Abstract
The Lower Rhine Graben (Central Europe) is a prime example of a seismically active low-strain rift zone characterized by pronounced anthropogenic and climatic overprint of structures, and long recurrence intervals of large earthquakes. These factors render the identification of active faults and surface ruptures difficult. We investigated two fault scarps in the Lower Rhine Graben, to decipher their structural character, offset and potential seismogenic origin. Both scarps were modified by anthropogenic activity. The Hemmerich site lies c. 20 km SW of Cologne, along the Erft Fault. The Untermaubach site lies SW of Düren, where the Schafberg Fault projects into the Rur River valley. At the Hemmerich site, geomorphic and geophysical data, as well as exploratory coring reveal evidence of repeated normal faulting. Geophysical analysis and palaeoseismological excavation at the Untermaubach site reveal a complex fault zone in Holocene gravels characterized by subtle gravel deformation. Differentiation of tectonic and fluvial features was only possible with trenching, because fault structures and grain sizes of the sediments were below the resolution of the geophysical data. Despite these issues, our investigation demonstrates that valuable insight into past earthquakes and seismogenic deformation in a low-strain environment can be revealed using a multidisciplinary approach.
Many of the state-of-the-art techniques in tectonic geomorphology and palaeoseismology were first developed and refined in tectonically active, arid regions with a high level of seismic activity (e.g. Oldham 1909; Wallace 1977, 1986; Sieh 1978; Crone 1985; Sieh et al. 1989; Crone et al. 1997; Friedrich et al. 2004). In such environments, a combination of high-resolution optical and terrain satellite-data analysis and field mapping is often sufficient to map the surface expression of potential active faults, to enable first-order tectonic studies and to identify sites suitable for palaeoseismological trenching or drilling. In contrast, in humid, soil-mantled regions the unambiguous recognition of seismogenic surface manifestations is challenging as surface erosion, runoff, periglacial processes and vegetation may modify scarp morphology (Fig. 1) or produce similar morphological features (e.g. Pierce & Colman 1986; Bull 2008). This problem is further compounded by urbanization and agricultural land-use, which may accelerate the destruction of geomorphic evidence of earthquake ruptures (e.g. Puzachenko et al. 1990; Almond et al. 2010). In addition, trench site selection in these regions is complex, because it is often influenced by legal issues, land ownership and costly excavation on cropland to reduce economic impact. These problems may be exacerbated in those tectonically active continental interiors that are located in humid climate zones and far away from plate boundaries. Owing to long earthquake-recurrence intervals and low slip rates in these environments, the recognition of potentially active faults and seismogenic fault scarps is a major problem and influences the efficiency of hazard assessments (e.g. Camelbeeck & Meghraoui 1996; Camelbeeck et al. 2008; Stein & Liu 2009; Stein et al. 2015; Landgraf et al., this volume, in press).
Different preservation potentials of coseismic surface ruptures in relation to climatic and anthropogenic processes. (a) Fault scarp of the 1954, MW 7.3 Fairview Peak earthquake, Northern Nevada; picture taken in September 2008. Note the good preservation of the free face and only little fault-scarp degradation. (b) Map view of the 1872, MW 7.4 Owens Valley earthquake, California; white arrow depicts surface-rupture trace. The rupture cuts through alluvial-fan sediments and a basaltic cinder cone (central black arrow); image courtesy Google Earth, copyright DigitalGlobe. (c) Surface rupture of the MW 7.1 Canterbury earthquake, Christchurch, New Zealand. Big white arrows depict surface trace of the rupture on undisturbed field. Note the immediate plowing and flattening of the surface scarps by a tractor a few days after the earthquake (left field). Small white arrows indicate direction of slip (right-lateral) shown by small offset feature. Image courtesy Russel A. Green and New Zealand Natural Hazard Research Platform (http://www.naturalhazards.org.nz). (d) Photograph (view to the east) of the Wissersheimer normal fault scarp in the central Lower Rhine Graben, south of the city of Kerpen. The height of the scarp at this location is 4.5 m.
Thus, if severe erosional overprint of fault scarps leads to either smoothing of a fault scarp profile or the formation of fluvial scarps at pre-existing seismogenic scarps, additional observations are necessary to unambiguously identify the fault zone and offset markers in the outcrop and in the subsurface. In this case near-surface geophysical methods offer a non-destructive means to identify and characterize faults with limited surface expression (e.g. Verbeeck et al. 2000; Vanneste et al. 2008; McCalpin 2009; Gold et al. 2013; Pueyo Anchuela et al. 2015).
In the tectonically active, low-strain regions of Central Europe a combination of geomorphic, geophysical and geological analyses has proven effective in the identification of seismogenic structures (Meghraoui et al. 2000; Verbeeck et al. 2000; Demanet et al. 2001; Vanneste et al. 2001; Peters et al. 2005; Štěpančíková et al. 2011). In particular, electromagnetic mapping and seismic refraction profiling combined with electrical tomography (i.e. resistivity and chargeability measurements) and ground-penetrating radar (GPR) have been successfully applied in locating faults and furnishing offset estimates. The most efficient combination of methods, however, depends on the individual on-site conditions. For instance, the quality and penetration depth of GPR measurements depends on the ground conductivity, which in turn is controlled by clay content, soil texture, the presence of water and the amount of iron-bearing minerals in the sediments (Olhoeft 1998; Bristow & Jol 2003; Daniels 2005).
Here, we present results of geophysical prospecting techniques combined with geomorphic and geological data for two case studies in the central (Erft Fault, Hemmerich site, hereafter referred to as site A) and southwestern (Schafberg Fault, Untermaubach site, hereafter referred to as site B) sector of the seismically active Lower Rhine Graben in northwestern Germany (Fig. 2). The principal goals of our studies were (1) to assess geophysical and geological methods used for detecting Quaternary, i.e. active faults (Wallace 1981), (2) to locate and place constraints on fault geometry, (3) to recognize structures and strata diagnostic of seismogenic faulting and (4) to evaluate the influence of any non-tectonic overprint on fault-scarp morphology.
