Origin of block accumulations based on the near-surface geophysics

3.1 ERT

We selected the multi-electrode resistivity method of measurement and subsequent processing by ERT as the main exploration technique in both sites. This method was especially selected because of the following reasons: (1) it has been widely used in previous studies of block accumulations with different origin [13,38,39,40,41,42,43,44,45], (2) it has a high resolution [46,47], and (3) it has a large depth range [24,48,49], as well as (4) it processes data relatively fast [16,50]. High-quality grounding is crucial to acquire reliable data [51]. Therefore, in Skalka site, we grounded the electrodes using sponges saturated in electrolyte (hypersaline solution) (Figure 3a) to ensure sufficient conductive contact with the surface. This approach has been previously verified [12,52]. We carried out the ERT measurements in Skalka site in 2009, and the first results were reported by ref. [53], which marked the initial stage of the block accumulation research. In our study, we used the original data from 2009 and reprocessed them using a more advanced approach (i.e. using a newer version of processing Res2Dinv software [v3.56.22 instead of v3.54.44] and advanced data interpretation skills). We used the Wenner–Schlumberger (WS) electrode array in both sites since it gives sufficient resolution, depth range, and sensitivity to both horizontal and vertical structures. It is also robust towards geoelectrical noise [16,54,55,56]. In Slunečná, we also used the Wenner alpha (Wα) electrode configuration, which, due to its lower sensitivity, possess higher robustness to noise [16,54]. We applied stacking and standard deviation to improve data quality. Basic information on the measurement geometry is summarised in Table 2.

Figure 3

Various geophysical methods ERT (a and b), SSR (c), GPR (d and e), EMI (f and g), used for studying the block accumulations in Slunečná site (b, e, and g) and the central block accumulation in Skalka site (a, c, d, and f).

We processed the measured data using Res2Dinv tomographic inversion software, and topographic corrections were applied. All data were filtered using RMS statistics tool and the most erroneous or outlying values were excluded (approximately 6–9% of data). We inverted data from both arrays (WS and Wα) and we used a joint inversion model. We applied both standard (L2 norm) and robust (L1 norm) inversions [57]. However, models using robust blocks did not deliver satisfactory explanatory power. Therefore, we further interpreted the standard inversion models. We optimised damping factor values [49,58,59,60] to overcome the inherent non-uniqueness of the inverse problem and to avoid fitting noise in the ERT measurements and, therefore, suppress the effect of the heterogeneity of the block fields within inverse resistivity models.

3.2 SSR

SSR measurements are generally used to evaluate the elastic properties of block accumulations [11,20,61,62,63]. In this study, we used geophones with a frequency of 10 Hz to measure the SSR. The source of the seismic signal was an 8 kg sledge hammer (Figure 3c) and a metal pad, which we used to determine the initial time t₀. In both localities, we used distant shotpoints providing sufficient coverage of the measured profile margins by seismic beams and signal summing to acquire better quality data. Data were processed using (1) the plus-minus method [64] to determine the first refraction boundary which we expected to be the lower limit of the investigated block accumulations and (2) seismic tomography (ST) [63,65] which shows overall geological conditions based on seismic velocity distribution. First arrival times were picked in SeisImager-Pickwin software (Geometrics, Inc./OYO Corporation). Data quality check was performed using a reciprocal time check function in SeisImager-Plotrefa (Geometrics, Inc./OYO Corporation) with maximum reciprocal time error 5% and, if needed, the first arrivals of seismic signal were further adjusted. We used ST in the Rayfract program (Intelligent Resources, Inc.). The advantage of this software is the calculation of tomography taking Fresnel zones into consideration [66,67], so the calculated velocity slices (showing the distribution of seismic P-wave velocities) better match the measured records. Both layered and gradient models were created in Rayfract. However, both models yielded similar results. As determining the thickness of block accumulations was one of the main goals of the study, we preferred layered models, which should better depict a layered environment. Topography corrections were used in all models. The interfaces obtained by the plus-minus method [64] were added to the ERT models for both localities to relate the SSR results to ERT results (like [68]).

