Investigating Sediments and Rock Structures Beneath a River Using Underwater ERT

Cent. Eur. J. Geosci. • 4(1) • 2012 • 81-93 DOI: 10.2478/s13533-011-0052-0

Central European Journal of Geosciences

Investigating sediments and rock structures beneath a river using underwater ERT

Research Article

Jannis Epting∗1, Andreas Wüest1,2, Peter Huggenberger1

1 Applied and Environmental Geology, Department of Environmental Sciences, University of Basel, Bernoullistrasse 32, 4056 Basel, Switzerland 2 Present address: Keller + Lorenz AG, Tribschenstrasse 61, 6005 Luzern, Switzerland

Received 27 October 2011; accepted 14 January 2012

Abstract: This article presents hydrogeophysical investigations performed in a well-developed, long-term hydrogeological gypsum karst research site where subsurface evaporite dissolution has led to the subsidence of a river dam and an adjacent highway; both constructed on gypsum-containing rock, southeast of Basel, Switzerland. An observation system was set up to improve the protection of surface and subsurface water resources during remedial construction measures of the highway and in order to understand the processes, as well as the temporal evolution, of rock water interaction (ﬂow and dissolution). However, no detailed hydrogeological information beneath the river could be derived from the previous investigations. To supplement the basic knowledge on this area, underwater Electrical Resistivity Tomography (ERT) measurements were conducted in the river bed upstream of the dam. The ERT-data are interpreted together with drill-core information and a conceptual 3D-Model of the area behind the dam and beneath the river. Results help to delineate weathered zones, associated faults and the thickness of sediment deposits behind the dam, as well as to locate voids within the local karst system. The combination of the ERT and modeling allows the optimization of future site-speciﬁc remedial construction measures. Keywords: Dam site • Subsidence • Karst evolution • Conduit development • Underwater ERT © Versita sp. z o.o.

1. Introduction

present hydrogeophysical investigations derived from a well-developed long-term hydrogeological research site for gypsum karst [3,4]. The common procedures for controlling gypsum karst de- Infrastructures constructed on soluble geologic formations velopment beneath subsided infrastructure are deep cut- are prone to subsidence [1, 2]. Karst features develop off trenches backfilled with impermeable material, close- much more rapidly when found within gypsum-bearing for- spaced grout curtains that are intended to fill all cavities, mations than they do in carbonate formations. While the as well as piling measures to support the infrastructure. characterization and modeling of flow in heterogeneous Such remedial measures profit from the understanding of and porous media has been investigated intensively, we ∗ the principal processes that lead to karstification and the subsequent localized weathering of rock and preferential E-mail: [email protected] flow paths (conduits) to some degree. 81 Investigating sediments and rock structures beneath a river using underwater ERT

Non-destructive geophysical methods can result in a more comprehensive and detailed site characterization than could be achieved by drilling alone, especially in complex environments such as karst areas and unstable sites where invasive techniques, such as drilling, cannot be performed [5–9]. ERT surveys also play an important role in investigating water-covered areas; as they can eﬃciently produce continuous underground images [10–12]. In this study, the results of underwater ERT within a river bed upstream of a dam are presented. At the site, subsurface evaporite dissolution has led to subsidence of a river dam and an adjacent highway, both constructed on gypsum-containing rock. Furthermore, the integration of underwater ERT into a 3D model of the river sediment deposits and the underlying weathered gypsum rock is presented. Results are discussed together with previous hydrogeological and hydrogeophysical investigations [3, 4] to describe the current state of the rapidly developing gypsum karst. 2. Settings and previous investigations

The area of investigation is located along the Birs River Figure 1. Investigation area in the urban agglomeration of Basel southeast of Basel, Switzerland (Fig. 1). The hydrology showing the remedial measures as well as ERT profiles. is strongly affected by an artificial river dam and the pro- duction of hydropower from a small hydro-electric power plant. The dam, as it is today, was constructed in the 1890’s [13]. However, documentation of anthropogenic im- pacts in this region, including the deviation of water for a.s.l. River-groundwater interaction is dominated by the early manufacturingth business in Basel, reaches back as hydraulic river head and variations of river bed conduc- far as the 11 century [14]. tance upstream of the dam during flood events. Upstream of the dam, river water infiltrates into the highly The stratigraphic column includes the lithological se- permeable fluvial gravel and into the weathered bedrock, quences for the geological and hydrogeological model- where it follows the hydraulic gradient around and being and extends from the Quaternary river gravel to the neath the dam and exfiltrates downstream into the river. Gipskeuper sequence (Fig. 2). Quaternary gravel, silty These processes enhance karstification in the soluble com- flood deposits, as well as artificial fillings beneath the ponents of the Gipskeuper and result in an extended highway overlie the Triassic and Jurassic strata on the weathering zone within the bedrock as well as in the de- right side of the river. On the map in Figure 2, the Qua- velopment of preferential flow within cavities. Such fea- ternary sequence has been removed, and the complex pat- tures play a major role in the evolution process of a rapidly tern of lithological changes in some parts of the investiga- evolving karst system. As a result, these processes have tion area is illustrated. These sequences consist of marls enhanced karstification in the soluble components of the and clays (Obere Bunte Mergel), dolomites (Gansinger gypsum-containing rock and have led to the subsidence Dolomit) and sandstones, marls (Schilfsandstein/Untere of the dam and the highway of up to 21 cm over the Bunte Mergel), and, for most of the investigation area, last 30 years. To prevent further subsidence, construction Gipskeuper. Gipskeuper is made up of a series of evap- measures were carried out in two major project phases in orite layers and intercalations of marls. The lithological 2006 and 2007 [3, 4]. term Gipskeuper as used in this paper generally includes The vertical drop to the base level downstream of the dam the mineral gypsum and also refers to anhydrite, which, in is 7.3 m. As there is sufficient water supplied by the the deeper subsurface, is the more common anhydrous form Birs River, the pondage is practically constant at 266.2 m of calcium sulfate. In its non-weathered appearance the 82 J. Epting, A. Wüest, P. Huggenberger

