Subsurface investigations of landslides using geophysical methods geoelectrical applications in the Swabian Alb ( Germany )

Subsurface investigations of landslides using geophysical methods geoelectrical applica¬ tions in the Swabian Alb (Germany) Landslides occur frequently all over the world, causing at times considerable economic damage, injuries and even death. In order to improve hazard assess¬ ment, common landslide types of a given region need to be investigated in detail. While traditional techniques of subsurface investigation are expensive and only provide point information, geophysical methods are suitable tools for gathering 2D and 3D information on the subsurface quickly, reliably and cost-effectively. In this study, the suitability and limitations of 2D resistivity for the determination of landslide extent, structure and soil moisture conditions are presented. For this purpose, two identical profiles were taken during a two-month period. Significant differences in electrical resistivity (>1000 Qm) due to varying soil moisture conditions were observed. Using various inversion parameters, it was possible to model two distinct subsurface images. Regrettably, the sliding plane could not be detected reliably, possibly due to the homogeniety of the landslide material and under¬ lying bedrock. 208 Geographica Helvetica Jg. 61 2006/Heft 3 Erkundung des Untergrunds von Hangrutschungen unter Verwendung von geophysikalischen Methoden geoelektrische Anwendungen in der Schwäbischen Alb (Deutschland) Gravitative Massenbewegungen treten häufig und weltweit verbreitet auf. Sie verursachen hohe öko¬ nomische Schäden und fordern zahlreiche Tote. Um Gefahrenanalysen zu verbessern, sollten die für eine Region charakteristischen gravitativen Massenbewe¬ gungstypen im Detail untersucht werden. Während traditionelle Techniken sehr teuer sind und nur punk¬ tuelle Informationen liefern, stellen geophysikalische Methoden geeignete Techniken dar. um schnell, gün¬ stig und zuverlässig 2Dund 3D-Informationen über den Untergrund zu erhalten. In dieser Studie werden die Möglichkeiten und Limi¬ tierungen der 2D-Geoelektrik hinsichtlich der Bestim¬ mung der Grenzen, Struktur und Bodenfeuchtig¬ keitsverteilung einer gravitativen Massenbewegung untersucht. Zwei Aufnahmen von identischen Profilen zeigen aufgrund der unterschiedlichen Bodenfeuch¬ tebedingungen enorme Veränderungen in den elektri¬ schen Widerständen (>1000 Qm) innerhalb von zwei Monaten. Die Verwendung unterschiedlicher Inversi¬ onsparameter ermöglichte zwei verschiedene Abbil¬ dungen des Untergrunds. Leider konnte die Gleitfläche nicht verlässlich bestimmt werden. Es wird angenom¬ men, dass die Rutschmasse und das darunter liegende Festgestein ähnliche Eigenschaften aufweisen. Etudes de subsurface des glissements de terrain ä l'aide de methodes geophysiques. Applications geoelectriques dans I'Alb souabe (Allemagne) Les glissements de terrain sont des phenomenes fre¬ quents qui causent des dommages economiques et fönt des victimes dans le monde entier. Dans le but d'ameliorer l'estimation du risque lie ä ce genre de phenomenes, une etude des differents types de glisse¬ ments doit etre menee. Alors que les techniques traditionnelles d'investigation de subsurface sont onereuses et ne fournissent que des informations ponctuelles, les methodes geophysiques sont des outils adequats qui permettent de collecter des informations en 2D et 3D de la subsurface de facon rapide, peu onereuses et fiable. Cette etude presente les potentiels et les limitations de la resistivite 2D utilisee pour determiner l'extension, la structure et les conditions hydrogeologiques. Deux profils identiques mesures sur une periode de deux mois montrent une importante difference de resistivites electriques (>1000 Qm), essentiellement due ä la Variation de teneur en eau du sol. Utilisant divers parametres d'inversion, deux images distinctes ont pu etre obtenues, sans toutefois que la surface de glissement puisse etre detectee de facon precise, ce qui laisse penser que les proprietes du materiel constituant le glissement de terrain et de la röche sous-jacente sont homogenes. Dipl.-Geogr. Rainer Bell. PD Dr. Thomas Glade, JanErik Kruse, Geographisches Institut, Universität Bonn, Meckenheimer Allee 166, D-53115 Bonn, Germany. e-mail: rainer@giub.uni-bonn.de thomas.glade@uni-bonn.de j-e.kruse@gmx.de Dipl.-Phys. Alejandro Garcia, Institut für Geologie, Fachbereich Geophysik, Universität Bonn, Nussallee 8, D-53115 Bonn, Germany. e-mail: jodidocorreo@yahoo.com Prof. Dr. Andreas Hördt, Institut für Geophysik und extraterrestrische Physik, Technische Universi¬ tät Braunschweig, Mendelssohnstrasse 3, D-38106 Braunschweig, Germany. e-mail: a.hoerdt@tu-braunschweig.de Manuskripteingang/received/manuscrit entre le 13.1.2006 Annahme zum Druck/accepted for publication/accepte pour Timpression: 14.9.2006


