Introduction

Geophysical subsurface characterization of slopes prone to mass movements has an important meaning in landslide hazard evaluation (Mc Cann and Forster 1990; Caris and Van Asch 1991; Bogoslovsky and Ogilvy 1977; Jongmans and Garambois 2007; Pilecki 2017; Whiteley et al. 2019). Active seismic methods are commonly used on landslides, mainly P-wave refraction (Grandjean 2006; Samyn et al. 2012; Grit and Kanli 2016; Uhlemann et al. 2016; Huntley et al. 2019; Whiteley et al. 2019), multichannel analysis of surface waves (MASW) (Kanli 2010; Grit and Kanli 2016; Huntley et al. 2019; Whiteley et al. 2019) and rarely S-wave refraction (Bièvre et al. 2012; Uhlemann et al. 2016; Whiteley et al. 2019). In recent years, several applications of seismic interferometry (SI) with ambient noise have also been used on landslides (Jongmans et al. 2009; Renalier et al. 2010a, b; Mainsant et al. 2012; Pilz et al. 2013; Larose et al. 2015; Harba and Pilecki 2017; Hussain et al. 2017; Whiteley et al. 2019).

Body waves, in comparison with surface waves, have broader applications in geoengineering (Szreder et al. 2008; Pilecki 2008; Pilecki et al. 2017); however, methods based on surface waves such as MASW and SI quickly develop improving range and resolution of investigation (Kanli et al. 2006, 2008; Chmiel et al. 2019). The use of low-frequency noise below 0.4 Hz allows to image geological subsurface up to approximately 400–500 m (Czarny et al. 2019).

The comparison of MASW and SI in geoengineering applications has not been the subject of numerous studies (Isakow et al. 2015). Both methods allow to estimate S-wave velocity field based on inversion of dispersion curves of Rayleigh surface wave. They provide interesting results related to the imaging of changes in S-wave velocity field, courses of lithological boundaries or boundaries which are referred to the landslide’s slip surfaces. Moreover, the dynamic elastic moduli can be calculated if the P-wave velocity and volumetric density are known.

The approach of combining SI and MASW was presented by Cheng et al. (2015) in conditions of Han River levee and western Junggar Basin test sites in China. Authors proposed a weighted stacking technique which allows to retrieve greater signal-to-noise ratio of surface wave from seismic noise cross-correlation. What is more, they applied a bidirectional shot mode, which is based on the virtual source method, to sort common virtual shot gathers repeatedly. Such procedure allowed them to increase the processing efficiency and significantly increase the usefulness of seismic noise data. The joint of dispersion curves obtained from passive methods used by authors with active MASW results is presented in Fig. 1. The active MASW data allowed to obtain a wide frequency range of approximately 17–45 Hz; however, SI (as PSM) showed the Rayleigh dispersion curve for lower frequency below 15 Hz and for greater phase velocities. Both methods complemented each other very well in retrieving the information of Rayleigh surface wave and in consequence in imaging the geological subsurface.

Fig. 1
figure 1

Joint of dispersion curves from the results of passive methods: passive seismic method (PSM), spatial autocorrelation (SPAC), roadside passive MASW (roadside) and the active MASW (black dots) on Han River levee test site in China (Cheng et al. 2015)

Full size image

Similarly, Eikmeier et al. (2016) combined MASW and passive seismic method in a densely built-up area with high traffic noise in Sao Paulo in Brazil. In comparison with MASW results, the dispersion curves from passive seismic records sampled lower frequency and consequently it allowed to obtain greater range of depth models. Authors proposed the joint use of seismic noise and sledgehammer sources. Additionally, they applied the joint inversion of the fundamental and higher modes of dispersion curves. The proposed technique significantly improved the subsurface model by increasing the depth of recognition and made the model more consistent with the geological medium.

Considering that the result of MASW and SI is the S-wave velocity field, it is difficult to find studies comparing results of both methods applied in the same geological engineering conditions on landslide. In the study, we present a comparison of data acquisition, processing and interpretation between both methods applied in geological engineering conditions of Just-Tegoborze landslide in southern Poland. The study area is located in Carpathian flysch, in complicated geological medium prone to mass movements (Pilecki et al. 2007, 2008; Kogut et al. 2018; Stanisz and Pilecki 2018). We discuss similarities and differences in both methods, especially data processing and S-wave velocity field interpretation for landslide subsurface.

This study is a continuation of work presented by Harba and Pilecki (2017) on Just-Tegoborze landslide. In conditions of high seismic noise generated by high traffic with many heavy trucks on the state road, active seismic data from commonly used methods such MASW are difficult to interpret. However, such conditions are favourable for SI method, being effective even with 60-min recording of seismic noise. Therefore, the comparison of SI and MASW measurements in high seismic noise conditions is significant in case of estimation of S-wave velocity field and the evaluation of landslide hazard.

Geological setting

The study was carried out on Just-Tegoborze landslide which is located on the south of Poland in Malopolska Voivodeship (Fig. 2).

