Managed aquifer recharge and sedimentological characterization within the complex esker deposits in Pälkäne, Finland
- 161 Downloads
It is proposed to construct a managed aquifer recharge (MAR) plant (capacity of 70,000 m3/day) to provide potable water for over 300,000 inhabitants in Tampere and Valkeakoski region in Southern Finland. The plant includes the infiltration of surface water in a Quaternary glaciofluvial ridge (esker) with subsequent withdrawal of the water from downstream wells. We present the sedimentological characterization and rearranged MAR design of the 3 km long unconfined esker aquifer in Pälkäne (capacity of 20,000 m3/day) that were needed to overcome the effect of discontinuities in the hydraulic connections along the Taustiala bedrock threshold as implied by a non-operational preliminary one-layer groundwater flow model. A more detailed sedimentological characterization was obtained by ground penetrating radar and a tracer test indicating flow paths and residence times of the infiltrated water. Four main hydrogeological units were distinguished. The bedrock threshold divides the groundwater flow into two preferred flow paths. The sedimentological interpretation was applied to determine the variation of hydraulic conductivities for the new seven-layer groundwater flow modeling. Sedimentological information together with hydrogeological studies was used in designing the sites and capacities for infiltration and withdrawal.
KeywordsManaged aquifer recharge Esker Aquifer Sedimentology Ground penetrating radar Depositional model Hydrogeology Groundwater flow model
Finnish lakes are typically humic, containing natural organic matter (NOM) derived from the decay of vegetation in the forests and peatlands of the catchment areas. NOM can be detrimental in drinking water as it is a possible source for bacterial growth. The removal of NOM is one of the main tasks in producing drinking water from surface waters in Finland. As an alternative for chemical water treatment, managed aquifer recharge (MAR) can be used for drinking water production.
There are currently 26 MAR plants in Finland (Jokela and Kallio 2015). During MAR, NOM is removed by physical, chemical, and microbial processes (Kortelainen and Karhu 2006; Kolehmainen et al. 2007; Kolehmainen 2008). A typical Finnish MAR plant includes infiltration of lake or river water in an esker and withdrawal of water from wells a few 100 m down-gradient. The infiltrated water should have a residence time of at least 1 month before withdrawal to provide time for processes needed to lower NOM concentrations by break down or remove NOM as evidenced by the operational results from several MAR plants (Helmisaari et al. 2005; Jokela et al. 2017).
Eskers in Finland are glaciofluvial formations which were commonly deposited by streams in tunnels beneath the ice during the final deglaciation of the Scandinavian ice sheet. Typically, an esker consists of 20–50 m of gravel and sand that is covered by a thin humic soil layer (< 10 cm). In general, eskers are preferred areas for potable water MAR treatment. However, they can also be centers of population and are considered recreational or nature conservation sites or they can be sources for extraction of gravel. When MAR plants are being planned, these interests may be conflicting. Public participation is an important feature of MAR planning in Finland (Jokela and Valtonen 2010; Kurki et al. 2013). Site selection and the actual planning for the exact locations for infiltration and abstraction areas should be based on sound research and modeling to ensure reliable MAR operation.
Tavase Ltd. is a wholesale company owned by six municipalities located in southern Finland. The company aims to construct an MAR plant east of Tampere to provide potable water for the municipalities of Tampere and Valkeakoski region. The number of people living in the region exceeds 300,000. The location and design for the MAR plant have been under extensive research. The MAR plant will consist of three MAR sites in separate esker aquifers, two of which are located in the municipality of Kangasala and one in the neighbouring municipality of Pälkäne. The capacities of the aquifers are 22,000, 28,000, and 20,000 m3/day, respectively, and thus, the total capacity of the MAR plant will be 70,000 m3/day. Raw water will be taken from Lake Roine in Kangasala and pumped to the infiltration areas located 1 km (Kangasala) to 7 km (Pälkäne) from the lake. Three infiltration methods will be used: basin infiltration will be applied for approximately one-third of the total capacity (in one of the Kangasala aquifers), whereas parallel well infiltration and sprinkling infiltration systems will be constructed for infiltration of the rest. Sprinkling infiltration comprises aboveground pipe network (infiltration rakes) through which water is distributed on top of natural forest soil. Well infiltration will be the primary infiltration method, but sprinkling infiltration can be used when necessary. No pretreatment will be needed for basin and sprinkling infiltration. However, an optional drum sieving was designed for well infiltration (Jokela et al. 2017). Before distribution to the trunk mains of the distribution networks, water will be disinfected by ultraviolet (UV) radiation, chlorination or both, and the alkalinity and hardness will be adjusted at the shareholder municipalities’ waterworks.
