Forsmark Site: M3 modelling and 2D visualisation of the hydrochemical parameters in Forsmark groundwater

This work represents the stage 2.3 of the hydrochemical evaluation and modelling of the Forsmark data. This comprises M3 modelling and 2D visualisation of the data along the boreholes. The following conclusions can be drawn: - M3 modelling helped to summarise and understand the data, by using as variables the major elements and the isotopes δ18O and δD. - Previous alternative models and the experience from Forsmark 1.2, 2.1 and 2.2, helped to clarify different previously unsolved issues such as: the use of variables, tests with different endmembers, the use of only groundwater data in order to build a bedrock hydrochemical model. - The visualisation of the mixing proportions along the boreholes helps to understand the distribution of the data in the domain and to check and compare the results of different models; and therefore to chose the model which best describes the measured data. - The different M3 modelling tests resulted in the following conclusions: a) When calculating mixing proportions only samples from the boreholes will be used, b) the altered meteoric end-member which best describes the more shallow groundwater compositions is defined by a representative upper bedrock sample; the Littorina end-member employed the existing modelled compositions; the Deep saline and glacial end-members compositions were tested by means of a feasibility study and employed in the modelling. - Three models were presented. All the models are good and can be used, but the best is to use the one that fits the conceptual model best and the hydrogeochemical understanding. - The use of Littorina, Glacial, Deep Saline and Altered Meteoric end-members makes possible the comparison of different sites such as Laxemar and Forsmark. - All the data used in the M3 modelling and the results of the modelling and visualisation along the boreholes are presented in SKB database SIMON. - The extended data do not affect the results of the modelling of the Forsmark 2.3 dataset (the difference between the mixing proportions calculated with both models is less than 1%)