Fig. 2
figure 2

Location of the study area of the Just-Tegoborze landslide (Harba and Pilecki 2017)

Full size image

Landslide covers an area of 21.3 ha below St. Just Pass between Jodlowiec Wielki Mt and St. Just Mt. The average landslide’s slope is 11.8° and the landslide’s azimuth is 130°. Geologically, the study area lies in Magura Nappe in the Outer Carpathians. Landslide bedrock (flysch bedrock) is made of variegated shales, shales and sandstones and is covered discordant by weathered rock and clayey colluvium (Fig. 3a).

Fig. 3
figure 3

Geological engineering cross section of the part of Just-Tegoborze landslide with location of SI and MASW profiles (a); lithostratigraphic profile from 3I/P borehole (b); symbols’ description refers to a and b (based on Harba and Pilecki 2017; Isakow et al. 2015)

Full size image

Failure surfaces were indicated on the study area at about 10 and 12 m depth within colluvial deposits and about 14–17 m depth between colluvium and less weathered flysch bedrock (Harba and Pilecki 2017). Seepages were observed on the depth between 0.7 and 5.8 m in 3I/P borehole on the study area (Fig. 3b). The main reasons of mass movement activation on Just-Tegoborze landslide are water infiltration, the occurrence of variegated shales in landslide’s bedrock and disturbed local tectonics associated with overthrust of sub-Magura beds by Magura beds (Pilecki and Harba 2015). High traffic intensity on the state road no. 75, which intersects the Just-Tegoborze landslide, has a significant impact on slope stability as well.

Methodology

Seismic acquisition was carried out along two profiles which crossed at the point of the 3I/P borehole on 29 June 2015 (AM–AM′ and BM–BM′ profiles for MASW) and on 9 July 2015 (A–A′ and B–B′ profiles for SI) (Fig. 4).

Fig. 4
figure 4

Study area on the part of Just-Tegoborze landslide with seismic profiles (based on Harba and Pilecki 2017)

Full size image

Seismic acquisition varied significantly between MASW and SI. In MASW method, seismic energy was released in a metal plate using a 5-kg sledgehammer. Geophones of 4.5 Hz natural frequency were used with 3-m intervals. Profiles AM–AM′ and BM–BM′ had 52 m length each. However, in SI method local seismic noise generated by intense vehicle traffic was the seismic source. Seismic noise came from three directions which allowed to obtain sufficient seismic noise recordings during 60 min. It was registered by 12 Güralp CMG-6TD three-component broadband seismometers installed along profiles A–A′ and B–B′. Intervals between seismometers were 10 m, 15 m and 20 m. Profiles A–A′ and B–B′ have 65 m and 55 m, respectively.

MASW data processing started with band-pass filtration due to seismic noise generated by high traffic on the state road. In order to extract Rayleigh wave energy, the frequency–wavenumber (f–k) transform was calculated. After that, muting was applied to improve signal-to-noise ratio. However, SI data processing methodology differs from the MASW one. After band-pass filtration, amplitudes of vertical component of seismic noise vibrations were normalized using 1-bit normalization. Then, cross-correlation of seismic noise between every pair of receivers was applied which is the base of SI method. Cross-correlation led to obtain series of empirical Green’s functions between pairs of receivers for Rayleigh surface wave. Every receiver was then represented by so-called common virtual source gather (CVSG) as a bidirectional shot with causal and acausal signal parts. Subsequently, f–k transform and muting were applied. As a result of data processing in MASW and SI methods, dispersion images of the Rayleigh surface wave were obtained. Interpretation for both methods was carried out according to the same methodology. Based on the analysis of dispersion curves, fundamental modes of the Rayleigh surface wave were picked. Then, inversion of picked modes was applied using genetic algorithm which resulted in 1D S-wave velocity models. Genetic algorithm is a nonlinear global search optimization method which uses the genetic and evolution principles in order to obtain the optimum model from a set of models (Dal Moro et al. 2007; Salamanca et al. 2017). The last step of interpretation includes model visualization as the 2D S-wave velocity sections for A–A′, B–B′, AM–AM′ and BM–BM′ profiles.

Table 1 shows a summary of similarities and differences between acquisition and data processing and interpretation in MASW and SI methods. The basic assumptions of the MASW method were broadly explained by Park et al. (1999) and Dal Moro (2015). However, basic principles of SI are described by Weaver and Lobkis (2001) and Wapenaar and Fokkema (2006).

Table 1 Comparison of acquisition and processing and interpretation parameters between MASW and seismic interferometry applied on Just-Tegoborze landslide
Full size table

Results and analysis

Figure 5 presents examples of seismic signals and its amplitude spectra used for processing and interpretation in the MASW and SI methods. In case of SI, the period of 60 min of seismic noise registration was assumed as sufficient in tests and used as a seismic source in further processing steps (Fig. 5c). In both methods, amplitude spectra were normalized to that of MASW in order to compare energy of both seismic sources. It shows that maximum value of seismic noise amplitude spectrum was not greater than about 70% of maximum value of MASW source signal spectrum (Fig. 5b, d).