Glaciofluvial aquifers often show complex internal structure and related variation in hydraulic properties that are difficult to predict for groundwater flow modeling purposes and MAR designs. This is due to the overlapping mosaic of deposits by changing depositional environments associated with rapidly changing processes and intense deformation at short distances at the ice margin, where most of the deposition takes place (Mäkinen 2003a). Moreover, those areas that have experienced intense erosion and redeposition by shore processes during the glacio-isostatic land uplift provide difficulties for the interpretation of glaciofluvial sedimentation (Mäkinen and Räsänen 2003). Therefore, the aquifer characterization in glaciofluvial landforms should always be based on the depositional model that includes the understanding of the depositional stages (Sharpe et al. 2002). This understanding is then used to predict sediment distribution and related hydraulic properties in areas with sparse data, and to define main hydrogeological units that enable data input straight into groundwater flow models (Sharpe et al. 2002; Artimo et al. 2003).
The sedimentological understanding can be achieved by applying ground penetrating radar (GPR) sedimentology that enables synthesis of the existing data (Artimo et al. 2010). GPR surveys are fast and cost-effective, and provide continuous profile data over wide areas with possibilities for pseudo 3-D interpretation (Bridge and Hyndman 2004; Artimo et al. 2010). The limited penetration in thick deposits can usually be compensated by the understanding of the depositional environments and stages enabling integration of universal facies models.
The important step in connecting the sedimentological information with groundwater flow modeling (grid size) is scaling, which emphasizes the delineation of major depositional units and architecture instead of detailed sedimentary facies distribution. In the case of esker deposits and associated landforms, the most crucial task is to determine the location and dimensions of the high-conductivity esker core and its relation to bedrock topography and groundwater table. Finally, the sound basis for groundwater flow modeling and MAR design is through integration of sedimentological understanding and infiltration/tracer test results that are applied to verify hydraulic properties as well as the scaling procedure (Artimo et al. 2010).
The aim of this paper is to transfer geological data of a complex depositional system to a reliable groundwater flow model for MAR feasibility assessment. We focus on sedimentological characterization that was needed after the preliminary one-layer groundwater flow model (MODFLOW code) failed to adequately simulate MAR operation due to an excessive 3 m drop in groundwater level between the infiltration area and the pumping wells. Instead, a seven-layer groundwater flow model with 25 × 25 m cell size was constructed to better represent the hydrogeological environment and groundwater flow conditions. The new flow model should form the basis for MAR design and be able to simulate MAR operation for capacity of 20,000 m3/day. The sedimentological characterization and determination of hydraulic conductivity values for the groundwater flow model relies on GPR-determined sedimentology that is used for the synthesis of the different geological data sets, and supplemented with infiltration and tracer test results.
Materials and methods
The study was conducted in the Quaternary Isokangas-Syrjänharju unconfined esker aquifer in the municipality of Pälkäne. The esker (glaciofluvial ridge) is part of a long interlobate esker chain that was deposited between two differently behaving ice masses during the last deglaciation phase of the Scandinavian ice sheet. The Syrjänharju esker in the SE end of the study area forms a distinct ridge, whereas the Isokangas esker in the NW part of the area exhibits more widespread deltaic landforms within generally higher lying bedrock topography. The ice margin terminated in the Yoldia Sea phase of the ancient Baltic Sea with the highest shore around 155 m a.s.l. (above sea level) as indicated by the delta levels (Eronen and Haila 1990). During the rapid glacio-isostatic land uplift, the esker became subjected to powerful shore processes that have markedly modified the original morphology of the esker, especially the distal parts of the esker deltas.
Groundwater observations and monitoring wells
The study area has 67 observation wells and most of them have been drilled down to the bedrock. Groundwater (natural) head (Fig. 1) has been extensively monitored for many years. Before the infiltration and tracer test, 15 groundwater elevation automatic measuring devices were installed in selected wells (Fig. 1). Measurement data were transferred via GSM system to Insta Ltd server from, where it was available for monitoring purposes. Manual groundwater level measurements were also carried out daily/weekly according to the monitoring programme. Several natural springs by the lake (Fig. 1), revealed by the colder water spots, form an important groundwater discharge area included in the model domain. Thus, the groundwater observations were supplemented with gyro-stabilized and computer controlled Flir infrared camera survey (resolution 320 × 240 pixels and temperature resolution of 0.08 °C) using helicopter with 70 and 100 m flight altitudes.