3.3 GPR

GPR was used as the method using electromagnetic signals to reveal the possible depth and internal structures of the block accumulation [19,22,69,70,71,72,73]. In both locations, we performed measurements using the Malå Ramac system of GPR with X3M control unit and shielded antennas of 250 and 100 MHz (Figure 3d and e). The use of shielded antennas was important in Slunečná, which is forested. The trigger wheel or time-mode trigger could not have been used owing to the extreme conditions of block accumulations, which caused continuous motion with constant velocity impossible. Therefore, we used a 20 cm manual trigger in all cases (Figure 2c and d). The measured data from both locations were processed using ReflexW 6.0 software (Sandmeier software) similar to [74]. We introduced the topography of the environment into all GPR sections. We used the EM wave propagation velocity of 0.09 m/ns based on the assumption of non-homogeneous unsaturated stacking of loose blocks with predominantly air-filled inter-block voids [9,19,75]. We processed the radargrams using the following processing sequence: (1) Resorting – X-Flip Profile (due to extremely difficult operation with antennas and to save time, some profiles were measured in a reversed way), (2) static corrections – move start-time (first arrival time adjustment), (3) 1-D filter – subtract mean (dewow), (4) 2-D filter – background removal, (5a) gain – energy decay and (5b) gain – Y-manual gain (in a vertical direction), (6) 2-D filter – running average, and (7) topographic migrations. In some cases, topographic migrations provide more continuous reflections and, thus, its interpretations are mostly unambiguous [19].

3.4 EMI

We carried out EMI measurements to obtain information on the conductive and/or resistive properties of block accumulations [8,11,76,77,78]. In contrast to the ERT, EMI measurements do not require coupling with the subsurface using metal electrodes (see Figure 3f and g). The device uses contactless measurements based on the principle of EM induction [79,80]. We utilised a CMD Explorer conductivity meter (GF Instruments, Brno), which allowed us to take measurements in two different regimes, i.e. both vertical and horizontal orientations of the measuring dipole coils axes with different calibrated penetration depths (measuring position “Hi” refers to vertical orientation with a full depth range and “Lo” refers to horizontal orientation with a half depth range). In Slunečná, we also used a different conductivity probe CMD MiniExplorer (GF Instruments), which is designed for more detailed survey (as it measures in shallower depths compared with CMD Explorer, for details see below). Owing to the heterogeneity of the examined environment and the expected high range of the measured values, we used manually controlled data collection (sampled at 1 m intervals) in both localities (Figure 2e and f) to obtain more accurate results with direct control on data quality based on the measured standard deviation ( σ ) of up to 5%. Owing to the more accessible nature of the terrain, continuous data collection was also applied at the Slunečná site for experiments. We measured the apparent resistivity values at the depth levels (Explorer – Hi: −2.2, −4.2, and −6.7 m, Lo: −1.1, −2.1, and −3.3 m; MiniExplorer – Hi: −0.5, −1.0, and −1.8 m, Lo: −0.25, −0.5, and −0.9 m) calibrated by the manufacturer. We displayed the resultant apparent resistivity curves for the depth levels using Grapher 9 (Golden Software).

4.1 Slunečná site

We delineated block accumulations with different characteristics through geomorphological mapping of the block accumulation clusters in the Slunečná locality (Figure 4a). Most of the accumulations were less dense (types B1, B2, and B3, where blocks were not in a visible contact with each other), except for a single dense accumulation (situated in 228–265 m, Figure 5a), where blocks were in contact with each other – type B5 [4]. All accumulations were mostly composed of blocks with the longest axis of up to 1 m, with sparsely incorporated larger and smaller blocks. The distribution of blocks with different size was random. The spaces between the blocks were all filled with finer matrix and the two southernmost block accumulations had the highest slope angle.

Figure 4

Location of block accumulations at the Slunečná (a) and Skalka (b) sites. The types of block accumulations and position of the geophysical profile are shown. For block accumulation types description, see Figure 2. The surroundings of the mapped block accumulations at the Slunečná site are formed of a mixture of scattered islets of B1 and B2 block accumulation types.