Seismotectonic framework of the Lower Rhine Graben. Instrumental and historical earthquake data from Leydecker (2011). Fault data from Vanneste et al. (2010). Subsidence data from Schäfer (1999). Fault abbreviations: BEE, Beegden Fault; BEL, Belfeld Fault; BEN, Benzenrade Fault; BIR, Birgel Fault; BOC, Bocholt Fault; ECK, Eckelrade Fault; ELS, Elsloo Fault; ERF, Erft Fault; ERP, Erp Fault; FEL, Feldbiss Fault; GEL, Geleen Fault; GEU, Geulle Fault; GRO, Grote Brogel Fault; KAL, Kalken Fault; KAS, Kaster Fault; KIR, Kirspenpicher Fault; KON, Koningsboscher Fault; KOT, Kotenforst Fault; LAU, Laurensberg Fault; LUE, Luelkener Fault; MUE, Muenstergewand Fault; NEE, Neerbeek Fault; PEE, Peelrand Fault; REP, Reppel Fault; ROE, Roettgener Fault; RUR, Rurrand Fault; SAN, Sandgewand Fault; SCH, Schafberg Fault; STE, Steinstrass Fault; STO, Stockheimer Fault; TEG, Tegelen Fault; VEG, Veghel Fault; VEL, VeldhovenFault; VIE, Viersener Fault; WEG, Wegberger Fault; WIS, Wissersheimer Fault; ZAN, Zandberg Fault. Digital elevation model derived from SRTM data (90 m resolution).
At site A, we carried out remote-sensing terrain analysis using LiDAR (Light Detection And Ranging) data combined with geological field mapping and shallow geophysical ground-penetrating surveys to identify an active segment of the Erft Fault (Fig. 3). At site B, we combined geomorphological, geophysical and palaeoseismological trenching techniques to document distributed deformation identified in coarse gravel layers related to surface rupture in an exposure overprinted by fluvial activity.
Topographic and geological setting of the study sites. (a) Shaded relief map of the Hemmerich site derived from airborne LiDAR data (DTM1, geodetic survey of Northrhine Westphalia (Landesvermessungsamt Nordrhein Westfalen). Data are displayed using hillshade effect with illumination from the NE. Resolution: max. 1 m/pixel, vertical accuracy ±0.2 m. (b) Geological map of the Ville Horst study area, modified after the geological map of Northrhine Westphalia (von Kamp 1987). (c) Shaded relief map of the Untermaubach site derived from airbourne LiDAR data (DTM5 m, provided by the geodetic survey of Northrhine Westphalia). Data are displayed using hillshade effect with illumination from the NE. Resolution: max. 5 m/pixel, vertical accuracy ±0.2 m. (d) Geological map of the Rur valley study area, modified after the geological map of the Northern Eifel (Knapp & Hager 1980).
Regional setting
The Lower Rhine Graben is a seismically active, low-strain continental rift located in a moderately humid climate of central Europe. During the Pleistocene this region was repeatedly dominated by periglacial conditions (Klostermann 1992). Today, owing to its temperate climate with a mean annual temperature of 9–11°C and precipitation values of 600–800 mm a−1 (German-Weather-Service 2012), as well as the nutrient-rich soils developed on Pleistocene loess, the Lower Rhine Graben is intensely used for agriculture (Jones et al. 2005; Blume et al. 2016). More than 50% of the region is used as farmland and about 20% is covered with forest. Furthermore, the Lower Rhine Graben has one of the highest population densities in Europe, with an average of 576 inhabitants/km2 (Strohmeier et al. 2007; LÖGD-Report 2010).
Seismotectonic and palaeoseismic studies in the Lower Rhine Graben have revealed very slow Pleistocene and Holocene fault-slip rates, ranging between 0.01 and 0.23 mm a−1 (Ahorner 1968; Geluk et al. 1994; van den Berg 1994; Meghraoui et al. 2000; Vanneste et al. 2001; Houtgast et al. 2005; Grützner et al. 2016). Long-term denudation rates range from 0.05 to 0.08 mm a−1 (Schaller et al. 2002, 2004; Meyer et al. 2008) and regional present-day soil erosion rates range from 0.5 to 1 mm a−1 (Bork & Lang 2003). Because both long- and short-term erosion rates are on the same order of magnitude as fault-slip rates, erosion and soil creep might disguise and alter tectonically induced surface features.
Although the long-term effects of active tectonics are clearly visible in the landscape of the Lower Rhine Graben in the form of degraded, gentle fault scarps (Fig. 2), structures unambiguously indicating coseismic surface rupture are virtually non-existent in the densely vegetated and populated landscape. Owing to the lack of surface exposure and the low slip rates on individual faults, a long-lasting debate has centred on the question whether faults in the Lower Rhine Graben move by coseismic rupture or by slow, protracted aseismic creep (Ahorner 1975, 2001; Camelbeeck & Meghraoui 1996, 1998; Vanneste & Verbeeck 2001b; Houtgast et al. 2003; Camelbeeck et al. 2007).
Indeed, differentiation between coseismic and aseismic fault motion in the landscape and associated sedimentary records in the hanging walls of faults in the Lower Rhine Graben region is difficult. In this area, one of the most drastic surface modifications is related to extensive underground coal and open-pit lignite mining. Collapsed mines in the main coal-mining district (Ruhr area) have caused subsidence of several tens of metres with local subsidence exceeding 1 m a−1 during the last 60 years (Spreckels 2000; Wegmuller et al. 2000). In addition, these processes have triggered numerous shallow mining earthquakes (Hinzen & Reamer 2007; Bischoff et al. 2010). In the Ruhr area alone about 1000 mining-induced seismic events (Fig. 2) with magnitudes ranging from 0.7 to 3.3 are recorded every year (Bischoff et al. 2010). Open-pit lignite mining creates another set of processes that affect ground stability. For example, in the central portion of the Lower Rhine Graben open-pit mining requires regional-scale ground-water diversion, which in turn leads to ground-surface lowering via sediment compaction and sagging effects owing to the collapse of previously water-filled cavities. The consequence is massive regional subsidence in the vicinity of the open-pit mines (Fig. 2) that locally exceeds 10 cm a−1 (Schäfer 1999). Results of geodetic and palaeoseismic studies suggest that this process induces aseismic creep at rates of more than 1 cm a−1 along some sections of the Erft and Rurrand faults (Camelbeeck et al. 2001; Kümpel et al. 2001; Vanneste & Verbeeck 2001a, b; Görres et al. 2006; Görres 2008).