Introduction
Landslides may be considered as common natural hazards. in many cases leading to significant economic losses and even fatalities.Since recent landslide activi¬ ties in many regions often result from reactivations of old landslides, it is important to detect and investigate older landslides in more detail in order to gain insight into landsliding processes characteristic for a particu¬ lar region.Such information could possibly be used to improve current landslide hazard assessment.
Reliable information on the extent, structure, sliding plane location.moisture conditions, ground water table and the degree of activity are essential for the careful assessment of landslide hazards.Traditional techniques (e.g.drillings) are expensive and often not suitable for the rugged terrain of a landslide.In addition, such investigations only provide point infor¬ mation.In contrast, geophysical methods are much cheaper and faster, having the added bonus of delivering 2D or even 3D information.According to McCann  & Forster (1990), it would appear that geophysical methods can deliver the necessary information for hazard assessment of landslides.An overview on the applicability of geophysical methods for geomorphol¬ ogy is to be found in Schrott et al. (2003).
In landslide studies.geophysical methods have been successfully applied over the last forty years, making use of resislivity (e.g.Denness et al. 1975; Donnelly et al. 2005), self-potential (e.g.Lapenna et al. 2005), low frequency eleclromagnetics (e.g.Schmutz el  al. 2000), ground-penetrating radar (e.g.Roch et al.   2005).seismic methods (e.g.Bogoslovsky & Ogilvy   1977; Glade et al. 2005).and gravity (e.g.Del Gaudio et al. 2000).Several studies exist comparing different geophysical methods (see Bichler et al. 2004; Cutlac   & Maillol 2004; Sass et al.).From these, it is appar¬ ent that each method has its specific field of applica¬ tion, as well as its limitalions (Tab.l).Thus, a combi¬ nation of various methods would seem appropriate for the investigation of complex structured landslides.Despite the improvements made in the implementa¬ tion of geophysical methods however, it is still crucial to support geophysical evidence with general geo¬ logical and detailed borehole information in order to obtain a more complete picture of the subsurface.
A typical problem area in the interpretation of results relates to the Variation of characteristic values within one material.As intrinsic Variation is often greater than Variation between materials, large overlapping of results can occur, in many cases preventing a definite correlation of investigated values with specific mate¬ rials.With reference to resistivity, moisture content would be the main factor causing heightened intrinsic Variation.As moisture content is a valuable aspect of landslide research, this is not necessarily a disadvantage of the method.Soil moisture studies have been carried out, for example, by Bogoslovsky & Ogilvy   (1977), Suzuki & Higashi ( 2001) and Hanafy & al   Hagrey (2006).
This study explores the hypothesis that 2D resistiv¬ ity allows identification of extent of recent landslide activity, of new and old landslide body structure (including the location of the sliding plane/s) and ena- bles the monitoring of moisture distribution within a landslide.The potentials and limitations ofthe method are addressed. 2

Study area
For the purposes of this research project, the Unter¬ hausen landslide with an approximate extent of 0.5km2 was investigated.It is located in the Central Swabian Alb (Fig. 1), a cuesta landscape composed of Jurassic sedimentary rocks (limestone overlying marls and clays).The average annual temperature is about 9°C and average rainfall ranges between 800 and 1000 mm.The settlement of the study area started in the early 1970s.
The extent of the old rotational landslide as mapped by Dongus (1977) is shown on Fig. 2. Damage on one of the houses is an indication that at least parts of the old landslide mass are occasionally reactivated.Results from the drillings and inclinometers taken at Lic01-03 indicate that the boundary of the reactivated landslide could be between LicOl and Lic02.Movement within the landslide appears to be rather complex.In late summer and early autumn 2005, a slow flowing movement was detected until a depth of 8.50 m (in borehole/inclinometer Lic02).During the extensive and rapid snowmelt in spring 2005, a sliding movement until a depth of 15.50 m was observed.Some of the massive limestone blocks from the escarpment feil down, stopping within the forested area upslope ofthe settlement.This deposi¬ tion area appears to be the old landslide head.partially suitable, -not suitable, depends on the site or needs further analysis Tab. 1: Suitability of various geophysical methods for different landslide types and landslide related features Eignung verschiedener geophysikalischer Methoden für unterschiedliche Typen von gravitativen Massenbewegun¬ gen und damit verbundene Aspekte Pertinence de differentes methodes geophysiques selon les differents types de glissements de terrain et caracteris¬ tiques associees Source: Bouillon (2005) and Hack (2000) (modified and adapted) 3 Methods

Drillings
To get information on the material and structure of the subsurface, three drillings were carried out at differ¬ ent locations (Fig. 2).Here, only the results of Lic02 are presented.The drilling was contracted to Goller Bohrtechnik, which used rotary drilling to extract a disturbed core with a diameter of 120 mm.

