Abstract
Schmidt-hammer rebound values (R-values) enable relative-age dating of landforms, with R-values relating to degree of weathering and therefore length of exposure. This method – recently termed as Schmidt-hammer exposure-age dating (SHD) – was applied to date five rock glaciers (size range, 0.01–0.12 km2) and one recent rockfall deposit at the study area Schöderkogel-Eisenhut, in the Schladminger Tauern Range (14°03′E, 47°15′N), Austria. The rock glaciers consist of gneiss or high metamorphic series of mica-schist that are comparable in their R-values. Four of them are relict (permafrost absent) and one is intact (containing patches of permafrost). On each of the five rock glaciers, SHD was carried out at 4–6 sites (50 measurements per site) along a longitudinal transect from the frontal ridge to the root zone. Results at all five rock glaciers are generally consistent with each other sharing statistically significant R-values along transects. The range between the highest and the lowest mean R-value at each of the five rock glaciers is 9.9–5.2. Using rock glacier length and surface velocity data from nearby sites, the rock glacier development must have lasted for several thousand years. Furthermore, by using SHD results from rock glaciers of known age from other sites in the region with comparable geology, approximate surface ages of 6.7–11.4 ka were estimated. This indicates long formation periods for all five rock glaciers. Our results suggest that many of the 1300 relict rock glaciers in central and eastern Austria were formed over a long period during the Lateglacial and Holocene period.
Keywords
Introduction
Active rock glaciers are common and characteristic large-scale flow features originating from thick debris accumulations (talus and/or moraine material) in high-relief alpine environments which are under cryogenic conditions for a long period of time. Surface morphology, extent and shape are the cumulative result of their entire formation period and therefore the climatic past (Barsch, 1996). Dating of such landforms might be an important source of palaeoclimatic information revealing for instance duration of periods favourable for rock glacier development or minimum ages of rock glaciers (e.g. Frauenfelder et al., 2005).
The Austrian Alps have a high density of rock glaciers. About 1500 rock glaciers have been mapped in the central and eastern Austrian Alps (Lieb, 1996) and were catalogued in the Rock Glacier Inventory of Central and Eastern Austria (Kellerer-Pirklbauer et al., 2010; Lieb et al., 2010). About 80% of them are considered to be relict. Most are located in the Hohe and Niedere Tauern Ranges (Figure 1) which cover about 9000 km2 of Austria (Kellerer-Pirklbauer, 2008b). Regional estimates of the Lateglacial and Holocene development of different generations of glaciers as well as rock glaciers in the Tauern Ranges are widely based on morphological criteria (e.g. altidudes of different stages of terminal moraines), snowline depression and estimates of the depression of the lower limit of permafrost (e.g. Fabiani, 1969; Lieb, 1987). However, information about ages and formation periods of specific rock glaciers in the Tauern Ranges are rare with only few exceptions (e.g. Kellerer-Pirklbauer, 2008a; Nagl, 1976). Furthermore, absolute dating for only two rock glaciers in the Austrian Alps has so far been carried out (Ivy-Ochs et al., 2009).

Geographical and geological setting of the study area and the Schmidt-hammer measurement sites. (a) Overview map and extent of the Hohe and Niedere Tauern Ranges. (b) The Schöderkogel-Eisenhut study area, Schladminger Tauern Range and its spatial relationship to the Schmidt-hammer measurement site on a moraine ridge in the Eselsberg Valley. (c) Detailed map of the Schöderkogel-Eisenhut study area with the Schmidt-hammer measurement points at the five rock glaciers. (d) Simplified geological map based on Thurner (1958) and Sapusek and Suette (1987). Hillshade with 10 m grid resolution provided by GIS Steiermark.