Figure 5

Results of the survey in the Slunečná site. (a) Types of block accumulation (modified after [4]) along the geophysical profile, black dashed lines in ERT and SSR models, radargram and EMI curves denote the boundary between different types of block accumulation. (b) Cropped ERT subsurface resistivity distribution model (5th iteration standard inversion, RMS error 2.1%). White solid lines represent first and second interfaces (see below). (c) Subsurface contour model of seismic velocities (RMS error 4.4%) with first and second interfaces derived from Hagedoorn’s [64] plus-minus method (solid white lines). Solid black lines represent isolines 2 and 5 km/s. (d) Radargram (vertical exaggeration: two times) obtained by GPR (f = 100 MHz) with a significant reflector (yellow dashed line) and maximum effective penetration depth (white dashed line). (e) Detailed segment of radargram where B5 block accumulation changes into B3 and vice versa. (f) Apparent resistivity depth (d) curves obtained by EMI (CMD Explorer, Hi).

The integration of the WS and Wα electrode configurations in the ERT resulted in a model of the subsurface resistivity distribution (Figure 5b) with higher density of the measured data and low RMS error (2.1%). This model shows an increase in the thickness of high-resistive bodies (resistivity from 7 to 27.6 kΩm). The thickness in the first half of the profile (0–195 m) reaching a maximum of 7 m. The maximum thickness in the second half (195–270 m) of the profile was 9 m. This pattern was interrupted several times (at 200–205 and 245–250) with lower resistivity (varying from 4.4 to 7 kΩm). The greater thickness of high-resistive bodies in the second half of the ERT profile most likely reflects an increase in the thickness of the block accumulations. Similarly, bodies with very high resistivities (over 20 kΩm) may indicate areas with higher block concentrations (accumulations with air-filled inter-block spaces, i.e. B5 category). The lower resistivity bodies likely represent block accumulations where inter-block spaces are filled with fine-grained matrix. At a depth of ca. 7 m lies conductive band, which is characterised by low resistivities (<4.4 kΩm). At a position of approx. 200 m, there is a zone that interrupts the low resistive environment of the second layer. An environment with the lowest resistivities (<4.4 kΩm) probably corresponds to the disrupted (likely weathered) parent rocks [1]. The relatively higher resistivity (4.4–11 kΩm) at a depth of approximately 15 m, which forms the third layer, most likely corresponds to intact rock.

The seismic model (with RMS error 4.4%) consists of three layers. The solid white lines in Figure 5c represent two subsurface refraction boundaries detected by the SSR plus-minus method [64]. The first interface within the inverse seismic velocity model is partially copied by an isoline of 2 km/s at an average depth of 7 m. The second interface is partially copied with isolines of 5 km/s. In parts of the profile around the 180, 220, and 250 m, both isolines drop to greater depths. This shift in isolines to varying depth may be caused by the selective weathering of bedrock [81,82,83]. The first (uppermost) layer with a constant thickness was delineated by isolines of approximately 2 km/s. The second layer was delineated by isolines of 2 and 5 km/s. The thickness of the second (middle) layer was found to change along the profile. The third (bottom) layer was bounded by an isoline of 5 km/s, which corresponds to the seismic wave propagation velocity in intact granites [84,85,86,87,88].

Two depth profiles of the subsurface environment were measured by GPR using antennas with f = 100 and 250 MHz, respectively. We obtained similar result for both antennas, but the 100 MHz antenna exhibited a higher penetration depth. Therefore, we have reported data obtained only from the 100 MHz antenna (Figure 5d). We achieved a penetration depth of approximately 6–9 m (Figure 5d). In the all resulting radargrams, it was relatively difficult to detect any unequivocal interface or significant changes in the subsurface environment, except for a structure at approximately 215 m (see yellow dashed line in Figure 5e) where the interfaces are markedly curved. The interfaces could represent a possible original surface of the currently infilled shallow depression there which also have a significant manifestation in SSR model (a rapid decrease in seismic velocity). In some places, the reflections reached deeper (e.g. at 220 and 255 m), which might indicate pockets of thicker debris (Figure 5d). The absence of interpretable reflection boundaries (but not a lack of reflections) is further discussed in the following sections.