Direct current (DC) resistivity
Based on the findings of a previous study (Sass et al.), the selection of the main geophysical method feil on 2D resistivity tomography.This resistivity method makes use of different resistivity values specifically characteristic to individual subsurface materials.Once subsurface resistivity distribution is established, this information can be related to characteristic resistiv¬ ity values of the individual materials, allowing finally.an interpretation of the possible structure of the sub¬ surface.Examples of typical resistivity values may be found, for example, in Reynolds (1997) and Knödel et al. (1997).
Results were obtained as follows: A constant current was sent through two current electrodes of a multielectrode array in the ground.Two potential elec¬ trodes were used to measure the resulting voltage differences.Measurements were carried out with dif¬ ferent electrode array configurations in order to pro¬ vide a tomography-like resolution.Finally, distribution of subsurface resistivity in 2D could be established by inverting resistivity values (Loke & Barker 1995).
Refer to Reynolds (1997) or Knödel et al. (1997) for further details on the approach.
For this study, an ABEM Lund imaging System with a Terrameter 300 device was used.All profiles were measured applying Wenner array geometry.During the course of the year.three profiles were laid: two in a forest in exactly the same location to allow for investi¬ gation of different situational influences.and the third longitudinal to the slope to determine the subsurface structure of the landslide al the location where move¬ ments caused damage on the house (Fig. 2).The pro¬ files can be characterised as follows: Forest Profile 1: 41 electrodes, 5 m electrode spac¬ ing, 200 m length.penetration deplh approximately33 m, current 0.2 mA, 12/04/2005, after a period of heavy snowmelt.
Forest Profile 2:41 electrodes. 5m electrode spacing, Data inversion was carried out using the Software Programme RES2DINV (Loke & Barker 1995), particularly as it allowed the inclusion of data on the local topography for data processing.Tire inver¬ sion routine used by RES2DINV is based on the smoothness-constrained least-squares method.For Longitudinal Profile 1, besides Standard inversion (Profile la), a model run-through was made including results of borehole Lic02 and assuming horizontally elongated structures (Profile lb).The latter routine alternative minimises absolute changes in resistivity values, thereby enabling a clearer contrast between interfaces of different resistivity regions (Geotomo  2004).Additionally, the inversion was constrained by a vertical to horizontal flatness filter ratio setting of 0.5 and a bedrock depth of 15.15 m at the location of borehole LIC02.This too, created a sharper boundary at this depth.Grenzen der allen Hangrutschungen (kartiert im Masssteib 1:50000 von Dongus 1977) und der jüngeren Hangnilschungsaktivitäten. Zusätzlich sind die Lokalitäten der aktuellen Untersuchungen dargestellt.Carte de l'extension de l'ancien glissement de terrain (cartographie au 1:50000 par Dongus 1977) et eles glisse¬ ments plus recents.Les sites d'investigation sont ineliques.