Accurate dating of rock glaciers is difficult and is best achieved by using a combination of relative and absolute dating methods, as proposed by Haeberli et al. (2003). Absolute dating methods applied for dating rock glacier surfaces (e.g. cosmogenic dating) are still at an initial stage of development, are expensive and have their own suites of assumptions and errors (e.g. Ivy-Ochs et al., 2009). Reliable radiocarbon dating, for instance, is impossible at high altitudes because of the lack of suitable organic material. In contrast, relative dating methods (e.g. Schmidt-hammer exposure-age dating (SHD), lichenometry and weathering-rind thickness) are substantially cheaper and hence have been used more frequently. Despite this fact, many rock glaciers are clearly too old for reliable lichenometric dating. Therefore, this specific method is generally unsuited to rock glacier dating. Schmidt-hammer rebound values seem to correlate with weathering rind thickness almost linearly (Laustela et al., 2003). Furthermore, the chronological resolution of relative dated rock glacier surfaces is often only relatively coarse (Kellerer-Pirklbauer, 2008a). Recent approaches using the Schmidt-hammer as a calibrated-age dating tool in Holocene glaciated environments by using age control points, revealed predicted age errors ranging from ±246–632 years (Matthews and Owen, 2011) to ±355–455 years (Shakesby et al., 2011), ±500–900 years (Matthews and Winkler, 2011) and ±600–700 years (Shakesby et al., 2006).
In this study, Schmidt-hammer exposure-age dating (SHD) was used for relative age dating of five rock glaciers in the Schladminger Tauern Range, one of the five subunits of the Niedere Tauern Range, Austria. For the first time, dating results of rock glaciers located in the Niedere Tauern Range are presented. Our SHD results are compared with dates from a moraine ridge of Lateglacial age 8 km E of the study area and with R-values on rock glaciers in the neighbouring Hohen Tauern Range. Therefore, this paper aims to improve the understanding of rock glacier formation and development during the Lateglacial and Holocene period in central Austria.
Study area and relevant landforms
Overview
The study area is located in the Schöderkogel-Eisenhut area, Schladminger Tauern Range, Austria at 14°03′E, 47°15′N, an area of about 3 km2 (Figure 1). The study area consists of a series of cirques facing E to NE (Lieb, 1996). They are mostly occupied by rock glaciers or tarn lakes. The study area is confined in the W by a NW–SE trending mountain crest with the mountain summits Schöderkogel (2500 m a.s.l.), Breitdach (2451 m a.s.l.), Arkogel (2441 m a.s.l.) and Eisenhut (2456 m a.s.l.). Based on Thurner (1958) and Sapusek and Suette (1987), bedrock in the area is dominated by mica-schist and gneiss with some amphibolite and marble lenses (Figure 1).
Important for this study and for the applied SHD method is the fact that the type of mica schist occurring south of the main alpine divide in our study area is difficult to distinguish from the gneiss because of a smooth transition between the two (Metz, 1976; H Proske and T Untersweg, May 2011, personal communication). This type of mica-schist comprises high metamorphic series and is compact and hence very similar to the gneiss in the study area (Metz, 1976). Both the high metamorphic mica-schist and gneiss, tend to produce boulders with medium-grained surface textures and hence comparable surface roughness characteristics (Untersweg, 2000).
From temperature data from the meteorological station in Oberwölz (20 km to the E, 810 m a.s.l.) provided by the Central Institute for Meteorology and Geodynamics and using a vertical temperature gradient of 0.0065°C/m, the mean annual air temperature in the study area during the period 1971–2000 at 2000 m a.s.l. is estimated as −1.3°C. This suggests that permafrost could still be present considering air temperature alone. Considering, furthermore, the ground cooling effect of coarse boulders (Gruber and Hoelzle, 2008) occurring to lower elevations in the study area, one might expect permafrost occurrence even at relatively low elevations with positive mean annual air temperatures.
At present, the Schladminger Tauern Range has no glaciers and there are only a few notable-sized perennial snow patches. Recent climate change has reduced the spatial extent as well as the number of perennial snow patches substantially. At the end of the Pleistocene during the Lateglacial period (14.5–11.6 ka according to Ivy-Ochs et al., 2008), valley glaciers in the Schladminger Tauern Range retreated, were replaced by cirque glaciers and finally disappeared. Rock glaciers were formed in the ice-free cirques and developed during the later part of the Lateglacial as well as during the Holocene period (Lieb, 1987, 1996). Most of the rock glaciers in the Schladminger Tauern Range are relict today and contain no permafrost. Only a few rock glaciers in the highest and most sheltered locations still contain patches of permafrost (Kellerer-Pirklbauer and Kühnast, 2009; Kellerer-Pirklbauer et al., 2010).