Borehole data and gravimetric surveys
Ground penetrating radar sedimentology
The continuously moving ground penetrating radar (GPR) survey provides information from the geological structures and groundwater levels (Bridge and Hyndman 2004; Artimo et al. 2010). The sedimentological interpretation of GPR profiles is supported with areal reference from geomorphological interpretation and local lithological reference from the borehole logs. In this study, three GPR surveys with total length of 36 km were completed after preliminary groundwater modeling and at different stages using GSSI SIR-3000 georadar by Geo-Work Ltd. The use of 100 MHz antenna (19 km) provided detailed image of the main sedimentary structures with penetration depth of about 15–25 m and forms the framework for the recognition of the depositional stages during the deglaciation.
The penetration depth was improved to 30–40 m in good ground conditions using 40 MHz DUAL antenna (17 km) to better detect sediment stratigraphy and bedrock topography, especially within the Taustiala bedrock threshold. However, the increased penetration means poorer definition of the sedimentary structures, so that only the large-scale features can be determined. Thus, the survey also focused on the delineation of the esker cores and adjacent hidden kettle holes within the thicker sediments in the SE part of the system including the pumping well sites K3 and K4. The kettle holes were formed by melting of buried ice and related collapse of glaciofluvial sediment. They were later levelled and partly filled by the ancient shore processes, so that they now appear only as hidden structures. Finally, the GPR sedimentology forms a geological framework for synthesis of all data sets including groundwater observations as well as infiltration and tracer tests.
Combined infiltration and tracer test
The combined infiltration and tracer test in the Isokangas esker in Pälkäne was carried out as a recirculating test, in which groundwater was recirculated from the downstream pumping wells to the infiltration area. Pumping and infiltration were done for 9 months with maximum rate of 7000 m3/day. Infiltration was done with deep wells in combination with sprinkler infiltration at the surface. In the tracer investigation, 11 mg/l a water-soluble tracer (NaI) was added to the infiltrated water when steady-state conditions were achieved. Sodium iodide was chosen as tracer, because it is very soluble, has negligible retardation in mineral soil, is easily detectable, and is harmless to man and pets even in relatively high concentrations. Tracer was injected to the infiltration wells. Tracer concentration was monitored in 45 observation wells (cf. Fig. 1) along the expected flow paths between infiltration and the two production wells, K3 and K4. In addition, other wells and springs were monitored and the total number of observation points was 100. The highest monitoring frequency was every second day. The tracer investigation was supervised by Geosigma Ltd, Sweden.
Groundwater flow modeling
The distribution of the hydraulic conductivities was predicted by the depositional model based on the sedimentological interpretation of the area (Fig. 3) supported by grain-size changes from GPR profiles and borehole data, and conductivity tests. Esker core with boulder gravel has the highest hydraulic conductivity values, whereas the esker margins as well as delta structures show lateral fining of sediments. The flow model was calibrated to natural groundwater head measurements and validated with infiltration tests, drawdown near the pumping wells and groundwater mound around the infiltration area. The groundwater flow modeling was done by Artimo et al. (2011).
The subglacial esker core and related crevasse deposits
Kettle holes and hidden kettle holes
Esker deltas or deltaic enlargements
The esker core is connected with seven repeated ice-marginal deltas or deltaic enlargements with spacing of about 500 m (Figs. 4, 5c). These deltas exhibit the changing position of the ice margin during the deglaciation and the main palaeoflow direction is indicated by large-scale forests (Fig. 5d). The level of the deltas indicates the highest shore of the area at 150–155 m a.s.l (Eronen and Haila 1990). The delta margins are heavily eroded by shore processes.
The proximal part of the deltas close to the esker core contains coarser material than the distal side and thus locally widen the zone of higher hydraulic conductivities. The proximal side is also associated with thicker esker deposits due to enlarged ice-marginal tunnels. The eastward directed deltaic enlargements on the NE side of the esker at 125–135 m a.s.l. contain till-like deposits several metres thick that probably represent debris flows associated with coarse-grained deltas. The deltaic deposits are also largely eroded by shore processes as indicated by the adjacent kettle holes filled with extensive shore deposits that are well visible in GPR data (Fig. 7).