Drilling
The Lic02 core of 16.00 m reached claystone bedrock at a depth of 15.15 m.The core material stemmed from the old landslide, consisting mainly of gravelly clay with interbedded weathered marl.
4.2 DC resistivity The calculated inversion modeis show an immanent error (RMS-error) that ranges between 2.2 and 7.9.
Consequently. the results can be classified as good and reliable.
Forest Profile 1 on the whole has very low resistivity values around 20 Qm (Fig. 3a).Only in the lower west¬ ern part are resistivity values very high.Sass et al., in a similar environmental setting linked high resistivity of around 300 flm and more with limestone blocks.
the surrounding clays and marls showing much lower restistivities.This could possibly be the Situation in the research area described here.the limestone blocks either being a part of the old landslide mass or having been introduced later through rock fall.Similarly, the lower resistivity values in this profile may represent either marls and/or clay.There appears to be a sliding plane at a depth of 15 lo 20 m.
The 2D of Forest Profile 2 seems to be quite differ¬ ent (Fig. 3b).Low resistivities.like those observed in Forest Profile 1, are only found at certain points.High resistivity values dominate, indicating that many more limestone blocks can be found within the old landslide body than initially measured.Unfortunately, no clear sliding plane could be detected.
Longitudinal Profile la shows nicely the higher resis¬ tivity of the talus slope, which mainly consists of lime¬ stone debris (Fig. 3c).However, the talus thickness could not be defined satisfactorily, ranging from 5- 8 m to 15 m.Further downslope, the limestone blocks appear to be smaller.From profile meter 54-60, the road causes high resistivity values for the first 2 to 3 m.
The three disconnected blocks in the central part of the profile could point to parts of old landslide masses.It is assumed that lower resistivity values between these blocks indicate areas of high moisture content.
In this scenario, the sliding plane was established at a depth of 15 m.When interpreting resistivity data, it should be kept in mind that inversion parameters can be changed and further information included.Thus, a second run-through of the inversion routine was carried out, using a priori information.robust filtering and enhanc- ing horizontal features (Longitudinal Profile lb).The different results of the two longitudinal profiles are shown in Fig. 3d.The greatest difference between them appears to be in the deeper layers of the profiles.For the constrained inversion routine (lb).clear bounda¬ ries for the talus slope (7-8 m thick) as well as for the landslide mass (11-12 m thick) could be identified.Furthermore, the relevant RMS-error is smaller here than for the Standard inversion routine (la).Although this is generally positive, it does not necessarily mean that the constrained inversion result is more reliable.

Discussion
The resistivity differences between the two forest pro¬ files are at times greater than 1000 Qm.This enormous ränge is mainly caused by changes in the moisture regime.The April measurement was heavily influ¬ enced by extensive and exceptional snow melting in spring of 2005.However, it is surprising lhat the high resistivity values taken to indicate limestone blocks were significantly lowered during a period of high moisture content.One explanation is that the permeability of the limestone blocks is such that percolation is intensified when enough water is available.Despite these uncertainties, the results confirm the possibility of monitoring soil moisture conditions in landslides using DC resistivity.
Like in most modelling studies.it is often possible to fit the data to an already existing geological model.This is exemplified by the longitudinal profiles to some degree.However.fitting results to a pre-existing model does not necessarily lead to a more realistic model.In the case described herein, the sliding plane and the loose material/bedrock inferface could not be ultimately determined using geoelectric resistivity and limited borehole information alone.Even where bore¬ hole information (Lic02) was available.the bedrock could not be detected.This could be due to the high clay content (30-70%) within the old landslide mass.the resistivity properties thereof being too similar to the bedrock.Thus. a multi-geophysical approach does appear necessary.Hecht (2003).for example.was able to detect sliding plane and bedrock interface using seismic refraction in a similar geological setting. 6

Conclusion
In this study, it was shown that geophysical meth¬ ods are valuable tools for the extraction of informa¬ tion about the subsurface.Although the extent of the landslide investigated herein could not be deter¬ mined fully.the suitability and limitations of certain resistivity methods could be demonstrated.It may be concluded that decisions about choice of geophysical method should be made on a case-to-case basis.taking individual landslide characteristics into account.as a one-time successful application in a particular area does not guarantee continued success.even if used in the same area for landslides within a similar geological context.
It is foreseen to continue monitoring water content within the landslide using 2D resistivity tomography at least on a monthly basis.The results will hopefully contribute towards a better understanding of differ¬ ent types of recent landslide activities.Furthermore, by comparing the resistivity results with rainfall totals and movement measured through inclinometers. it may be possible to determine critical moisture levels.This could contribute towards the development of early warning Systems for landslides using geoelectrical methods.Abstract: Subsurface investigations of landslides using geophysical methods -geoelectrical applica¬ tions in the Swabian Alb (Germany) Landslides occur frequently all over the world, caus- ing at times considerable economic damage, injuries and even death.In order to improve hazard assess¬ ment, common landslide types of a given region need to be investigated in detail.While traditional techniques of subsurface investigation are expensive and only provide point information, geophysical methods are suitable tools for gathering 2D and 3D information on the subsurface quickly, reliably and cost-effectively.
In this study, the suitability and limitations of 2D resistivity for the determination of landslide extent, structure and soil moisture conditions are presented.For this purpose, two identical profiles were taken during a two-month period.Significant differences in electrical resistivity (>1000 Qm) due to varying soil moisture conditions were observed.Using various inversion parameters, it was possible to model two distinct subsurface images.Regrettably, the sliding plane could not be detected reliably, possibly due to the homogeniety of the landslide material and under¬ lying bedrock.

Fig. 2 :
Fig.2: Comparison of extent of old landslide activity.as mapped byDongus (1977) at a scale of 1:50000.and recent landslide activities.Locations of recent investigation sites are indicated.