We used EMI measurements to construct the apparent resistivity curves. The curves were scattered, and the measured values were highly variable (ranging from 200 to 1,100 Ωm, see Figure 5f). The variations in values obtained using CMD Mini Explorer conductivity meter were even more fluctuated, and the measured values often dropped to negative values. The best quality results were obtained by CMD Explorer conductivity meter with vertical coil setting (Hi position) in the manual measurement mode (Figure 5f). Measurements in continuous mode did not yield better results. The rapid change in the trends of the curves was observed at positions 200 and 205 m. The curves of the depth of −2.2 and −6.7 were smoother in the 120–200 (205) m section and were more variable in the second half of the profile. This is likely caused by the rapidly changing thickness of the block accumulation.

4.2 Skalka site

In Skalka site, we determined the surficial characteristics of the central block accumulation by geomorphological mapping (Figures 3 and 5a). We found block concentration of the central accumulation to be very high when the blocks were in contact with each other (Figure 3) and the inter-block voids were not filled with the matrix (i.e. category B6 [4]). We found blocks with the longest axis of ∼1 m, sparsely distributed smaller blocks (tens of centimetres), and/or larger blocks (up to 2 m) on the surface of the central accumulation. Blocks in the downslope part of the accumulation had slightly smaller diameter than those in the upslope part (as well as [53]).

The model of ERT measurements in Skalka site (with RMS error 5.8%) demonstrates the presence of several extremely high-resistive bodies (resistivity varying from 55 to 102 kΩm) near the surface (Figure 6b). Their occurrence is mainly linked to the upslope parts of the profile. Along the downslope part of the profile, the resistivity of the uppermost layer generally decreases. However, approximately in the third quarter of the profile (i.e. at 165 m), there is another extremely high-resistive body (from 55 to 102 kΩm). There is a relatively lower resistivity environment (from 5 to 30 kΩm) under the layer of these bodies (approximately 20 m deep). The high-resistivity bodies in the resistivity model (at approximately 100, 115, 130, and 165 m) most likely correspond to the stacking of blocks with large air-filled voids. The occurrence of lower resistivities at depths of up to 5 m (at approximately 120 m and from 140 to 160 m) most likely correspond to areas where the inter-block spaces are filled with a fine-grained matrix. A sudden increase in subsurface resistivity at 95 m probably indicates the upslope limit of the block accumulation, and the sharp reduction in resistivity at 175 m marks its downslope limit. High resistivities (>55 kΩm) between 95 and 175 m could thus likely be caused by more air filed voids within block accumulation, which correspond with the B6 type description [4].

Figure 6

Results of survey in the Skalka site area. (a) Different types of block accumulation [4] along the geophysical profile, black dashed lines in ERT and SSR models radargrams and EMI curves denote the boundary between different types of block accumulation. (b) Cropped model of subsurface distribution of the resistivity obtained by ERT (5th iteration standard inversion, RMS error 5.8%). White solid line represents first interface (see below). (c) Contour model of seismic velocities below the surface (RMS error 4.5%) with the first interface derived from Hagedoorn’s [64] plus-minus method (solid white line). Solid black lines represent isolines 2.5 and 4.5 km/s. (d) Radargram obtained using GPR (f = 100 MHz) with maximum effective penetration depth (white dashed line). (e) Detailed segment of radargram where B6 block accumulation changes into B4 and vice versa. (f) Apparent resistivity depth (d) curves obtained by EMI (CMD Explorer Lo – black curve and Hi – remaining curves). Note that the curves of nearly identical depths do not follow the same trend and tend to go against each other.