Figure 2. Geological, tectonic map with removed Quaternary sequence (modified after unpublished data, Pfirter [49]), as well as lithostratigraphy, hydrostratigraphy, modeled geological units (modified after Bitterli-Brunner et al. [50], Gürler et al. [51], Pearson et al. [52], Spottke et al. [53]) and longitudinal cross section. Expression of regional geological formations: Obstusus Tone – clay; Arietenkalk, Rhät – limestone; Untere and Obere Bunge Mergel - greyish to reddish marls and clays; Gansinger Dolomit – Dolomite; Schilfsandstein – sandstone; Gipskeuper – gypsum Keuper.

Gipskeuper is characterized as having a low permeabil- wells for groundwater or subsidence measurements. In to- ity, however, with its weathered appearance, Gipskeuper tal, 12 observation wells were fitted with automatic data can be considered as a heterogeneous (karstified) aquifer. loggers for monitoring the following physical parameters: Areas below the dam and the highway are strongly weath- hydraulic head, temperature and electrical conductivity. ered due to gypsum dissolution in the Gipskeuper and are Additional lithostratigraphic information could be derived loosened over several meters of thickness. from reports made during the installation of the piles. Hy- The investigation area is characterized by the Eastern draulic connections and flow velocities within the inves- Rhinegraben Master fault accompanied by an◦ intense tec- tigation area− were investigated by a dye tracer test in tonic segmentation into compartments [15]. The Triassic 1996. Maximal1 groundwater flow velocities ranged from strata dip at an angle of approximately 45 to the West 85 to 111 md ; values typical for conduits within well- and are subdivided by a series of NNE-SSW normal faults developed mature karst systems [16]. (Fig.2). Fracture zones are associated with rock weakness and can locally increase permeability within sequences, The first investigational studies beginning in the 1990’s resulting in enhanced groundwater leakage and the de- focused on drilling campaigns providing 1D lithostrati- velopment of paths for preferential flow (Tectono-karstic graphic profiles from borehole logs. The primary purpose conduits). for drilling boreholes was to find significant permeable Multiple data sources were available for the investiga- zones within the underlying bedrock and existing cavi- tion area: (1) lithostratigraphic profiles from borehole ties. Although the probability of encountering cavities is logs as well as coarse geological information on piling fairly low and relies on a hit-or-miss approach, in total, measures and locations with supplementary cement injec- 7 cavities with diameters ranging from 0.3 to 2.7 m were tion (Fig.1); (2) continuous groundwater measurements detected at a depth of 15 to 18 m. The borehole data (Fig.1); (3) results from dye tracer tests; and (4) the na- suggests that the thickness of the weathered Gipskeuper tional geological map. A total of 24 vertical boreholes ranges from 2.2 to 14.8 m. Additionally, in the 1990’s were drilled in several investigation phases from 1993 several boreholes hydraulically connected aquifers. The to 2007. Most boreholes were equipped as observation connection is documented by drill-core protocols and was 83 Investigating sediments and rock structures beneath a river using underwater ERT