Setting of site A (Hemmerich site)
Site A is situated c. 20 km south of Cologne on the Ville Horst, a NNW–SSE-oriented ridge in the eastern central Lower Rhine Graben (Fig. 2). The Erft and Swist faults, two large NW–SE striking border faults of the Lower Rhine Graben, traverse the central and SE parts of the study site (Fig. 3a; Ahorner 1962). Fault-offset measurements (Ahorner 1962) revealed a maximum of 140 m of vertical displacement for the base of Early Pleistocene terrace gravels (‘ältere Hauptterasse’) at the SE end of the Ville Horst. Here, the Swist Fault accommodates the largest portion of Quaternary offset (c. 85 m, Ahorner 1962). Towards the NW, the Erft Fault increasingly accommodates displacement and accounts for the major portion of Quaternary offset (c. 100 m) about 8 km NW of the study site (Ahorner 1962).
The most prominent sedimentary deposits covering the Ville Horst are Early Pleistocene conglomerates of the Rhine River. On the hanging wall of the suspected fault scarp, the terrace gravels are overlain by Late Pleistocene loess and reworked loess deposits (von Kamp 1986). Tertiary sedimentary strata including Oligocene and Miocene sandstone, siltstone and lignite, as well as Pliocene sandstone are exposed at the flanks of the Rhine River and Swist Creek terraces as well as in several quarries, particularly along the SW flank of the Ville Horst (Ahorner 1962; von Kamp 1986).
Mining-induced fault creep has been reported for the Donatus Fault, a strand of the Erft Fault system, c. 5 km NW of the study site. Here, displaced walls and roads as well as continuous GPS measurements reveal vertical displacement rates of up to 22 and 6 mm a−1 of horizontal displacement (Görres 2008; Görres & Kuhlmann 2008).
Setting of site B (Untermaubach site)
Site B is situated c. 0.5 km SW of the village of Untermaubach, where the Schafberg normal fault projects into the Rur River valley. The Schafberg Fault is located at the southwestern transition of the Lower Rhine Graben to the Rhenish massif (Figs 2 & 4) and is predominately exposed in Lower Devonian mudstones and siltstones (Heimbach and Rurberg layers) and Triassic Buntsandstein sandstones (Knapp & Hager 1980). Ahorner (1962) reported vertical offset of the Devonian basement and Triassic cover rocks of the Eifel of several hundredmetres along the Schafberg Fault. Between the villages of Bogheim and Untermaubach (Fig. 3b) offset measurements and drilling revealed Early to Mid Pleistocene terrace gravel (‘Upper Main Terrace’) vertically displaced by 10–15 m (Quitzow & Vahlensieck 1955; Ahorner 1962; Richter 1962). The geomorphic expression of the Schafberg Fault is most obvious south of the village of Straß (Fig. 4), exhibiting a clear linear offset of several tens of metres in Devonian bedrock. Along strike, the trace of the fault is also emphasized by patches of Tertiary sand- and siltstone located on the hanging-wall block of the fault (Fig. 2). Quitzow & Vahlensieck (1955) suggested that Tertiary sediments originally covered the entire region and are now mostly eroded resulting from Quaternary vertical uplift of the Rhenish massif. South of Untermaubach, the Schafberg Fault traverses a narrow river valley surrounded by the Devonian rocks and Triassic cover rocks (Knapp & Hager 1980). The valley floor is covered by Late Pleistocene and Holocene fluvial deposits of the Rur River. This location is one of the few sites in the Lower Rhine Graben where a fault clearly cuts Holocene sediments (Kübler 2013). The deposits at this site comprise fluvial gravel of the youngest terrace sequence (‘Lower Terrace’) of the Rur Valley (Klostermann 1992). The gravels are covered by Holocene flood deposits (Fig. 3b), which consist of silty sand with clay-rich layers and abundant organic material. A soil has developed on these fluvial deposits. The site is situated on farmland very close to the recent channel of the Rur River, which suggests that the surface of the terrace riser and any possible tectonic surface structures have probably been impacted by both fluvial and anthropogenic processes.
Locations of geophysical and morphological measurements at (a) the Hemmerich site (site A) and (b) the Untermaubach site (site B), adapted from Streich (2003). White arrows and dashed white lines indicate the inferred fault scarps. Aerial photographs from the Geological Survey of Northrine Westphalia.
Methods
Geomorphic analysis
For site A, we used the DGM1 (Digitales Geländemodel l) subset provided by the geodetic survey of North-Rhine Westphalia, with a regular point distribution and a resolution of 1 m2/pixel (Fig. 3a). For site B, only the DGM5 subset with a ground resolution of 5 m2/pixel was available (Fig. 3c). Here, we additionally carried out topographic levelling using a total station (Figs 4b & 5). Measurements were done at each site relative to a fixed point along profiles approximately perpendicular to the visible scarps and inferred faults. The point spacing was 0.5–2 m along the profiles, and profiles were separated by 3–10 m.
Topographic analysis analysis of site B. (a) Terrain slope map of the study site; (b) topographic profiles of the main profile line.
Geophysical analysis
We performed geophysical measurements to both define the depositional characteristics and structure in inferred extensional fault zones (site A and B, respectively) and to identify an optimal trenching site for further geological analyses (site B). At site A, all survey lines strike 30° and involve a length of about 300 m. Because of land ownership constraints, all profile lines terminate at the NW–SE-oriented road, which is subparallel to the inferred fault scarp (Fig. 4a). Shallow electromagnetic mapping was conducted across a c. 160×100 m area (Fig. 4b).
Electromagnetic mapping
Electromagnetic techniques measure electric conductivities of the ground by inductive coupling between the ground and the measuring instrument (McNeill 1980). We used an EM38 device with parallel coil axes separated by 1 m and a fixed transmitting frequency of 14.6 kHz. We did all surveys in vertical dipole mode (i.e. vertical coil axes), which provides a penetration depth of about 1.5 m (McNeill 1980). We surveyed all sites with a point spacing of 0.5 m along profiles perpendicular to the suspected faults, and a profile spacing of 1–2 m. At site A, electromagnetic mapping was conducted on a 100×25 m area. Further electromagnetic and electrical resistivity tomography data were collected along a limited number of profiles adjacent to this area, and east of the main survey area (Fig. 4a). At site B, we carried out electrical resistivity tomography measurements on a 180 m-long, 65° oriented profile, approximately perpendicular to the inferred fault (Fig. 4b). The electrode spacing was increased to a maximum of 24 m in 3 m increments.