The model of seismic velocity isolines (with RMS error 4.5%) shown in the tomographic inverse section (Figure 6c) follows the slope topography, except for the upper half of the profile where they fall to greater depths. Based on the plus-minus method [64], a single interface was found that copies the slope topography along the entire profile at depths of 8‒10 m. This observation points to very low variability in the thickness of block accumulation and a double-layered environment. This refraction boundary follows an isoline of 2.5 km/s, which does not correspond to higher propagation velocity (>4.5 km/s) of seismic waves in intact gneisses [84,85,86,87,88]. The refraction boundary thus does not present the contact of the block accumulation with an intact rock. Below the block accumulation probably lies the layer of disrupted bedrock (e.g. by tectonics and frost shattering) and likely weathered parent rock, delineated by an isoline of 4.5 km/s, which corresponds to the velocity of propagation in the intact gneiss [85]. The total thickness of the accumulation and disrupted zone varies greatly over the course of the profile, exceeding 15 m at the deepest point in the first half of the profile.

We present the GPR results of Skalka site that were obtained only by 100 MHz antenna for the same reason as in the Slunečná site. Our results (Figure 6d) indicated similar depths with a narrow range between 6 and 9 m. However, the depth was approximately 7 m in majority of the profile. This is in accordance with the commonly accepted values of the effective depth range for a 100 MHz antenna, i.e. 6–7 m [22]. Furthermore, the reflections completely disappeared at greater depths, probably due to the loss of energy. As we reached the limits of the used antenna, we can only presume that we have reached the lower margin of the investigated block field. No major reflecting interfaces, indicating the presence of a possible internal structure (e.g. layering), were found in the sections. We found areas with stronger and weaker reflections (e.g. in Figure 6f) in the radargram after processing raw data (at approximately 140 and 160 m for the former and at approximately 120 m for the latter).

The apparent resistivity curves for the two coil orientations and for all depth ranges were constructed based on the EMI measurements. Despite the chosen approach of measurement (i.e. manual data acquisition), the measured values were highly variable (ranging from 400 to 1,600 Ωm). The CMD Explorer conductivity meter provides the best quality data (in all EMI settings), and a common trend could be interpreted from the curves acquired from its vertical coil orientation setting (Hi position). The curves acquired from similar depths using two different coil orientations (i.e. Lo [black] and Hi [green] in Figure 6e) did not show similar values of the apparent resistivities when mutually compared. An interpretation of the variable EMI curves, including a comparison with the results of other methods, is further discussed in the subsequent section.

5.1 Evaluation of the applicability of geophysical methods

Analysis of the results shows that the applied methods generated data with a variable quality. During the evaluation of their applicability, we considered the quality of the obtained data and considered the main difficulties of their acquisition, i.e. geophone and electrode placement, extremely difficult motion, and manipulation with the EMI probe or GPR antennas. Based on our outputs (see Table 3), we can state that (1) the SSR method is suitable for block accumulation research (similar to [20]), (2) the usage of ERT method is “conditional” depending on the quality of electrode grounding, (3) contactless EMI cannot be utilised, or with considerable limitations (we recommend that EMI should be used only as a complementary method), and finally, (4) the GPR method would be suitable to determine the thickness of block accumulation when uses low frequency (e.g. 20–30 MHz) antennas to achieve a greater maximum penetration depth [72].

5.1.3 GPR

Compared to ref. [74], the GPR did not provide any easily interpretable data. We could observe strong reflections (as in ref. [74]) in both radargrams, but the investigated environment was too heterogeneous. The inner structure (formed of loose blocks) was rather highly disordered (chaotic) for any anticipation of continuous reflecting interfaces (its graphical interpretation is provided in Figure 7). Therefore, it was not possible to accurately identify the assumed interfaces between the block accumulations and the underlying bedrock, as confirmed by other methods. Furthermore, the probable energy loss and attenuation of the residual signal at the depth of the assumed bedrock surface played a significant role in not allowing the reliable determination of accumulation thickness [22,92]. Similar problems were noted by ref. [93] in a landslide environment. In summary, the determination of the internal structure of block accumulation using GPR was rather unfeasible [72].

Figure 7

Schematic interpretation of the propagating EM waves, and the anticipated discontinuous GPR reflection boundaries on discrete blocks forming accumulations with an irregular architecture. Inter-block space is either air-filled (Skalka site) or partially filled with a fine-grained matrix (Slunečná site).