recently confirmed by geochemical and hydraulic data (Ta- rounding media as well as an interpretation and/or assign- ble1;[3]). ment of changes in physical parameters. Each method has The borehole data suggest that the occurrence of cavities, limitations in depth of exploration and resolution, depend- and consequently the development of conduits, is concen- ing on the settings. trated at the base of the weathered Gipskeuper (lixiviation Electrical Resistivity Tomography (ERT) can be effective front), where most of the solution cavities were encoun- in settings where faults and “weak” zones beneath the wa- tered. The majority of the encountered cavities contain ter layer can be expected. Among others, Dahlin [17], Loke clay, gravel and calcite fillings. During episodic floods, and Barker [18], Donner [19], Pellerin [20] and Khalil [21] these fillings are partially flushed, allowing more aggres- describe various ERT applications for environmental sci- sive water to enter the system. Consequently, the devel- ences and hydrological questions. Geophysical mapping opment of conduits occurs in response to the flooding of with ERT has been successful for investigating and map- passages. The map in Figure 2 also shows the course of ping features in karst terrains (e.g. [22–25]), exploring the Birs River in 1798 compared toth the situation in 1983. shallow subsurface cavities [26–28], within complex ge- The river was straightened in the 19 century and cut into ological areas [29, 30], and in urban areas (e.g. [31]). Fur- the Triassic bedrock, resulting in a narrow couloir. thermore, numerous ERT investigations have focused on However, such intrusive exploration methods deliver lim- dam leakage (e.g. [32]) and buried paleochannels [33, 34]. ited information on the geometry of karst features and Active faults are often imaged as low resistivity zones their connectivity; in addition they are costly, time- (e.g., [35–37]), so continuous imaging of resistivity struc- consuming and can leave hydraulic connections. In the ture can be a powerful tool for mapping active fault zones. current case study, the drilling of several boreholes re- In addition, since a resistivity survey is sensitive to ver- sulted in a connection of groundwater-bearing horizons in tical structure, the delineation of the location of vertical the gravel deposits and of the easily soluble evaporitic weak zones is auspicious. bedrock. This caused the accelerated vertical groundwa- For ERT surveys in water-covered areas, electrodes are ter movements to locally stimulate karstification. installed on the lake or river bed, or floated on the water Further hydrogeological and hydrogeophysical investiga- surface, however, direct contact with the earth is favorable tions, including groundwater and karst evolution model- as it amplifies the current into the earth, and with this ing [3, 4, 16] within the investigation area on the river the sensitivity to subsurface anomalies. Kwon et al. [38] bank, under different hydrologic and geotechnical bound- used ERT in a river to detect faults by employing floating ary conditions, both before and after the construction re- electrodes as well as electrodes on the river bed. In ad- mediation measures, allowed− the description of: (1) pref- dition to their field case study, Kwon et al. [38] presented − erential flow in the shallow subsurface1 (calculated leak- a numerical study evaluating the potential of underwater age rate1 range between 25 ls during base flow and up to ERT surveys when using floating electrodes or electrodes 90 ls during major flood events [4]); (2) zones that are re- placed at the bottom of the river bed with regards to elec- lated to groundwater flow around the dam, including flow trode spacing, sensitivity and electrode spreads. Nyquist dynamics; (3) drainage phenomena of karst features such et al. [39] used stream bottom ERT to map groundwater as cavities and conduits; (4) the weathering horizon within discharge and assess groundwater-surface water interac- the Gipskeuper; (5) near-surface faults and fractures that tion within streams. They show that patterns of groundwa- are associated with “weak” zones; and (6) buried pale- ter discharge can be mapped at the meter scale. Crook et ochannels. However, no information on the thickness of al. [40] used ERT imaging to characterize the architecture sediment deposits, weathered zones, or karst and fault of substream sediments, whereby they used electrodes in- features within the gypsum rock beneath the river behind stalled on the river bed. Their results provide an estimate the dam could be derived from these investigations. The of the sediment volume behind a log jam. current study focused on filling this deficit by applying In this case study, underwater ERT measurements were further non-destructive methods. conducted in the river bed upstream of the dam to sup- 3. Methods plement the basic knowledge within the outlined gypsum karst research site. 56 underwater graphite electrodes with a fixed spacing of 1 m for all measurements limit the maximal profile line to 55 m (Advanced Geosciences, Inc. There are a number of geophysical techniques which are (AGI), Sting/Swift R1 resistivity meter). From Loke [41], applicable to investigations of geological structures near the maximum prospective depth for the Wenner setup can the surface (e.g. [3]). All methods are based on mapping be calculated by multiplying the length of the profile line specific physical contrasts between the target and the sur- with a factor of 0.173. The minimum length of a profile 84 J. Epting, A. Wüest, P. Huggenberger

Table 1. Compilation of borehole information (Epting et al. 2009a).

OW monitoring filter section in status from (m a.s.l.) to (m a.s.l.) difference 1 2 3 inclinometer Gipskeuper void (filled) 259.48 257.18 2.30 - fluvial gravels and Gipskeuper connection sealed - - - 4 head, T, EC fluvial gravels and Gipskeuper (sep- connection unclear, siphon mechanism - - - 5 arate) 6 head, T, EC fluvial gravels and Gipskeuper void, connection 250.92 248.62 2.30 7 - Gipskeuper void (filled), connection sealed 254.20 251.50 2.70 8 - - void (filled), connection sealed 253.60 252.80 0.80 9 head, T, EC Gipskeuper - - - - 10 - - sealed - - - 11 head, T, EC fluvial gravels and Gipskeuper void (filled), connection 257.05 256.25 0.80 12 head, T, EC fluvial gravels and Gipskeuper connection, siphon mechanism - - - 13 - - sealed - - - 14 head, T, EC fluvial gravels siphon mechanism - - - 15 head, T, EC fluvial gravels and Gipskeuper connection unclear, siphon mechanism - - - 16 head artificial fillings and Gipskeuper connection unclear, siphon mechanism - - - 17 - - sealed - - - 18 head unweathered rock dry - - - 19 - - sealed - - - 20 head slope clay to Obere Bunte Mergel - - - - inclinometer - - - - - 21 head, T, EC weathered Gipskeuper void 252.17 250.77 1.40 void 249.87 249.57 0.30 66 head artificial fillings, Schilfsandstein, connection - - - 69 Gipskeuper 72 ------