Fine-grained sediments of esker margins
The margins of the esker (areas outside the outermost kettle holes) and related deltas are fringed with fine sand to silt and clay deposits that were deposited in proglacial water. These fine-grained deposits are located closer to the esker on the western side of the complex, whereas in the eastern side, they are further away from the esker ridge due to the mostly eastward deposited deltas. The western side also has clay–silt layers close to the land surface and are holding perched groundwater. These sediments weaken the radar impulse and restrict penetration depths at the margins of the esker system, but are not a relevant factor for the groundwater modeling.
Groundwater levels were monitored both in the infiltration and recovery areas. The maximum drawdown in the recovery area was 1.5 m in the observation wells at 30–50 m distance from the pumping well K3. The increase of groundwater level in the recharge area was 2–3 m. The maximum increase of 3 m was monitored in the close vicinity of infiltration well IK1. Stable groundwater levels were achieved with different infiltration rates in the recharge area. In general, the changes in groundwater levels were slightly smaller than were anticipated. Examination (Geosigma Ltd., unpublished) of the results from the various observation wells showed that tracer transport seems to occur about as expected along about 2/3 of the distance between the infiltration area and the pumping wells, as indicated by breakthrough data (not shown) from observation wells 104, 370, 273, and 340.
Main flow along the esker cores between boreholes 297–295–340–388–130 (Fig. 1).
Secondary flow first towards the NE between boreholes 339–340 and then towards the east via borehole 251 (Fig. 1). After the bedrock threshold, there is a flow connection from borehole 251 through the eastern margin esker deltas to boreholes 398–393–385–386–274 (Fig. 1).
Average residence time for groundwater flow from the infiltration area to the Taustiala bedrock threshold along the esker core was 47 days based on borehole 340. The hydraulic conductivity of the core is 700–800 m/day (8 × 10−3…9 × 10−3 m/s). Pumping of the infiltrated water was from wells K3 and K4. Both wells showed good specific yields (well K4 approximately 6000 m3/day with 1 m drawdown, data not shown), but only 4–5% of the tracer was recovered. This indicates that the esker core is the main aquifer and that the large kettle holes (with wells K3 and K4 in them) markedly influence and slow down the lateral groundwater flow.
Average residence times for the wells K3 and K4 were 84 and 104 days, respectively. Well K4 is located within the complex esker delta and kettle hole area on the eastern side of the core, whereas well K3 is located within more uniform sediment conditions closer to the coarse-grained core along the groundwater flow direction. Taking into account the high specific yield of the well K4, the lowering of the groundwater level at the beginning of the test was higher than expected. It seems that the kettle holes restrict the flow to some extent. This is also verified by the tracer test results indicating 23 days longer residence time and minor recovery for the well K4. Later, the well started more effectively to collect the infiltrated water. However, at this stage, most of the tracer was already passed along the high-conductivity esker cores as observed in downstream boreholes.
Based on the hydraulic conductivity measurements (Packer test), the conductivity decreases relatively fast outside the esker core as indicated by the well K3 with value of 300 m/day (3.5 × 10−3 m/s). Near the margins of the esker system (boreholes 128, 131, and 293), the conductivities drop to 10–15 m/day (1.1 × 10−4–1.7 × 10−4 m/s). The synthesis of the sedimentological and hydrogeological data implies that the groundwater discharges to the south of well K3 towards the lake with several marginal springs as implied by the tracer test results, while the main flow is towards the Syrjänharju delta, where the two parallel esker cores join the delta system.
Groundwater flow model
The seven-layer groundwater flow model is able to reproduce natural groundwater levels, flow paths, and residence times. However, the modeling has uncertainties in some parts of the margins of the model area due to lack of detailed data coverage. Several water production simulations were run with the model and the results prove its functionality during the infiltration conditions.
The sedimentological characterization and the more accurate data on the influence of the Taustiala bedrock threshold are well in accordance with the infiltration and tracer test results, and enabled the definition of the main hydrogeological units and hydraulic properties for the groundwater flow modeling. A similar sedimentological approach has been successfully applied for the glaciofluvial Virttaankangas MAR plant that has a capacity for 100,000 m3/day and provides drinking water for the Turku city district with 300,000 inhabitants (Artimo et al. 2003; Mäkinen 2003b; Artimo et al. 2010).
The NW bedrock threshold and the associated groundwater divide proved to persist during the highest infiltration volumes.
The Taustiala bedrock threshold in the middle of the area divides the area into two hydrogeological parts and shows a complex groundwater flow over it. Therefore, the MAR design had to be rearranged based on the groundwater flow model simulations.
The pumping wells K3 and K4 are located in kettle hole structures on both sides of the esker core (high hydraulic conductivity) and they markedly restrict the groundwater flow lateral to the core. On the other hand, this enables longer residence times (Artimo et al. 2010).