Electric tomography
In electric tomography surveys, two quantities, resistivity and chargeability, are measured simultaneously (Kemna et al. 2000; Daily et al. 2005). Here, we only show the resistivity data as for both sites this dataset provided a good image of the subsurface structures. For resistivity measurements, we used a standard four-electrode array, consisting of two current and two potential electrodes. We used dipole–dipole (at site A) and Wenner (at site B) electrode configurations. In a dipole–dipole array, current and potential electrodes are arranged in dipoles with equal spacing d and a multiple nd of the dipole spacing in between. Owing to the relatively small dipole spacing, a dipole–dipole array can better resolve vertical structures (Knödel et al. 2013). However, for large distances between the two dipoles the potential difference between the potential electrodes becomes very small, resulting in higher noise levels compared with a Wenner array. The initial electrode spacing at the Erft Fault was d=3 m, and n=1. Maximum distances after gradually increasing n, and increasing d whenever the noise became too high, were d=9 m, and n=3.33. In a Wenner array, the potential electrodes are located between the current electrodes, and the distances between adjacent electrodes are equal. Owing to the uniform electrode spacing, a Wenner array covers the ground uniformly, resulting in an unbiased image. However, contours of structures may not be very well resolved (Knödel et al. 2013). We started with an electrode spacing a=3 m and gradually increased a to 24 m.
Ground penetrating radar
We used Ramac GPR equipment with unshielded antennas of 50 MHz (Daniels 2005) and recorded profiles perpendicular to the suspected faults. Trace increments were 10 cm. The antenna was always oriented perpendicular to the survey lines. At both sites, we also carried out velocity soundings with a fixed transmitter and a moving receiver (Davis & Annan 1989). As the antenna separation was increased to over 7 m, and horizontal penetration was not larger than twice the vertical one, only the lower-frequency antennas were suitable for such measurements. The trace increment was 5 or 10 cm during different surveys, respectively.
Geological analysis
Coring
Coring was carried out along a scarp-perpendicular transect at site A. We used a motorized drill to sample soil and sediment to depths of 5 m. Soils and sediments overlying fluvial terrace gravels were penetrated, and most drill cores terminated near the top of the gravels. Continuous cores about 2.5 cm in diameter were collected.
Excavation
We opened one trench (65° orientation) at site B. For site A, we did not obtain excavation permits because of land ownership issues. We opened the trench parallel to the geophysical profile lines (Figs 4b & 6) where geophysical prospecting indicated shallow subsurface faulting (Fig. 10). The trench site was located on farmland in the narrow Rur River valley on a terrace surface (Lower Terrace) that is also the present-day flooding surface. Owing to its low elevation relative to the current river bed, the ground-water table at the trench site was situated about 1.5 m below the surface. For further details on trench-excavation strategy, see Kübler (2013). We recorded the sedimentary deposits by describing composition, lithology and sedimentary fabrics in detail. We further mapped deformation features including fissures and joints as well as rotated and fractured clasts exposed in the gravel deposits.
Drilling results shown in soil stratigraphic profile with the principal sedimentary units encountered and superposed on topography (Streich 2003).
Results
Geomorphic features
At site A, our LiDAR-DTM analysis reveals detailed information on the topographic structure, fluvial network, anthropogenic land use, and other morphological features of the Erft Fault site (Fig. 3a). The NE corner of the map is the lowermost part of the area characterized by a flat surface situated at 95–100 m above sea-level and gently inclined to the NE (4–5°). It corresponds to the Mid-Pleistocene Rhine River terrace (‘Mittelterasse’) (Quitzow 1956; Winter 1970). The topography rises abruptly towards the SW along a NW–SE-striking scarp, resulting from fluvial incision of the Rhine River (Ahorner 1962; von Kamp 1986). The orientation of the fluvial scarp morphology is very sinuous. SW of the scarp, the landscape rises to 160 m above sea-level and is characterized by the plateau-like topography on the NW–SE-oriented Ville Ridge (Ahorner 1962). Here, a subtle NW–SE-striking, SW-facing scarp separates the ridge. According to previous studies (Ahorner 1962; von Kamp 1986; Strecker et al. 2002b; Streich 2003), this scarp represents the surface expression of the SW-dipping Erft Fault. At the SE sector of the study area, the scarp is c. 4 m high. To the NW, the scarp height gradually increases, reaching a maximum of c. 10 m. The SW portion of the plateau in the centre of the study area is characterized by subtle topography with the exception of one deeply incised channel draining to the SW, and three incised tributaries flowing from the north, NE and SE into this channel. Towards the SW, elevation decreases rapidly to 30–35 m along a steep NW–SE-oriented scarp, which corresponds to the surface expression of the Swist Fault (Ahorner 1962). The scarp has been modified by lateral erosion and incision of the Swist stream. The lowermost point of the scarp is situated at the recent course of the Swist stream (110 m above sea-level). Towards the SE of the Swift, the topography gently rises again to a maximum of 125 m above sea-level.
At site B, analysis of the 5 m-LiDAR dataset did not allow to visualize the subtle topography of the terrace riser (Fig. 3c). Levelling results at the site, however, reveal two subparallel NNE-oriented, east-facing scarps at the surface of the valley floor where the Schafberg Fault projects into the valley (Fig. 5). To the south, the scarps disappear in the active bed of the Rur River; to the north the scarps are overprinted by road construction and poorly visible owing to vegetation cover. The trends of the two scarps are slightly different: the southwestern scarp strikes 125° and is slightly curved, whereas the northeastern scarp is nearly linear, striking 155°. The subtle scarps are 0.8–1.2 m high, respectively, and affect the hydrological conditions at the study site. In summer, plants along the orientation of the scarps often wilt owing to local drought conditions when runoff in the nearby stream is low, whereas in spring the water table is high in the suspected hanging wall east of the scarp (Strecker et al. 2002a).