For both GPR antennas, the maximum depth range reported by ref. [22] was achieved in both sites. Even if we agree that block accumulations can provide suitable conditions for reaching the maximum penetration depth of the GPR method [26,94], we could not assert with absolute certainty that the obtained penetration depth exceeded the thickness of the investigated block accumulations as we had probably reached the method limitations. Meanwhile, the sudden disappearance of reflection might be caused by a strong signal attenuation in the relatively more conductive environment composed mostly of disrupted, weathered, and water-saturated parent rocks. We can conclude that the overall thickness assumed for the studied block layer approximately matched the GPR penetration depth.

The use of lower-frequency antennas with a greater penetration depth could be a solution for the above-mentioned problem. Sass [25] reached a depth of over 50 m on a talus cone using a frequency of 25 MHz. Otto and Sass [12] reached a depth of 40 m using the same antenna. These depths could be sufficient to detect the surface of the parent rock in the both locations. However, the use of low-frequency antennas has several limitations. Although the low-frequency GPR antenna (e.g. 20 or 30 MHz) delivers sufficient penetration depth, the resolution of such measurement and the ability to distinguish geological structures are rather low due to its long wavelength [94,95]. In addition, such antennas are not shielded, and, therefore, they are highly sensitive to geoelectric noise and unwanted reflections from trees and operators [19]. Schrott and Sass [20] considered the use of unshielded antennas in wooden terrains to be almost impossible. The lower-frequency antennas also have larger size. Even the transportation of a 100 MHz antenna with the size of approximately 1 m² to the survey site was extremely difficult. The measurement itself was very demanding and sometimes even dangerous. In accordance to ref. [20], due to the difficult terrain, we experienced a significantly longer measurement time compared to the measurements in easily accessible sights.

5.2 Internal structure and tested hypotheses of origin and evolution

We reconstructed the accumulation’s internal structure and thickness and tested the most relevant hypotheses of origin and evolution in the two studied sites mostly on the basis of the data obtained from the SSR and ERT surveys. This is because the EMI and GPR methods provided ambiguous results possibly due to insufficient penetration depth of GPR and highly scattered EMI curves. The thickness of the block accumulations in both sites ranged from 7 to 9 m. In the integrated geological model (Figures 8 and 9), the surface of unweathered parent rock was determined as the second velocity boundary as derived from the Hagedoorn’s plus-minus method [64,96] at Slunečná site and the isoline of 4.5 km/s in the seismic velocity model of Skalka site. This approach suggests that the overall depth of the block accumulations with the eluvium ranged from 10 to 15 m in Slunečná site and from 8 to 20 m in Skalka site. The individual types of block accumulation [4] were interpreted in a vertical section, and the internal structure of the block accumulation was outlined based solely on the ERT results.

Figure 8

Integrated geological model of the Slunečná site interpreting the subsurface conditions and particular types of block accumulations along the geophysical profile. Red indexes reflect the following: I. = body of parent rock; II. = the most probable body of the disturbed parent rock; III. = body of block accumulation; and a, b – seismic refraction boundaries derived from the plus-minus method [64]. Results obtained by processing our data collected in 2017.

Figure 9

Integrated geological model of the Skalka site interpreting the subsurface conditions and particular types of block accumulations along the geophysical profile. Red indexes show the following: I. = body of parent rock; II. = the most probable body of disturbed parent rock; III. = body of block accumulation; and a – seismic refraction boundary derived from the plus-minus method [64]. Results obtained by processing our own data collected in 2017.

The integrated geological model of the Slunečná site (Figure 8) shows that the results of the geophysical survey agree well with the results of geomorphological mapping, except for the area from 240 to 260 m, where a discrepancy between the mapped surficial accumulations of B5 and relatively lower resistivity subsurface bodies can be observed.