The discharge along the esker core out from the MAR plant area and lake marginal spring discharges can be controlled by the infiltration and pumping adjustments based on groundwater flow model simulations.
The final MAR design (Fig. 10) eliminates the effect of discontinuities in the hydraulic connections along the Taustiala bedrock threshold and gives better possibilities to regulate groundwater levels and discharges towards the large lake in the south. The seven-layer numerical groundwater flow model reliably explains the observations made during the research. It is applicable in infiltration conditions and can be used for MAR simulations.
The sedimentological characterization using GPR as a key method was able to describe the complex structure and related groundwater flow paths over the bedrock threshold as well as to explain the poor recovery of the NaI during the tracer test. It also managed to estimate the distribution of hydraulic conductivities for the groundwater flow model that is applicable for reproducing the observed groundwater levels and discharges during the natural conditions as well as during the infiltration test. The seven-layer groundwater flow model indicates that preliminary MAR design should be rearranged and that the flow model is able to simulate MAR operation for planned volume of 20,000 m3/day.
We thank the personnel of the groundwater department at Pöyry Finland Ltd. for co-operation during the test trials and reporting and for preparation of maps. We also thank Suvi Rinne (FCG Ltd.) for preparation of maps.
JM did the sedimentological characterization and is the main writer of the paper; EK and PJ provided the information and expertise on MAR design, groundwater investigations, and flow modeling.
Compliance with ethical standards
Conflict of interest
The authors declare no conflict of interest. The founding sponsors had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, and in the decision to publish the results.
- Artimo A, Saraperä S, Puurunen O and Mäkinen J (2010) The Turku Region Artificial Infiltration Project, Finland—tools for enhanced aquifer characterization. In: Proceedings of the 7th international symposium on managed aquifer recharge of groundwater (ISMAR7), Abu Dhabi, 9–13 October, 2010, pp 93–100Google Scholar
- Artimo A, Saraperä S, Puurunen O (2011) Pälkäneen Syrjänharjun monikerroksinen pohjaveden virtausmalli. Raportti 18.3.2011, p 14 (The multi-layer groundwater flow model of the Syrjänharju area in Pälkäne, unpublished report, in Finnish)Google Scholar
- Bridge JS, Hyndman DW (2004) Aquifer characterization. SEPM Special Publication 80, 176 ppGoogle Scholar
- Eronen M, Haila H (1990) The main ancient shorelines. In Atlas of Finland, Alalammi P (eds) National Board of Survey, Geographical Society of Finland: Helsinki, Finland; Folio 124, Geology, Surficial deposits, Figure 17a, p 17Google Scholar
- Helmisaari HS, Derome J, Hatva T, Illmer K, Kitunen V, Lindroos A-J, Miettinen I, Pääkkönen J, Reijonen R (2005) Artificial recharge in Finland through basin and sprinkling infiltration: soil processes, retention time and water quality. In: Recharge systems for protecting and enhancing groundwater resources, Proceedings of the 5th international symposium on management of aquifer recharge, ISMAR5, Berlin, Germany, 11–16 June 2005. UNESCO IHP-VI, Series on Groundwater No. 13, pp 617–623Google Scholar
- Jokela P, Valtonen J (2010) Managed aquifer recharge and public participation. In: Proceedings of the 7th international symposium on managed aquifer recharge (ISMAR7), Abu Dhabi, 9–13 October 2010Google Scholar
- Kolehmainen R (2008) Natural organic matter biodegradation and microbial community dynamics in artificial groundwater recharge. Dissertation. Tampere University of Technology, Publication 781, Tampere, FinlandGoogle Scholar
- Mäkinen J (2003b) Development of depositional environments within the interlobate Säkylänharju-Virttaankangas glaciofluvial complex in SW Finland. Annales Academiae Scientiarum Fennicae Geologica-Geographica 165:1–65Google Scholar
- Pöyry Ltd. (2011) Tavase Oy imeytys- ja merkkiainekoe Pälkäneellä, loppuraportti. 30.3.2011. Pöyry Finland Oy. (Tavase Ltd. Infiltration and tracer test in Pälkäne, final report, Pöyry Finland Ltd., in Finnish)Google Scholar
- Sharpe DR, Hinton MJ, Russell HAJ, Desbarats AJ (2002) The need for basin analysis in regional hydrogeological studies: Oak Ridges Moraine, Southern Ontario. Geosci Can 29:3–20Google Scholar