Coring
Coring at site A revealed differently stratified layers NE and SW, respectively, of the fault scarp (Fig. 6). A 40 cm-thick, brown-coloured plowing horizon (Munsell colour 10YR4/3), comprising silt with homogenously distributed organic material, forms a veneer at the top. In places, the plowing horizon is covered by recent landfill. In the inferred hanging wall of the fault, an argillic brown soil was encountered. The B-horizon underlying the plowing zone consists of reworked, decalcified loess and contains occasional Mn and Fe concretions, indicating a fluctuating ground-water table. To the NE, near the topographic scarp, the upper part of the B-horizon frequently contains clay cutans. SW of the 40 m point along the profile, primary loess constitutes the C-horizon of the terrace-soil profile. At 65 and 90–100 m from the origin, primary loess was encountered immediately below the plowing horizon, indicating localized top-soil erosion owing to fluvial processes or erosion related to farming. In contrast, between these points the primary loess underlies decalcified loess, which is up to 2 m thick. The soil and loess in the hanging wall cover a formerly more pronounced relief of the gravel-terrace surface and overlie a unit of silty gravel, apparently a mix of loess and the underlying sandy terrace gravels. This unit mainly consists of silt and gravel, with less than 15% sand. Where the silty gravel underlies primary loess, it is calcareous at the top. The gravel components reflect the lithology of the adjacent Rhenish Shield, but also contain Buntsandstein and quartzite clasts. At the SW end of the profile, the silty gravel is about 1 m thick. We observed a gradual transition to the original, reddish sandy fluvial gravel at a depth of about 4 m. Further NE within the inferred hanging wall of the fault and c. 15 m from the origin, the silty gravel thickens to at least 2.8 m, and the transition to the original terrace gravel occurs at a depth even greater than at the SW end of the profile.
The inferred footwall in the NE, although probed by one drill core only, reveals a reduced soil profile without any primary loess cover. In the northeasternmost drill core we found sandy terrace gravel (Fig. 6) at a depth of 1 m, and silty gravel is reduced to a thickness of 0.15 m. Terrace gravel at shallow depth is also indicated by abundant coarse gravel clasts exposed on the field, probably brought to the surface by plowing. Considering surface topography, the top of the original terrace gravel lies at least 3 m deeper in the hanging-wall sector, only 30 m SW of this footwall point. The offset amounts to more than 6 m between the point in the inferred footwall and the SW end of the transect (Fig. 6).
Geophysical measurements
Electromagnetic mapping
At site A, the conductivity map (Fig. 7a) shows pronounced spatial variability. In the main survey area (Figs 4a & 7a), a pattern of alternating zones of high and low conductivities parallel to the topographic scarp was imaged. Owing to uneven ground conditions, this part of the conductivity map is very noisy. The pronounced zone of high conductivities in the NE is probably due to anthropogenic landfill, whereas alternations further SW may be related to changes of soil composition or changing soil-water content.
(a) Integral ground conductivity distribution of the uppermost 1.5 m at site A. Electromagnetic map (Streich 2003) superimposed on aerial photograph (2009 GeoBasis-DE/BKG, provided by Google Earth©). Bold dashed line illustrates a break in conductivity and inferred fault locations. Thin dashed lines outline similar but less pronounced zonation in adjacent survey areas. (b) Satellite image (2006 SPOT/CNES, RGB, provided by Google Earth©) of the Untermaubach Site. Location of the main profile is superposed. Note the difference in colour of the pasture corresponding to variations in conductivity. (c) Integral conductivity distribution of the uppermost 1.5 m at the Untermaubach site, obtained by electromagnetic mapping (Streich 2003).
Structures imaged in the northwestern survey area are smooth. However, structures in this area deviate from the uniform scarp-parallel zonation; pronounced conductivity variations also occur in SE–NW direction. The profiles between the two areas mapped add more complexity to the image. At the NE termination of these profiles, low conductivities were observed, whereas the northeasternmost parts of both adjacent surveyed areas reveal high conductivities. The profiles in the SE, although located further away from the main survey area, exhibit a zonation very similar to that of the main survey area. A linear structure, manifested as a sudden conductivity increase towards the SW, can be traced throughout the entire map. This inferred structure is roughly parallel to the topographic scarp. The location of the conductivity increase agrees well with the major seismic velocity contrast and the offsets observed in radar reflectors. Compared with the coring data, the conductivity increase matches an increase in thickness of the soil horizons.
At site B, optical satellite images illustrate the difference in soil properties through a different colour of the vegetative cover, and therefore indicate the location and orientation of potential subsurface anomalies (Fig. 7b). The electromagnetic map (Fig. 7c) shows that the anomalies visible on satellite images are also recognizable in the variation of conductivity values. Integral conductivities of the uppermost 1.5 m range from about 3 mS m−1 to more than 20 mS m−1, corresponding to resistivities of c. 330–50 Ωm. Three zones of increased conductivities, intersecting the main profile nearly at right angles, can be identified. A high-conductivity zone at the eastern part of the conductivity map is bounded by a sharp linear conductivity contrast, which trends 170° and cross-cuts the main profile 40 m from the origin. This contrast coincides with an increased topographic slope. The zone of low conductivities west of the sharp contrast is interrupted 30 m north of the main profile, but continues near the northeasternmost corner of the survey area. In contrast, the other two high-conductivity zones are not as straight and not bounded as sharply. One high-conductivity zone intersects the main profile at 72–83 m. This zone is curved and widens towards the SE. The third zone of high conductivity is located in the SW of the survey area; it is wider and not as pronounced, as the general conductivity level is higher there than in the eastern part of the mapped area. Another linear zone of slightly increased conductivity strikes 35° and cross-cuts all other structures. As it is narrow and perfectly straight, parallel to the trend of the road, and passes by a shed, this structure is likely to be of anthropogenic origin.