Table 2. Compilation of underwater ERT profiles. Grey: ERT Profiles that were combined to one 220 m long longitudinal profile line is 28 m (constrained by the river width) and the max- (We 1 to 7). The Dipole-Dipole layout was tested for comparison of layout types. We 8 is the closest profile to the imum length of the profile line is 55 m (constrained by dam followed by We 9, 10 and 11 (see Fig. 1) the electrode layout), resulting in an exploration depth of approx. 5 to 10 m. profile date length (m) comment Reaching this depth allowed the entire thickness of the Longitudinal profiles parallel to the river sediment deposits to be investigated down to the weathering zone, as documented by the boreholes and the depth We1 07.08.2008 55 Wenner layout of the subsurface structures of the dam [3, 4]. The rela- We2 06.08.2008 55 Wenner layout tively heavy electrodes and anchors at the end of the elec- We3 06.08.2008 55 Wenner layout − trode cables sufficed to secure the installation on the river Dd1 06.08.2008 55 Dipole-Dipole layout − bed. All measurements were carried3 1 out during low river We4 22.08.2008 55 Wenner layout discharge3 1 situations (4 to 7 m s , average discharge is We5 22.08.2008 55 Wenner layout 15 m s ) in July and August 2008 (Table 2). The topog- We6 22.08.2008Transverse profiles 55 perpendicular Wenner layout to the river raphy of the river bed and depth of the water along the We7 22.08.2008 55 Wenner layout survey lines was also recorded. Figure 1 shows the locations for the various ERT profiles We8 07.08.2008 49 Wenner layout lines within the river behind the dam, where the longitudi- We9 06.08.2008 43 Wenner layout nal profile line of 220 m (parallel to the river) - (Table 2). We10 22.08.2008 40 Wenner layout The single profiles were recorded 2 to 4 m from the right We11 02.07.2008 28 Wenner layout, not covering the total river width river board. The length of the four transverse profile lines across the river was adapted to the width of the river bed 85 Investigating sediments and rock structures beneath a river using underwater ERT

varying between 28 and 49 m (Table2). The transverse profile lines could not be continued outside the river because of the steep highway dam. Reynolds [42] summarizes the strengths and weaknesses of the commonly used electrode arrangements (arrays) for ERT (Wenner, Schlumberger and Dipole-Dipole). Gener- ally, the Wenner setup was chosen because of the high Figure 3. Distribution of measured resistivity values in comparison signal strength obtained within areas where major back- to the results of other authors. ground noise is expected and the good resolution of vertical changes in the subsurface. Also, Wenner arrays were the best in terms of signal to noise ratio [11]. A con- siderable disadvantage of the Wenner setup is the poor resolution of horizontal changes in the subsurface. One of the longitudinal profiles (Table 2) was also carried out the dam and beneath the river. The conceptual 3D-Model with a Dipole-Dipole setup; the gain in capturing horizon- was modeled using GoCAD© (Geological Objects 47-48). tal changes in the subsurface was only slightly improved. 4. Results Therefore, the Wenner setup was favored. Post-processing and data interpretation was carried out using the inversion program RES2DINV [43]. It automat- Figure3 shows the distribution of measured resistivity ically determines topographically corrected 2D electrical values in comparison to the results of other authors. Gen- resistivity models of the subsurface by inverting the data erally, two resistivity zones or layers can be distinguished obtained from electrical imaging [30]. A robust inversion (Fig.4): a resistive (100 to 500 Ωm) surface layer corre- was used as it is more suitable for detecting and sharp- sponding to the streambed sediments as well as two ver- ening linear features such as faults and contacts within tical structures corresponding to karst and or fault zone such complex geological settings of karst regions [3]. The fillings and a conductive (10 to 40 Ωm) bottom layer cor- inversion parameters were kept constant in order to retain responding to the bedrock. Nyquist et al. [38] identified the comparability of the various measurements. The in- three resistivity layers in similar settings: a resistive (100 version parameters for RES2DINV are summarized in the to 400 Ωm) surface layer corresponding to the streambed appendix. sediments, a conductive (20 to 100 Ωm) middle layer cor- As a large proportion of the current flows through the wa- responding to residual clay sediments, and (unrelated to ter layer, electrical resistivity in water-covered areas is the current case study) a resistive (100 to 450 Ωm) bottom greatly influenced by the water depth. Therefore, elec- layer corresponding to the carbonate bedrock. Crook et trical resistivity and geometry of the water column must al. [39] could show that, for similar settings, the bound- be known for an accurate robust inversion [12, 39, 44, 45]. ary between the lower two resistivity regions correlates The electrical resistivity and geometry of the water col- well with the interface previously logged between the al- umn is fixed in the earth model and the inversion program luvial gravels (95 to 1500 Ωm) and underlying weathered attempts to determine the electrical resistivity of the cells limestone (10 to 75 Ωm). that would most accurately reproduce the observed elec- For all measurement values of the apparent resistivity, the trical resistivity measurements [46]. The electrical con- calculated Root Mean Square (RMS) error was less than − ductivity ofµ the river water was measured by digital con- 5%, indicating that the measurements were undisturbed ductivity meter for1 each field day and varied between 390 and that the electrical resistivity models resulted in one − − and 417 Scm . This results in an average electrical plausible solution [19]. resistivity of1 the1 water column of 25 Ωm (conversion 1 Ωm Figure4 shows a three-dimensional fence diagram of the equals 1 S m ). The water layer within the investiga- inversion results from the longitudinal profile combined tion area varied between 0.4 and 1.85 m. As the height with the transverse profiles and the lithostratigraphic in- of the ponded water upstream of the dam is practically formation of 4 boreholes. The results of the ERT mea- constant at 266.2 m a.s.l. the depth variations of the wa- surements show that the underground can be divided into ter layer are brought about by the topography of the river two layers, a water-saturated layer of sediments with high bed, which was measured after installing the electrodes. resistivity and a weathered rock layer with low resistivity. The ERT-data are interpreted together with drill-core in- Furthermore, two anomalies in the ERT results indicate formation and a conceptual 3D-Model of the area behind vertical structures with high resistivity. The interpreta- 86 J. Epting, A. Wüest, P. Huggenberger

Figure 4. 3D fence diagram of ERT results illustrated together with lithostratigraphic information in boreholes (all axes in m). The topmost blue color stands for the water layer.