Fig. 5
figure 5

An example of characteristic seismogram from active MASW (a) and its amplitude spectrum (b); an example of characteristic registration of seismic noise used in seismic interferometry (c) and its amplitude spectrum (d) gained on Just-Tegoborze landslide. Amplitude spectra were normalized to MASW maximal amplitude spectrum

Full size image

Moreover, amplitude spectrum of seismic noise clearly shows that the dominant energy lies in a frequency band from approximately 7–20 Hz (Fig. 5d). However, data obtained from MASW acquisition indicate that useful signal, which was generated by a sledgehammer, has a range of frequency between approximately 10 Hz and 40 Hz (Fig. 5a, b). It shows that MASW signal has a wider frequency range but seismic noise has a lower limit 7 Hz of frequency range.

Dispersion images obtained from MASW and SI are shown in Fig. 6. Phase velocity ranges from 140 to 200 m/s for MASW and from 150 to 260 m/s for SI. Fundamental mode is well visible on both dispersion images, however, higher modes are better distinguishable on SI dispersion image. Fundamental mode of SI has the greatest energy for the frequency range of 8–22 Hz which is correlated with frequency band of local seismic noise generated by high traffic on the state road (Fig. 5). In case of MASW, fundamental mode has the greatest energy for the higher and wider frequency band of 12–34 Hz compared to SI. Resolution of dispersion spectra from MASW and SI is comparable.

Fig. 6
figure 6

Dispersion images with picked fundamental modes (M0) from MASW (a) and from seismic interferometry (b)

Full size image

Figure 7 presents S-wave velocity seismic sections obtained as a result of data processing and interpretation from MASW and SI methods. Landslide clayey colluvium is characterized by the values of S-wave velocity up to about 250–270 m/s. Lithologically, this is a very complicated layer of soil and mixture of clay and loam with rock fragments. The clayey colluvium is the most exposed to changes in water infiltration and very sensitive on water content which resulted in its rigidity changes and consequently S-wave velocity changes. According to MASW results, the thickness of clayey colluvium reaches about 8–10 m. However, on the sections from SI it reaches about 9–11 m. Such difference can be explained by the changes in water content due to storm rainfalls which had taken place before SI measurements.

Fig. 7
figure 7

S-wave velocity sections from MASW—longitudinal (a) and perpendicular (b) to landslide axis and from seismic interferometry with seismic noise—longitudinal (c) and perpendicular (d) gained on Just-Tegoborze landslide. Red thick dashed line indicates interpreted seismic border between clayey colluvium and weathered rock colluvium; black thick dashed line indicates interpreted seismic border between weathered rock colluvium and less weathered flysch bedrock (3I/P denotes borehole/cross point)

Full size image

At the depth of 12–16 m, probable lithological boundary between weathered rock colluvium and less weathered flysch bedrock is identified on SI S-wave velocity sections. It is characterized by S-wave velocity of approximately 340–350 m/s. Such boundary has been indicated as a main slip surface shown in Fig. 3. However, the depth range of MASW is limited to approximately 15 m and did not allow to distinguish the boundary between weathered rock colluvium and less weathered flysch bedrock. The slip surfaces in colluvial layers, known from geological engineering investigation, are not observed on either SI or MASW S-wave velocity sections.

Conclusions

In the study, we compared results of estimation of S-wave velocity fields by multichannel analysis of surface waves and seismic interferometry with use of ambient noise in the same geological engineering conditions of Just-Tegoborze landslide. The study allowed to formulate the following conclusions:

  • both active seismic source used in MASW and passive seismic source used in SI were successfully applied in Just-Tegoborze landslide geological engineering conditions. Time of acquisition was comparable; however, the MASW method was much more labour-consuming,

  • seismic noise generated by intense vehicle traffic and recorded for 60 min was a sufficient seismic source for SI processing and interpretation. However, this seismic noise was a disturbance in active MASW which decreases signal-to-noise ratio. Despite that, MASW data were successfully applied for data processing and interpretation,

  • dominant energy in amplitude spectrum of seismic noise contained a frequency range from approximately 7 Hz to 20 Hz. However, the MASW useful signal, which was generated by a sledgehammer, had greater values and wider band of frequency approximately from 10 Hz to 40 Hz. Maximum value of seismic noise amplitude spectrum was not greater than about 70% of maximum value of MASW source signal spectrum,

  • MASW and SI provided comparable resolution of dispersion curves,

  • fundamental mode of dispersion curve was well visible on both dispersion images; however, higher modes were better distinguishable for SI. In SI, fundamental mode had the greatest energy for the 8–22 Hz frequency band which was well correlated with the band of local seismic noise. In MASW, the frequency range of fundamental mode was from 12 to 34 Hz,

  • both methods resulted in 2D S-wave velocity estimation of landslide’s subsurface. It showed the similar course of seismic boundary between clayey colluvium and weathered rock colluvium. The differences are probably related to varied water content within clayey colluvium because of the storm rainfalls before the SI measurements. The boundary between weathered rock colluvium and less weathered flysch bedrock was observed only on SI S-wave velocity sections,

  • slip surfaces in clayey colluvium were not observed on either SI or MASW S-wave velocity sections,

  • in general SI was characterized by greater depth range than MASW. However, lateral resolution was better for MASW where intervals between geophones were smaller than intervals between seismometers in SI and source signal comprising higher frequencies.