Resistivity
At site A, the inverted resistivity cross-sections of the main profile (Fig. 8a) reflect the different stratigraphic units encountered during coring. A narrow, steep, deep-reaching zone of minimum resistivities at about 13 m correlates very well with the clay unit and suggests that this unit may continue further downward and even widen at depth. Minimum resistivities in this zone are as low as 10 m. The location of the low-resistivity zone on the main profile interferes with the landfill. However, coring shows that the fill is not thicker than 0.45 m above the clay zone, and magnetic data indicate that its main body is located 10–20 m away from the resistivity profile. Thus, the landfill is not likely to exert a notable influence on resistivity in the depth range where the pronounced anomaly indicates clay (about 4–10 m). SW of the prominent anomaly caused by the clay, the main profile indicates the presence of a high-resistivity unit at depth, which can be associated with terrace gravel. Within this unit, resistivities rapidly increase downward SW of 75 m, where maximum values of several hundred Ωm are attained. In the NE part of the section, the transition to high resistivities is more gradual, corresponding to a thicker horizon of reworked silty gravel in this steeper-sloping portion of the profile. A zone of moderate values is evident within this deep unit at 52–66 m from the origin, below the depth penetrated by coring. This unit is overlain by a layer with resistivities almost as low as in the steep, narrow anomaly. The low-resistivity top layer corresponds to the silty, clayey soil and primary loess (Fig. 6). The top layer, not exceeding a thickness of 1 m immediately adjacent to the pronounced minimum-resistivity anomaly, thickens stepwise towards SW. One step is visible at 72 m, where, according to the land owner, a change in soil properties occurs. Beyond 88 m from the origin, resistivities indicate an average thickness of the upper layer between 4 and 5 m, which is thicker than the soil thickness penetrated by coring and which may encompass the upper part of the silty gravel. The boundary layer inferred from resistivities undulates: the underlying high-resistivity unit bends upward at 97 m and downward at about 90 and 109 m, respectively.
Results of resistivity and coring at site A (Streich 2003). Profile locations are shown in Figure 4a.
At site B, the inversion results of electric tomography (Fig. 9a) show that ground resistivities along the main profile range from roughly 80 to 500 Ωm. Resistivity values in the top c. 1.5 m are consistent with the results obtained from electromagnetic mapping. Two principal units can be distinguished. A layer with high resistivities, c. 4–5 m thick, overlies a moderately resistive unit. A thin irregular unit of moderate resistivities forms the top. At the coarse resolution of the profile, this unit is interrupted twice by upward-directed bends of the underlying principal resistor. It also encompasses several depressions, which exhibit the lowest resistivities in the section at 40, 80 and 140 m along the main profile (Fig. 9).
Results of ground resistivity (a) and ground penetrating radar (b) at site B along the main profile with topography incorporated, two times vertically exaggerated (Streich 2003). The location of the profile line is shown in Figure 4b. Note the three zones of minimum resistivity at 140, 80 and 40 m, respectively, marked by white arrows. B: GPR Antenna frequency 50 MHz. Brown lines depict surface. Depth axis shows subsurface depth for a ground velocity of 0.08 m ns−1. Note the symmetrically dipping reflectors at 140 m (1), the asymmetrically dipping reflectors at 80 m (2) and the offset reflectors at 40 m (3) indicated by the white quadrangles.
Ground penetrating radar
GPR sections along the main profile at site B reveal a pronounced, strongly structured reflector at shallow depths (Fig. 9b). The horizon correlates well with near-surface resistivity anomalies (Fig. 9a). Near 40 m from the origin, the reflector exhibits an offset from a depth of less than 1 m in the SW to approximately 1.5 m in the NE. The offset is located at the SW margin of a moderate resistivity anomaly. Further NE, the reflector exhibits some variation in depth, being generally shallower than in the pronounced offset zone, yet not as shallow as immediately SW of 40 m. Another depression in the radar reflector appears about 80 m from the origin. In accordance with the coring data, the southwestern slope of this depression is steeper than the northeastern slope. Furthermore, a wide, shallow trough and an overall weaker signal at 133–153 m agree well with the resistivity anomaly in the same distance range (Fig. 9b). At 150 ns, corresponding to a depth of approximately 6 m, another reflector is indicated in the NE half of the profile. As the depth of this reflector approximates the top of the deep low-resistivity unit revealed by electric tomography, this reflector may be associated with the top of the Devonian basement.
In summary, geophysical prospecting at site B reveals three anomalous zones characterized by resistivity minima in the resistivity-depth model, high near-surface conductivity in the conductivity map and offsets of radar reflectors (Figs 8b & 9a, b). The NNW–SSE trend of the anomalous zones, identified by field surveys and geophysical mapping, coincides with the strike of the fault inferred from geological maps (Fig. 3b) and aerial photographs (Fig. 7b). This orientation is also parallel to the trend of the topographic steps observed in the field (Fig. 5). The anomalous zone intersecting the main survey line at c. 80 m is curved in areal view (Fig. 7b, c), but most pronounced in vertical sections: the offset of the radar reflector is largest and resistivity exhibits a pronounced minimum (Fig. 9b). Conversely, the zone in the NE of the survey area shows the most pronounced spatial conductivity contrast that also coincides well with the observations made based on aerial photo mapping (Fig. 7b). The third anomalous zone further SW and upslope is most symmetrical and widest, but least pronounced.
Trenching at site B
The trench excavated at site B exposes a 4–5 m-thick cover of unconsolidated fluvial deposits overlying Devonian sedimentary rocks (Fig. 10a, b). The basement rocks consist of platy mud- and sandstone of the Heimbach Formation (Upper Lower Devonian; Knapp & Hager 1980). The trench exposed the top of the basement at a depth of 5 m at metre 25 of the trench log (Fig. 10c). At 15 and 1.5 m of the trench log, the top of the basement was documented in drill cores at depths of 6.0 and 6.2 m, respectively (Fig. 10c). The sedimentary succession can be subdivided into four main units from oldest to youngest (units A–D). The lowermost unit A comprises a c. 1.5 m-thick layer of clast-supported coarse gravel with predominately prolate clasts up to 60 cm in diameter in a silt- and clay-rich matrix. Clasts mainly consist of sandstone, shale and quartzite. At the top, the unit A layer exposes interspersed clay lenses with abundant organic material possibly related to a reworked soil horizon (unit A′, Fig. 10c). This horizon is overlain by unit B, a 1–2.5 m-thick inhomogeneous clast- and matrix-supported coarse-gravel layer with interbedded sand and fine-gravel lenses. Clast lithology does not significantly differ from unit A, but in unit B clasts are smaller and better rounded. Maximum clast sizes do not exceed 30 cm. Unit B is in turn overlain by a matrix-supported clayey and sandy to fine-gravel layer (unit C) with cross-bedded sand lenses. Furthermore, the trench exposes two asymmetric channels, which coincide with the position of the two easternmost surface scarps and geophysical anomalies 2 and 3 (Fig. 9). The channel fill consists of sandy silt and clay-rich layers (unit D, Fig. 10).