We do not favour the hypothesis of in situ weathering, because there are no rock outcrops, tors, or cliffs upslope the block accumulations and the thickness of accumulations ranged from 7 to 9 m (compare with maximum 3 m of in situ weathered mantle described by ref. [1]. We declined the hypothesis of sorted slope [31] based on the geomorphological mapping, because the accumulations are developed as restricted polygon (Figure 4) with no upslope and downslope continuation. None of the subsurface layer can be interpreted as the layer of periglacial cover bed concept because of their significantly larger thickness (compare several metres on Figure 8 and several tens of centimetres in refs. [32,33,34]). On the other hand, the highest category of the block accumulation is lying on the slope with the highest slope angle and vice versa. Based on the previously mentioned statements, we incline to gelifluction origin (probably in a cold periods of Pleistocene) of block accumulations in Slunečná site with different degrees of washing out of fines due to the slope angle [4,5].

The middle layer (second layer in Figure 8) cannot be of glaciogenic origin, although part of the Bohemian Massif was glaciated during the last ice age [97]. However, the site is located far from the hypothetical extent of the nearest palaeoglaciers [29,98]. Moreover, the periglacial cover bed concept or glaciogenic origin of the middle layer does not match our results in a geophysical characteristic, especially seismic velocities. The measured velocities (over 2.5 km/s in Figure 8) do not match either with the velocities of 0.3–0.6 km/s [88,99,100] anticipated for fine-grained aeolian sediments (typical for middle layer of cover bed [34]) or with the typical velocities of glaciogenic sediments (i.e. till) with <2.5 km/s [85,88,101,102]. These values are significantly lower than the measured velocities in the studied site. The low resistivities in the middle layer might be also caused by stagnant seepage water [98,103]. However, considering the position of the block accumulations at approximately 70 m above the valley floor on the Slunečná slope, we do not favour such an interpretation. Therefore, the middle layer is most probably disturbed bedrock, which gradually changes to the unweathered bedrock.

The integrated geological model reconstructed for Skalka site (Figure 9) shows that the subsurface extent of the block accumulation determined by the geophysical survey does not fully correspond to its surficial manifestation found by geomorphological mapping. The accumulation emerges at 106 m in the profile and submerges at 176 m. The geophysical survey may indicate the presence of a block accumulation at 95 m. The discrepancy in these results indicates the presence of a well-developed block accumulation covered by soil, with no surficial manifestation. Such outcomes confirm the importance of integrated interpretations and cross-validation of geophysical and geomorphological methods.

Based on the geomorphological mapping and geophysical survey, we incline to complex polygenetic origin of the block accumulation in Skalka site [3], which were probably created by in situ weathering and gelifluction transport. Because of the in situ weathering of bedrock and disintegration of rock scarps or tors lying upslope would not probably create a 7–9 metre thick layer of loose block (compare a few metres described by [1]), some material probably had to be transported by gelifluction from above lying gently sloped plateau, where intensive cryoplanation could took place [35]. This hypothesis is in accordance with the reduction of block size downslope along the profile. The accumulation is probably not a relict block stream [104], because these landforms most often lay on the bottom of valleys and are longer than wider [3,104]. Occurrence of relict rock glacier could explain the accumulation origin, but there are no morphological manifestations on the surface (such as steep front slope or transverse and longitudinal ridges described by refs. [3,105].

The middle layer in the upper part of the profile (second layer in Figure 9) is (similar to Slunečná site) most probably disturbed bedrock, which gradually changes into the unweathered bedrock. We do not support the glacial origin of this layer for the same reason as in Slunečná site. As well as the middle layer in Figure 9 cannot represent the middle layer in the cover bed concepts [32,33,34], due to its thickness and seismic velocities. The accumulation position almost on a high plateau without any peat bog and with a high slope angle excludes low resistivities caused by the stagnant seepage water.

In summary, we consider the allochthonous periglacial explanation of block accumulations origin in both sites with different degrees of washing out of fines in Slunečná site and with an interaction of in situ frost shattering in Skalka site. The research showed that the combination of suitable geophysical methods is valuable in the clarification of geomorphological hypotheses and confirmed the importance of using a multidisciplinary geophysical survey. In addition, Table 3 provides a clear comparison of the applicability of the methods for the investigation of block accumulations and the associated difficulties.