tion of the results of the ERT measurements are covered 2 to 3 m and correlate with the lithostratigraphic informa- in the discussion. tion derived from the boreholes. The thickness of the sed- 5. Discussion iment deposits increases from upstream regions towards the dam. These findings correlate with the progression of the surface of the gypsum rock (c.f. Fig. 2). (2) The bottom of the sediment deposits corresponds to an The discussion is focused on the occurrence of two ERT erosion surface of the river preceding the dam construc- anomalies: (1) karst features such as cavities, conduits, tion. The undulating interface of low and high electrical fractures and fault zones generally result in an increase in resistivity values might be explained by the former pro- electrical resistivity if they are filled with air (near-infinite gression of paleochannels and pool sequences. Findings electrical resistance), and, providing there is a electrical agree with the results of the previous ERT measurements resistivity contrast with the surrounding rock, a decrease and information from geological as well as historical maps. if they are filled with clay and water. Although clay frac- (3) Zones with relatively low electrical resistivity values tions will decrease electrical resistivity more than water, up to a maximal 40 Ωm, beneath the sediment deposits, in the field their influence cannot be determined due to could be associated with the weathered gypsum. Low the shape of the features and the fact that the degree of electrical resistivity values result from water with high the filling is, most of the time, unknown; and (2) electrical solution content within water-saturated clays. Especially resistivity contrasts between various sedimentological se- within these zones, the weathering process resulted in the quences and their degree of weathering. In the following, removal of gypsum and the remains of the clay component. the observed features are discussed in detail: Zones with comparably higher electrical resistivity values (1) Zones with relatively high electrical resistivity values between 40 and 100 Ωm on the western part of the first above 100 Ωm are associated with sediment deposits be- transverse profile could be explained by the more resis- hind the river dam. The rather high electrical resistivity tant Schilfsandstein (c.f. Fig. 2). Findings correlate with values could be explained by the existence of coarse fluvial the lithostratigraphic information derived from the bore- gravel. The thickness of the fluvial sediments range from holes at the river board and information from the geo- 87 Investigating sediments and rock structures beneath a river using underwater ERT

Figure 5. Conceptual 3D-Model derived from ERT results and interpreted lithological and fault features. A (view from NW): ERT-Profiles and drill-core information; B (view from NW): ERT-Profiles, drill-core information and interpreted lithology of the bedrock (pink: Gipskeuper; violet: Schilfsandstein; cf. Fig. 2); C (view from SW): ERT-Profiles, drill-core information, interpreted lithology of the bedrock (transparent pink: Gipskeuper; transparent violet: Schilfsandstein; cf. Fig. 2) illustrated together with faults and the progression of the Gipskeuper- Schilfsandstein interface (dark and light grey, respectively); D (view from NW): ERT-Profiles, drill-core information, interpreted lithology of the bedrock (transparent pink: Gipskeuper; transparent violet: Schilfsandstein; cf. Fig. 2) illustrated together with faults and the progression of the Gipskeuper- Schilfsandstein interface (dark and light grey, respectively); and E (view from NW): water- (blue), river bed- (brown) and bedrock- surfaces (pink and violet) illustrated together with faults and the progression of the Gipskeuper- Schilfsandstein interface (dark and light grey, respectively).

88 J. Epting, A. Wüest, P. Huggenberger

6. Conclusions logical map. As the underwater ERT measurements only reach to a maximum depth of 10 m, the detection of sharp boundaries between weathered and non-weathered zones The conducted underwater ERT measurements and high was not possible. electrical resistivity contrasts in the subsurface enabled (4) The subsurface image shows that the high resistivity the separation of the river sediment deposits from the un- anomalies develop vertically under the river bed. Within derlying weathered gypsum rock. Furthermore, results en- the longitudinal profile at approx. 50 m, as well as in the abled the description of the progression of paleochannels second transversal profile at approx. 18 m, regions can be and pool sequences, as well as karst features in combi- observed where the interpreted weathered gypsum rock is nation with the local fault system. Through the analysis vertically cut through and high electrical resistivity values of the images from five resistivity profiles anomalies ap- occur. The reason for these structures beneath the river peared to be connected to each other and the strike di- bed could be gravel-filled sinkholes, possibly in combi- rection could be derived. These results help to delineate nation with the local fault system. Similar karst features the thickness of sediment deposits behind the dam, and were observed in the river bed further south. Furthermore, to locate distinct karst features that promote preferential this karst feature could be part of a conduit system that flow, which play a major role in the karst evolution pro- reveals a siphon mechanism according to the hydrological cess. The observations suggest that the karst system is characteristics of the river stage as has been described already in a well-developed, mature state. in Epting et al. [4]. The location of the anomaly on the The main limitations of underwater ERT measurements in- longitudinal profile correlates with a mapped small-scale clude the restricted investigation depth for cross sections fault on the surface which is associated with large-scale in narrow rivers and the requirement of high electrical re- fault activities (Fig. 2). The location of the anomaly on sistivity contrasts in the subsurface. For the present case the second transversal profile correlates with the mapped study, the applicability was supported by previously con- boundary of lithofacies (Fig. 2). Both locations interpreted ducted surface ERT measurements that already indicated as karst features or faults are associated with geologically high subsurface heterogeneity and electrical resistivity weak zones. Both anomalies extend down to the bottom contrasts. For cases where structural subsurface hetero- of the resistivity image, which implies that the structures geneity but no electrical resistivity contrasts are to be extend much deeper. expected, as in coarse gravel environments with low elec- Figure 5 shows the final interpretation5A result as a concep- trical conductivity and water with low mineralization con- tual 3D-Model derived from ERT results and interpreted tent, other hydrogeophysical investigation methods, such lithological and fault features. Figure 5Bshows the loca- as GPR (Ground Penetrating Radar) could be employed. tion of the longitudinal and four transversal ERT-profiles Due to the multiple data sources and hydraulic data from within the modeled 3D body. Figure illustrates the observation wells and high-resolution 3-D hydrogeologi- ERT-profiles and the5C interpolated surface of the bedrock, cal models, it was possible to partially eliminate ambigu- including the lithologies of the Gipskeuper and the Schil- ity in data interpretation and to describe the relationship fsandstein. Figure shows the progression of the East- between the different observed features in a spatial con- ern Rhinegraben Master fault together with two small- text. scale faults derived from the geological map (Fig. 2). The The results give the opportunity to optimize future inves- small-scale fault in front was extended according to the tigations and remedial construction measures to extend vertical anomaly◦ observed in the longitudinal ERT-profile existing observation networks for subsidence monitoring (see above). The derived strike direction of this anomaly and to stabilize the dam; including the implementation of is approx. 300 WNW, and corresponds to to the strike optimal localized grout curtains and stabilizing piles. The direction of the mapped fault further to the south. Both regions where fault or karst features were detected should faults indicate a steep dip. Furthermore, the progression be especially considered as “weak” zones and reinforce- of◦ the interface of the lithologies of the Gipskeuper and ment prioritized. However, during remedial construction the Schilfsandstein is visible. An approximate dipping5D of measures and reinforcements water protection issues must 45 to the West that was derived from geological mapping be considered in an appropriate way. is also indicated within the ERT-profile. Figure illustrates the various modeled surfaces, including the water-, This subsurface information now allows the adaptation of river bed- and bedrock- surfaces as well as faults and the geometries and distribution of aquifer properties in the progression of the Gipskeuper-Schilfsandstein inter- existing hydrogeological karst evolution models that were face. The volume of3 sediment being held behind the dam set up for process modeling and to document the develop- is approx. 3.2E04 m . ment of the karst system. 89 Investigating sediments and rock structures beneath a river using underwater ERT

