. Bearing in mind these three observations, it is judged that the intrinsic uncertainties in the positions of the boundaries between rock units on the surface geological map and in the boreholes at depth /Munier and Stigsson 2007/ are of little significance for the modelling of rock domains, particularly in the local model volume. The uncertainties in the modelling work that concern the extension at depth of virtually all the rock domains outside the local model volume and the quantitative estimates of the proportions of different rock types in most of the rock domains remain. However, these uncertainties are not judged to be significant for the understanding of the geological relationships inside the target volume.

5.9.2 Deterministic model for deformation zones

Deterministic modelling, not least of deformation zones, is strongly dependent on an accurate posi-tioning of boreholes at depth. The uncertainty calculated for the spatial position of boreholes in all three dimensions generally increases somewhat with depth and is more significant in the horizontal plane than in the vertical dimension /Munier and Stigsson 2007/. The estimated uncertainty in the position of, for example, a possible deformation zone in a borehole does not exceed c. 30 m in the horizontal plane. In most cases, the uncertainty is less than 10 m in the horizontal plane and less than 6 m in the vertical dimension. These uncertainties are approximately of the same order of magnitude as the uncertainty in the position of low magnetic lineaments /Isaksson and Keisu 2005, Isaksson

Figure 5‑44. Refined base model for modelling the gravity response from rock domains, version 1.2, inside the regional model area (solid red line). Rock domains are numbered (cf. Figure 5-24a). Responses are only shown within the boundary buffer area (dashed red line). The residual between the input gravity anomaly and the model anomaly [mgal] has been calculated for each survey station and is shown on the map. Blue colours indicate a mass surplus in the rock domain model and red colours indicate a mass deficiency. The candidate area is shown with a solid magenta line. The local model areas for version 1.2 and stage 2.2 are shown with a solid blue line and a dashed yellow line, respectively.

orientation of sets of fractures in some deformation zones, when data corrected after the completion of model stage 2.2 have been used /Stephens et al. 2008/. The zones selected for this analysis lie in the north-western part of the tectonic lens (zones ENE0060A, WNW0123, A2, A8 and F1).

The judgement in the modelling work to match a particular low magnetic lineament to a particular deformation zone in a single-hole interpretation, as well as the assumption concerning the down-dip extension of steeply dipping or vertical zones are intrinsic weaknesses in the modelling procedure.

These features affect directly the estimate of the dip of the modelled zone, its thickness and its predicted intercept at 400−500 m depth. However, the predictions provided by local model stage 2.2 for the position and character of the intersections of deformation zones in the four cored boreholes not used in this model have been verified highly satisfactorily during model stage 2.3 /Stephens et al. 2008/. Furthermore, the additional ground magnetic data acquired during stage 2.3 only have a limited effect on the position and trace length of deformation zones at the ground surface in model stage 2.2 /Stephens et al. 2008/. As predicted in model stage 2.2 /Stephens et al. 2007/, no new deformation zones with a trace length longer than 3,000 m at the ground surface have emerged.

Nevertheless, if Forsmark is selected as the site for the deep geological repository, then it will be necessary to complete minor modifications to the deterministic geological models prior to construc-tion work.

All the observations summarised above provide a basis for confidence in the deterministic models for deformation zones, notwithstanding the intrinsic weaknesses noted above. It is suggested that the successful predictability of the occurrence and character of deformation zones is related to the significant bedrock anisotropy that was established over 1,850 million years ago, when the bedrock was situated at mid-crustal depths and was affected by penetrative, ductile deformation under high-temperature metamorphic conditions.

The following summarises the two remaining more significant uncertainties in the modelling of deformation zones.

• The size of gently dipping fracture zones. The significance of this uncertainty has been reduced somewhat through the selection of the target volume /SKB 2005c/. However, it remains for the gently dipping zones that are present above 500 m depth in this volume. It also needs to be kept in mind that the absence of gently dipping zones in approximately the north-eastern half of the regional model volume is coupled directly to an inherent data bias in the regional model, since reflection seismic data are not available for this area.

• The orientation and size of the possible deformation zones in the single-hole interpretations that have not been modelled deterministically. Since it has not been possible to link these geological features to a low magnetic lineament or seismic reflector and since they commonly occur along short borehole intervals, it is judged that they predominantly represent intersections with minor zones.

5.9.3 Statistical model for fractures and minor deformation zones Limitations

The data available from surface outcrops lie entirely within fracture domains FFM02 and FFM03.

Borehole data come from both these two domains as well as from domains FFM01 and FFM06.

Since the coupled size/intensity models for the outcrop scale and the TCM models are based wholly or partly on outcrop data, the models have the least uncertainty when applied to fracture domains FFM02 and FFM03. Application to FFM01 is more uncertain, since there are no outcrop data to sup-port the size model, although there are borehole data for quantifying relations between intensity, rock type and depth. The limited volume of data from domain FFM06 was not used in the uncertainty analysis. As a result, the application of the model to FFM01 has a much higher uncertainty than the application of the model to FFM02 or FFM03. Application to FFM06 is uncertain to a degree that could not be quantified with the data available.

The tectonic fault model (TFM), which is calibrated in part on the ground magnetic lineament data, has a higher degree of uncertainty than the other models. There is a mismatch, an over-prediction, between the number of larger fractures predicted by the model and the number obtained by

measure-ments. However, the detection reliability of the method has not been quantified. It is therefore possible, until the opposite can be demonstrated, that the ground magnetic measurements fail to detect the portion of the fracture array which shows weak magnetic signatures. Thus, the detection reliability of the magnetic lineament data and the TFM alternative model based on them should be carefully assessed by any user.

Key uncertainties

The key identified uncertainties are:

• Does the tectonic continuum exist, allowing for the development of a single model to encompass borehole, outcrop, ground magnetic lineament and deformation zone data?

• If the tectonic continuum does not exist, and there are distinct populations of joints and

reactivated joints that differ from fault/deformation zone related fractures, then what is the upper size limit to the joints or the lower size limit to the deformation zone related fractures?

• What is the impact of fracture intensity variations by rock type? Can intensities for rock types be combined, and if so, what magnitude of uncertainty does this produce?

• Does the mean pole of a fracture set vary spatially within a fracture domain? If so, what

uncertainty is likely to be induced if a constant mean orientation is used for each set within each fracture domain?

Complementary studies on fracture mineralogy

Two complementary studies are in progress at the current time, the results of which have not been assessed in this report. These studies were designed to reduce uncertainties in some aspects of the fracture mineralogy (see section 5.2.5). The first of these studies concerns a more quantitative evalu-ation of the amount of different minerals along fractures. The second aims to provide more detailed information bearing on the significance and origin of fractures that lack a mineral coating or filling and wall-rock alteration.