Results of the trench analysis at site B. (a) Photo mosaic of the northern trench wall; dashed line indicates the original surface before excavation and removal of the upper soil horizon. Trench log of the main units and deformation features of (b) the west-northwestern part and (c) the east-southeastern part of the northern trench wall. Stars depict the locations of fractured clasts.
The eastern channel has a total width of 12±2 m and a maximum sediment thickness of 2.5 m (Fig. 10). The channel exhibits an internal horizontal stratification characterized by variations in organic-matter content, clay and Mn- and Fe-oxides. The base of the channel lies below the present-day ground-water table and shows the typical features of a gleyic soil horizon (Gr), i.e. a horizon characterized by reducing conditions exhibiting a greenish blue colour owing to the presence of ferrous oxide (FeII) (Scheffer & Schachtschabel 2002). This horizon furthermore contains a high concentration of organic matter in the form of retransported plant remains, charcoal and wood. The superseding horizon is characterized by Mn- and Fe-oxides in the form of black and reddish concretions ranging from <1 mm to up to 2 cm in diameter.
The western channel had a total width of 10±1 m and a maximum sediment thickness of 3 m. The channel is highly asymmetric with a steep western flank (45±5°) and a shallower eastern flank (15±3°). The base of this channel lies also below the ground-water table, exhibiting the same features as the eastern channel. The overlying deposits predominately consist of clayey silt with a few isolated clasts exhibiting different zones of vertical and lateral clay transport caused by fluctuations of ground-water and percolating water. Furthermore, a moderately well-preserved subchannel within the upper channel bed is exposed (Fig. 10).
Deformation features
Owing to the lack of laterally continuous layering within the gravel deposits and intense sediment alteration produced by ground-water fluctuation, the recognition of sediment deformation within the gravel deposits required detailed mapping of the trench walls. Sediment deformation in this fault exposure is not restricted to a single fault plane, but is rather characterized by fractured and offset clasts within a fabric of rotated clasts, predominately across a c. 10 m-wide fault zone (Fig. 10). In addition, there is less prominent, but still recognizable deformation across a c. 30–40 m-wide deformation zone below the eastern surface scarp (Fig. 5). Gravel deformation decreases significantly with distance from the fault zone and is not recognizable below the western channel. The cumulative offset across the fault zone of 1.0±0.2 m is taken up by numerous small intersedimentary fractures, each accommodating a few centimetres of displacement.
Secondary evidence for coseismic activity at the trench site is provided by liquefaction phenomena exposed in both the hanging and footwalls of the fault. Liquefaction occurs exclusively in fine sand and silt layers of unit C (Fig. 10b, c). The fine-grained channel fill (Unit D) shows no signs of deformation and truncates the deformed gravel unit B below the eastern scarp (Fig. 10c).
Discussion
The structural and geomorphic setting as well as historical and instrumental earthquake data for the Lower Rhine Graben (Hinzen & Oemisch 2001; Leydecker 2011) render the Erft and Schafberg Fault sites well-suited locations to assess rupture effects related to large Holocene earthquakes (Strecker et al. 2002a, b). In the context of the low deformation rates, humid climate conditions and the dense population, our study demonstrates the effectiveness of a multidisciplinary approach to identify potentially active faults in this type of environment.
High-resolution optical remote sensing, digital elevation model (DEM) analysis and electromagnetic mapping at both study sites reveal anomalies that indicate fault activity. At site A, sectors of high conductivity contrast match well with the position and orientation of the surface scarp identified on LiDAR data (Fig. 3a). At site B, conductivity results confirmed the location of structures identified in optical imagery (Fig. 7b). However, electromagnetic mapping did not yield new insights – the strongest effect was shown where the anthropogenic landfill was in contact with the scarp. Since data acquisition and processing for electromagnetic mapping are much more time-consuming compared with first-order satellite data analysis, it is unlikely that the results obtained justify the effort. In particular, LiDAR data combined with optical satellite data of submetre resolution (e.g. Worldview, Quickbird, etc.) may be a faster and equally reliable method of choice for areas such as the Lower Rhine Graben to obtain first-order constraints on potential fault locations.
Using resistivity tomography we detected lateral discontinuities as well as subsurface structures along the profiles surveyed. Surface-offset estimates at site A were derived from topographic analysis of the LiDAR dataset. Electrical tomography shows a distinct contact suggesting a single fault that coincides with the surface scarp that, combined with coring data, provided first offset estimates of 3–6 m of a subsurface marker that probably represents the top of the gravel terrace (Fig. 8).
At site B, identifying the location of the fault was more difficult. Topographic analysis indicates two potential scarps; geoelectric and GPR data reveal three subsurface anomalies indicating lateral discontinuities in the sediment layers (Fig. 9). However, the strong fluvial overprint of the Holocene terrace riser makes a differentiation between tectonically v. fluvially generated features, regarding both the subsurface anomalies and surface scarps, very difficult. However, trenching revealed a clear difference between the NE and SW scarps; only below the NE scarp, we observed sediment-deformation features that would suggest distributed seismogenic rupturing in a fault zone. The SW scarp is most likely the result of fluvial incision and subsequent filling by the meandering Rur River. This distinction could not have been made from the geophysical data alone (Fig. 11a, b). However, GPR data show differences in the geometry of the anomalously dipping reflectors. While the SW reflectors suggest an asymmetric channel with no clear vertical offset, the NE reflector is clearly offset and downthrown to the NE, with a total displacement of about 1 m (Figs 9 & 11b). Also, the difference in strike and shape of the two surface scarps in map view (Fig. 5) is consistent with a fluvial origin of the western scarp owing to its arcuate shape – like an old meander of the Rur River – whereas the eastern scarp is perfectly straight.