Acknowledgements

2000,Technische Universität Berlin [9] Fenning P.J., Brown A.J., Nind D., Geophysical sur- The authors would like to thank Rebecca Page for lan- veys across a ground subsidence feature, in Proceed- guage editing the manuscript and Horst Dresmann for as- ings for SAGEEP 2000, Washington D.C., 2000, 857- sisting in the setup of the conceptual 3D-Model in Go- 866 CAD. The field surveys were supported by Fabian Wigger, [10] Chung H.J., Kim J.H., Park K.P., Kwon H.S., Choi Johannes Darnutzer, Max Respondek, Piotr Halicki, An- H.S., Kim K.S., Kim J.S., Application of geophys- nette Affolter, Rebecca Page and Hanspeter Wüest. ical results to designing bridge over a large fault: Geotechnical Problems in Asian Countries, Proceed- References ing of Fourth Asian Young Geotechnical Engineers Conference, 2001, 45–48 [11] Kim J.H., Yi M.J., Song Y., Cho S.J. Chung S.H., Kim K.S., DC resistivity survey to image faults beneath [1] Gutiérrez F., Gypsum karstification induced subsi- a(SAGEEP) riverbed: Symposium on the Application of Geo- dence: effects on alluvial systems and derived geophysics to Engineering and Environmental Problems hazards (Calatayud Graben, Iberian Range, Spain). , 2002, 13IDA10 Geomorphology, 1996, 16, 277-293 [12] Cho S.J., Kim J.H., Yi M.J., Lee S.K., Son J.S., Sato [2] Lamont-Black J., Younger P.L., Forth R.A., Cooper M., EffectiveProc. 110 methodologyth SEGJ Conference of DC resistivity survey for A.H., Bonniface J.P., A decision-logic framework for imaging the underground structure at water-covered investigating subsidence problems potentially at- area: , 2004, 180–183 tributable to gypsum karstification. Engineering Ge- [13] Golder E., 100 years Birswuhr Neue Welt (in ology, 2002, 65, 205-215 German). 1984, Baudepartement Basel-Stadt, Tief- [3] Epting J., Huggenberger P., Glur L., A Concept for bauamt Integrated Investigations of Karst Phenomena in Ur- [14] Fechter D., Topography with the consideration of the ban Environments - Merging geophysical and hy- cultural and conventionalth history (in German). In: drometrical investigations with 3D hydrogeological Basel in the 14 century, 1856, Basel modeling for applied urban hydrogeology within a [15] Schmassmann H., Baselbieter and Western Aargauer gypsum karst area. Engineering Geology, 2009, DOI Tafeljura (in German). In: Jäckli H., Kempf T., Er- 10.1016/j.enggeo.2009.08.013. läuterungen zum Blatt Bözberg-Beromünster 1. Hy- [4] Epting J., Romanov D., Huggenberger P., Kaufmann drogeologische Karte der Schweiz, 1972, 93-100 G., Integrating Field and Numerical Modeling Meth- [16] Birk S., Liedl R., Sauter M., Identification of local- ods for Applied Urban Karst Hydrogeology. Hydrol- ized recharge and conduit flow by combined analysis ogy and Earth System Sciences, 2009, 13, 1163– of hydraulic and physico-chemical spring responses 1184 (Urenbrunnen, SW-Germany), J. Hydrol., 2004, 286, [5] Gabbani G., Lavorini G., Pacini L., Geophysical and 179-193, doi:10.1016/j.jhydrol.2003.09.007 geotechnical investigation of an areal landslide in the [17] Romanov D., Kaufmann G., Hiller T., Karstification Tuscan Northern Apennines (Italy) with an extensive of aquifers interspersed with non-soluble rocks From consolidation design. In: Proceedings for the EEGS- basic principles towards case studies. Engineering ES 2000 Annual Meeting, 2000, Bochum, Germany Geology, 2010, DOI 10.1016/j.enggeo.2010.09.008 [6] Lapenna V., Perrone A., Pietro L., Piscitelli S., Rizzo [18] Dahlin T., 2D resistivity surveying for environmental E., Sdao F., 2D resistivity and 3D self-potential to- and engineering applications. First Break, 1996, 14, mographies in the study of the Giarrossa landslide 275-283 (southern Italy). In: Proceedings for the EEGS-ES [19] Loke M.H., Barker R.D., Rapid least-squares inver- 2000 Annual Meeting, 2000, Bochum, Germany sion of apparent resistivity pseudosections using a [7] Sretenovic B., Lokin P., Sretenovic N., Djordjevic M., quasi-Newton method, Geophys. Prosp., 1996, 44, Kisin S., Some possibilities of electrical scanning 131-152 method in landslides exploration, in Proceedings for [20] Donner F., Modern measurement and evaluation pro- the EEGS-ES 2000 Annual Meeting, 2000, Bochum, cedures for resistivity geoelectrics, with application Germany examples (in German). Zeitschrift für angewandte Ge- [8] Yaramanci U., Kiewer M., Geoelectrical Character- ologie 1997, 42 isation of the Opalinus ClayFormation in the Un- [21] Pellerin L., Applications of electrical and electromag- derground Rock Laboratory ‘Mont Terri’. Mont Terri netic methods for environmental and geotechnical in- Project, Technical Note 99-27, ED-C Experiment, 90 J. Epting, A. Wüest, P. Huggenberger

