Evaluating transfer-function models to understand groundwater level impacts - A study of head changes due to drainage and infiltration in an underground infrastructure project

Abstract

When analysing the effect of underground construction on groundwater levels, process based models are often applied. They generally require a detailed conceptualisation of the study area supported by a large amount of information and a thorough description of the flow system. The use of time-series driven transfer function-noise models is a less time consuming approach to assess the effect from groundwater drawdown. They require fewer types of input data and are easier to apply and could therefore show potential in groundwater modeling to complement or replace process based models. This study aims to evaluate how transfer function-noise models can be used to evaluate groundwater head changes, as a result of infiltration and drainage from an underground infrastructure project. The study uses an area surrounding the Haga service tunnel, part of the Västlänken project in Gothenburg, as a site study. The transfer function-noise models are created with the use of groundwater head observations, estimated recharge, leakage measurements and infiltration measurements. After a completed calibration process, the model outcome is used to perform stress simulation scenarios, cluster analysis of residuals and parameters as well as an analysis of parameter correlations. The stress simulations are performed in order to quantify the impact from infiltration and drainage on the groundwater heads. Furthermore, the cluster analysis is performed to establish the spatial impact in the area and to evaluate how the model is parameterised. Lastly the analysis of parameter correlations is performed to examine the possible relations of the model parameters against site physical characteristics. This study demonstrates that transfer function-noise models show potential to be used to identify and quantify the stress that has caused a disturbance, identify missing or inaccurate stress measurements and to simulate different stress scenarios. The study also show that the groundwater levels in the site study area have been impacted by the service tunnel and the spatial impact could be established with a cluster analysis.