USING MULTIPLE POINTS STATISTICS (MPS) IN SWEDEN TO OPTIMIZE CHANCE OF FINDING DEEP, LOCAL GRAVEL AQUIFER AND ESTIMATE GRAVEL THICKNESS

Grevie vattentäkt is an important water supply area belonging to the water utility VA SYD. It is located in the Southwestern part of Sweden. Groundwater is abstracted from a quaternary gravel layer at the bottom of a partly buried valley, the Alnarp Valley. The gravel is deposited in an ancient river system on a surface of chalk. The chalk defines the bottom of the Alnarp Valley. The gravel is superimposed by sandy and clayey sediments and is confined to certain parts of the old river system and is widely but locally distributed. In some areas the gravel isn’t present at all, and where present, the thickness differs from a few meters up to 8 meters at best. The groundwater in the gravel layer is mostly artesian. The underlying chalk contains chloride, a relic from the depositional environment. A growing need for water, combined with rising chloride content in many existing abstraction wells, leads to the need of making new abstraction wells, to secure the future water supply.

New wells are complicated and expensive to make, providing a pronounced need for as much certainty as possible, for finding the gravely aquifer when pointing out new well locations.

Airborne electromagnetic (AEM) measurements and ERT profiling falls short due to the geological settings: Approximately 40 m of clayey glacial sediments overly 30 m of relatively sandy sediments. At the bottom of this sandy unit, the gravel layer is found on top of limestone. Furthermore, the limestone is fractured in places, and shows relatively high levels of chloride, which locally are found in the soft sediments as well.

Based on these conditions, it was decided to use geostatistical approach (MPS) to estimate gravel thickness and to create a “traffic light map”, showing high, medium, and low probability of finding gravel. This map is based on 133 realisations – or probable geological models, based on existing data and information. The MPS simulation utilizes a training image as well as soft and hard data – AEM and well data. For the AEM data, a new inversion approach has been used, a probabilistic inversion method, using the conceptual geological model as a prior in the inversion.

The presentation reviews and discuss the workflow, including relevant decisions and assumptions, data preparation, the inversion strategy and more, and how the simulation results are being used today.