vestigations. Surveys in Geophysics 2002, 23, 101- ogy 2002, 49, 441-449 132 [35] Maillet G., Rizzo E., Revil A., Vella C., High resolution [22] Khalil M.H., Geoelectric resistivity sounding for de- electrical resistivity tomography (ERT) in a transition lineating salt water intrusion in the Abu Zanima area, zone environment: Application for detailed internal west Sinai, Egypt. Journal of Geophysics and Engi- architecture and infilling processes study of a Rhône neering 2006, 3, 243-251 River paleo-channel. Marine Geophysical Researches [23] Sumanovac˘ F., Weisser M., Evaluation of resistivity 2005, 26, 317-328 and seismic methods for hydrogeological mapping in [36] Fujita K., Ikuta O., Resistivity structureEarth Planets of the centralSpace karst terrains. Journal of Applied Geophysics 2001, part of the Yamasaki fault studied by the multiple 47, 13-28 electrodes resistivity method, , [24] McGrath R.J., Styles P., Thomas E., Neale S., Inte- 2000, 52, 567–571 grated high-resolution geophysical investigations as [37] Unsworth M., Bedrosian P.A., Electrical resistivity potential tools for water resource investigations in structure at the SAFOD site from magnetotelluric karst terrain. Environmental Geology 2002, 42, 552- exploration. Geophysical research letters, 2004, 31, 557 L12S05, doi:10.1029/2003GL019405 [25] van Schoor M., Detection of sinkholes using 2D elec- [38] Kwon H., Kim J., Ahn H., Yoon J., Kim K., Jung C., trical resistivity imaging, Journal of Applied Geo- Lee S., Uchida T., Delineation of a fault zone be- physics 2002, 50, 393-399 neath a riverbed by an electrical resistivity survey us- [26] Kaufmann G., Romanov D., Geophysical investigation ing a floating streamer cable. Exploration Geophysics of a sink in the northern Harz foreland (North Ger- 2005, 36, 50–58 many). Environmental Geology 2009, 58, 401-405 [39] Nyquist J.E., Freyer P.A., Toran L., Stream Bottom [27] El-Qady G., Hafez M., Abdalla M.A., Ushijima K., Resistivity Tomography to Map Ground Water Dis- Imaging subsurface cavities using geoelectric tomog- charge, Ground water 2008, 46, pages 561–569 raphy and ground-penetrating radar. Journal of Cave [40] Crook N., Binley A., Knight R., Robinson D.A., and Karst Studies, 2005, 67, 174-181 Zarnetske J., Haggerty R., Electrical resistivity [28] Leucci G., De Giorgi L., Integrated geophysical sur- imaging of the architecture of substream sedi- veys to assess the structural conditions of a karstic ments, Water Resour. Res., 2008, 44, W00D13, cave of archaeological importance. Natural Hazards doi:10.1029/2008WR006968. and Earth System Sciences 2005, 5, 17-22 [41] Loke M.H., Electrical imaging surveys for environ- [29] Soupios P., Papadopoulos I., Kouli M., Georgaki I., mental and engineering studies. 2000 Vallianatos F., Kokkinou E., Investigation of waste [42] Reynolds J.M., An Introduction to Applied and Envi- disposal areas using electrical methods: a case study ronmental Geophysics. 1997, Wiley, New York. from Chania, Crete, Greece. Environmental Geology [43] Loke M.H., Res2dinv Software User‘s Manual, Ver- 2007, 51, 1249-1261 sion 3.57. Geotomo Software, Penang, Malysia, 2007 [30] Griffiths D.H., Barker R.D., Two-dimensional resistiv- [44] Kim J.H., Yi M.J., Song Y., Cho S.J., Lee S.K., Son J.S., ity imaging and modelling in areas of complex geol- Chung S.H.,Mulli-tamsa A study on the DC resistivity method to ogy. Journal of Applied Geophysics 1993, 29, 211-226 image the underground structure beneath river or lake [31] El-Hussain I., Holbrook J., Sneed C., Integrated geo- bottom: , 2002, 5, 223–235 (in Korean physical and geological methods to delineate buried with English abstract) paleochannels in the New Madrid seismic zone of [45] Snyder D.D., MacInnes S.C., RaymondSymposium M.J., Zongeon the southeastern Missouri. In: Proceedings for SAGEEP K.L.,Application Continuous of Geophysics resistivity to profiling Engineering in shallow and Envi- ma- 2000, Washington D.C., 2000, 253-262 rineronmental and fresh Problems water environments: (SAGEEP) [32] Wise’n R., Bjelm L., Dahlin T., Resistivity imaging as a preinvestigation methodth in urban environments. , 2002, 13GSL4 Proceedings EEGS-ES 6 Meeting, 2000, CH05 [46] Loke M.H., Lane J.W. Jr Inversion of data from electri- [33] Al-Saigh N.H., Mohammed Z.S., Dahham M.S., De- cal resistivity imaging surveys in water-covered ar- tection of water leakage from dams by selfpotential eas. Explor Geophys, 2004, 35, 266–271 method. Engineering Geology 1994, 37, 115-121 [47] Mallet J.L., Discrete Smooth Interpolation in Geo- [34] Baines D., Smith D.G., Froese D.G., Bauman P., metric Modeling. Computer-Aided Design 1992, 24, Nimeck G., Electrical Resistivity Ground Imaging 178-191 (ERGI): a new tool for mapping the lithology and ge- [48] Mallet J.L., Geomodeling. 2002, Oxford University ometry of channel-belts and valley-fills. Sedimentol- Press, New York.

