River-Aquifer Interaction in the Uppsala Esker - a Modelling Study of a Proposed Drinking Water Production site

Sammanfattning: The Swedish municipalities of Gävle and Älvkarleby need new sources of drinking water as the population grows. Gästrike vatten AB has employed the consultant firm Midvatten AB to assess the possibility of a new groundwater extraction site on the Uppsala esker between Älvkarleby and Skutskär in northern Uppland county. It has been observed that the natural recharge to the aquifer in the Uppsala esker might be too low to compensate for a future groundwater extraction and that there is a risk of induced infiltration from the river Dalälven if the water table is lowered. River water might bring organic contaminants into the aquifer and negatively affect the groundwater quality.A solution proposed by Midvatten is to infiltrate the esker with river water free from organic contaminants at infiltration sites. This artificial infiltration is estimated to create new groundwater to compensate for the extraction and stop river water from reaching the extraction wells. There is however, a need to estimate the magnitude of infiltrating river water when the infiltration sites are active.The aim of this study was to estimate the flow of water between the river and a section of the Uppsala esker for a test period during 2017, specifically, the infiltration from the river. In addition to this, changes in flow depending on proposed pumping and infiltration scenarios were modelled.A MODFLOW model was developed in the graphical user interface Groundwater Modeling System (GMS) and its performance was validated against observed aquifer head. The model could accurately represent the head close to the river but was less accurate with increasing distance from the river. Average infiltration from the river varied from 3 to 25 l s-1. The calculated infiltration depended on which extraction well or artificial infiltration site was active and the rate of flow.It was concluded that the distribution of hydraulic conductivity in the aquifer was not sufficiently detailed. A solution could have been to use stratigraphic data from borehole logs instead of a general quaternary deposits map as basis for the distribution of hydraulic conductivity. Artificial infiltration close to the river prevented large volumes of induced infiltration. The accuracy of the model could have been improved if the results were compared to other methods such as particle-tracking, tracer tests and with measurements of the streambed such as seepage meters.