Comparison of electric tomography and GPR to decipher the nature of the two channels excavated during trenching at site B. Black arrows depict suspected fault position; (a) Ground resistivity, and (b) GPR results of site B fitted to scale of the palaeoseismic trench (outlined in thick dashed lines). (c) Trench logs of the two channel fills; note the intensive sediment deformation in the eastern channel and vertical offset of sedimentary layers, compared with the lack of sediment deformation in the western channel. Map key shown in Figure 10.
The geomorphic, geophysical and geological data collected indicate that active faults affect both study sites. At site A, the fault strike and surface offset can be traced in the topography and matches previous mapping of the Erft Fault (Ahorner 1962; von Kamp 1986). The Erft Fault constitutes one of the main boundary faults of the Lower Rhine Graben with a relatively large Quaternary offset of around 100 m (Ahorner 1962). It is thus likely that the structures studied at the site, which are formed in Early to Mid Pleistocene gravel deposits, represent several coseismic events resulting in a cumulative vertical offset of c. 6 m (Fig. 8). This hypothesis is supported by the analysis of geomorphic data, soil stratigraphic units revealed by coring and subsurface imaging by various geophysical data showing that the surface scarp coincides with subsurface anomalies and offset marker horizons (Figs 6 & 8). Results of resistivity tomography (Fig. 8) are supported by coring data (Fig. 6) and appear to image the position of the inferred fault at depth, 13–14 m from the profile origin.
Despite these promising observations, results from geophysical prospecting remain ambiguous when viewed in isolation, even though they display striking, apparently correlative anomalies. Coring data and regional Late Pleistocene–Early Holocene palaeo-environmental information is important to further evaluate and reconstruct the youngest tectono-sedimentary history of the site. The coring data document a reduced soil-profile thickness in the elevated footwall in the NE, and a stratigraphically complex hanging-wall profile, which can be explained in the context of faulting and diffusion processes at the scarp site.
The resistivity sections do not image the ground exactly. Root mean square (RMS) errors of the inversion were 14.6% for the dipole–dipole array and 7.1% for the Wenner array along the same line, and absolute resistivity values vary depending on the parametrization of the inversion. A high noise level associated with a dipole–dipole array may represent one cause for this result. Poor electrode coupling to the dry top-soil may have increased the errors further. However, without further analysis involving trenching, it will not be possible to fully decipher the latest rupture history of the Erft Fault.
In contrast to site A, which lies on an erosional surface of Early Pleistocene gravel beds (Fig. 3a), site B is located on a thin veneer of Holocene gravels deposited on Devonian basement rocks (Fig. 3b). Given the location of the fault at the transition from the Lower Rhine Graben to the Rhenish massif and the thin sedimentary cover, the Schafberg Fault reflects a relatively short phase of tectonic deformation from the early Holocene to the present. Two pronounced anomalies of soil thickness in the steepest sloping areas coincide with low-resistivity zones and prominent offsets in a radar reflector (Fig. 9). With resistivity and GPR analysis it was possible to identify subsurface structures associated with the surface scarps. However, with the exception of the channel structures that correspond to Unit D (Fig. 9), neither of the geophysical techniques was useful to clearly differentiate the individual sedimentary units defined in the trench. This is probably due to the subtle differences in grain size and lithology that coincide with relatively small differences in their geophysical properties and thus cannot be detected in detail with geophysical methods.
Trenching exposed two asymmetric channels filled with fine-grained flood deposits below both scarps (Fig. 11) that clearly underwent different formation processes. While the resistivity data did not allow further interpretion of the origin of the structures (Figs 9 & 11a), GPR data show some differences that, in combination with trenching results, allow for better differentiation between a fluvial or tectonic origin (Fig. 11b, c). Trenching exposed clear differences in the deformation record of the eastern and western scarp. Although both scarps are associated with elongated subsurface channels filled with fine-grained flood deposits, only the eastern scarp shows clear signs of tectonically driven sediment deformation and offset of sedimentary strata, as well as offset of the contact between Quaternary deposits and Devonian basement rocks (Fig. 10c).
Conclusions
We have identified two active faults in the central and southwestern sectors of the Lower Rhine Graben. At both sites, the seismogenic nature of the surface structure cannot be evaluated by a single method alone. We showed that an optimal characterization of the Quaternary faulting history is achieved by several methodological approaches. While electromagnetic mapping did not furnish new insights regarding the location and strike of the faults, the combination of remote sensing, electric tomography and GPR were the most promising non-invasive methods to locate faults and identify offset markers at both study sites. However, neither of the geophysical measurements used was accurate enough to differentiate between individual sedimentary units. The differentiation between tectonic, fluvial and anthropogenic processes in causing subsurface anomalies and their links with structures exposed at surface is only possible in combination with geological and palaeoseismological methods. At site A, electric tomography and coring results demonstrate that the surface scarp coincides with offset reflectors and soil stratigraphic layers that are most likely resulting from repeated faulting. At site B, the combination of GPR and trenching allowed a clear distinction between fluvial incision and distributed tectonic faulting in the Holocene causing vertical displacement of c. 1 m.
We conclude that a combination of geomorphological, geophysical and geological methods is required to identify active faults in regions characterized by small fault-slip rates and strong anthropogenic and/or climatic overprint. Faults with only subtle geomorphic expression and strong fluvial overprint can be detected with detailed integrated geophysical and trenching approaches.
Acknowledgments
Geophysical prospecting was carried out within the framework of the SAFE project (Slow Active Faults in Europe) funded by the European Research Council, EU project no. 2000.220055, awarded to M. Strecker. Trench excavation was funded by DFG-project (German Science Foundation) ‘Active intraplate deformation in central Europe: Paleoseismology of the Lower Rhine Graben’ granted to A. Friedrich (DFG FR 1673) and M. Strecker (STR 373-18/3). We thank F. Schäbitz of Cologne University for support during the coring campaigns. Furthermore, we thank R. Gold and an anonymous reviewer for helpful comments that helped to improve the manuscript. LiDAR data were kindly provided by the Geodatenzentrum NRW (© Geobasis NRW 2010).
- © 2017 The Author(s). Published by The Geological Society of London. All rights reserved
References
Active faulting in a populated low-strain setting (Lower Rhine Graben, Central Europe) identified by geomorphic, geophysical and geological analysis