91 Investigating sediments and rock structures beneath a river using underwater ERT

[49] Pﬁrter U., Geologie und Morphologie des Birstals [52] Pearson F.J., Balderer W., Loosli H.H., Lehmann B.E., von St. Jakob bis Münchenstein. Diplomarbeit Matter A., Peters T., Schmassmann H., Gautschi A., Geologisch-Paläontologisches Institut, Universität Applied Isotope Hydrogeology-a case study in north- Basel (in German), 1973 ern Switzerland. 1991, Elsevier, Studies in Environ- [50] Bitterli-Brunner P., Fischer H., Explanations map mental Science 43, 439pp. Blat Arlesheim 1067 (in German). Geological Atlas [53] Spottke I., Zechner E., Huggenberger P., The south- of Switzerland, 1989 eastern border of the Upper Rhine Graben: A 3D [51] Gürler B., Hauber L., Schwander M., Geology of the geological model and its importance for tectonics and surrounding area of Basel with suggestions usage op- groundwater ﬂow. International Journal of Earth Sci- tions of geothermal energy (in German). In: Beitrag ences 2005, 94, 580-593 zur Geologischen Karte der Schweiz, 1987, N.F. 160, 33p. Appendix: Inversion settings for the underwater ERT data processing in RES2DINV.

Parameter Value Initial damping factor Minimum damping factor Line search option 0.16 Convergence limit for relative change in RMS error 0.015 Number of iterations Always Vertical to horizontal flatness filter ratio 0.4% Model for increase in thickness of layers 5 Number of nodes between adjacent electrodes 1 Flatness filter type, Include smoothing of model resistivity Default Reduce number of topographical datum points 4 Carry out topography modeling Model changes only Type of topography trend removal No Type of Jacobian matrix calculation Yes Increase of damping factor with depth Average Type of topographical modeling Gauss-Newton Robust data constrain 1.1 Cutoff factor for data constrain None Robust model constrain Yes Cutoff factor for model constrain 0.05 Allow number of model parameters to exceed datum points? Yes Use extended model? 0.02 Reduce effect of side blocks? Yes Type of mesh No Optimize damping factor? No Time-lapse inversion constrain Normal Type of time-lapse inversion method No Thickness of first layer None Factor to increase thickness layer with depth Simultaneous Use finite element method 0.5 Width of blocks 1.1 Make sure blocks have the same width Yes RMS convergence limit Normal width Use logarithm of apparent resistivity Yes Type of inversion method 1% Proceed automatically for sequential method Yes IP damping factor Concurrent Use automatic IP damping factor No Limit resistivity values No Upper limit factor No Lower limit factor Yes Type of reference resistivity 50 0.02 Average 92 J. Epting, A. Wüest, P. Huggenberger

Model refinement Combined Marquardt and Occam inversion Type of optimization method Normal Convergence limit for Incomplete Gauss-Newton method Not Use data compression with Incomplete Gauss-Newton Gauss-Newton Use fast method to calculate Jacobian matrix. 0.003 Use higher damping for first layer? No Extra damping factor for first layer Yes Type of finite-element method No Factor to increase model depth range 2.5 Resistivity variation within water layer Triangular Optimize Jacobian matrix calculation 1 Automatically switch electrodes for negative geometric factor Minimize variation Force resistance value to be consistent with the geometric factor No Shift the electrodes to round up positions of electrodes Yes No No