Present permafrost distribution has been modelled using two different empirical approaches for the Schladminger Tauern Range. The empirical relationship between mean annual air temperature and permafrost existence shows that <1% of the study area is potentially underlain by discontinuous and <7% by sporadic permafrost. In contrast, the empirical values of probable lower limits of discontinuous permafrost occurrence in a nearby region in combination with correction values showed that only about 0.2% were potentially underlain by discontinuous permafrost (Kellerer-Pirklbauer, 2005). These results clearly show the low importance of permafrost in the region at present. In contrast, the snowline in the region at some time in the past must have been substantially lower (order of 400–500 m) in order for rock glacier formation to have occurred.
Studied Landforms
The landforms studied in the Schöderkogel-Eisenhut area comprise five rock glaciers and one recent rockfall deposit. The five rock glaciers vary between 0.01 and 0.12 km2 in size and are 190–750 m in length. Four rock glaciers are regarded as relict as they contain no permafrost and lack detectable horizontal surface movement. The four relict rock glaciers are the Gamskar (here abbreviated as GA or as mu162 according to the rock glacier inventory by Lieb, 1996), Eisenhut (EI/mu164), Sulzkogel (SU/ mu165) and Breitdach (BR/ mu166) rock glaciers. The Schöderkogel rock glacier (SC/mu167) is considered as intact (sensu Barsch, 1996) still containing patches of permafrost but without information about ongoing movement. Horizontal surface movement has not been measured for this rock glacier.
Additionally to the permafrost modelling results described above, Kellerer-Pirklbauer (2005) detected permafrost in the study area at the root zone of the rock glacier SC based on measurements of the bottom temperature of snow cover (BTS) with BTS values down to −4.2°C below a complete winter snow cover of at least 1 m in thickness. Furthermore, spring water temperature measurements for GA of 1.9–2.2°C (Rode, 2009) also indicate possible permafrost existence (following the approach by Haeberli, 1975) in the study area.
At the rock glaciers GA, EI, SU and SC, gneiss is the dominant bedrock type in the source headwall area with a 90–100% areal cover. In contrast, at BR, mica-schist prevails (90%) mixed with amphibolite. Figure 2 shows exemplarily the field situation at the two rock glaciers GA and SC.The rockfall deposit covers an area of 0.004 km2 and consists solely of mica-schist (Table 1).

Two of the five studied rock glaciers: (a) SC; (b) GA. Schmidt-hammer measurement sites are indicated for both rock glaciers. Stippled lines mark alignment of the longitudinal transects. Photographs by M Rode.
Details of sampled landforms. Rock glacier codes in brackets are according to the Rock Glacier Inventory of Central and Eastern Austria (Kellerer-Pirklbauer et al., 2010; Lieb et al., 2010). Bedrock type is according to Thurner (1958) and Sapusek and Suette (1987).
In order to provide an estimation of the absolute ages of the R-values from the rock glaciers, it was necessary to carry out additional SHD measurements on rock surfaces of known age. Therefore, it was also attempted to date a terminal moraine ridge of Lateglacial age 8 km E of the study area (Figure 1b). No landform was suitable as an age control point in the close vicinity to the study area. The distinct terminal moraine ridge is curved in plan view and crosses the entire Eselsberg Valley. It is separated into two equally sized parts by the erosional forces of the Eselsberg Creek. SHD was carried out on the ridge at the orographic left side at 1535 m a.s.l. This ridge has a length of about 100 m and a height of up to 5 m. According to Lazar et al. (1988), the moraine was formed during the Gschnitz Stadial. This stadial was later dated to 15.4 ka BP based on the locus typicus moraines in the Gschnitz Valley, Western Austria (Kerschner and Ivy-Ochs, 2007). The moraine ridge in the Eselsberg Valley is mostly covered by a complete vegetation cover developed on soil.
Methods and data base
The Schmidt-hammer method has been applied as a relative dating tool in geomorphology since its first application by Matthews and Shakesby (1984). Since first use, this method has been increasingly applied in periglacial and glacial studies for the relative dating of bedrock and boulder surfaces including, for example, moraines (e.g. Matthews and Shakesby, 1984; Rune and Sjåstad, 2000; Winkler, 2009; Winkler and Shakesby, 1995), rockfall deposits (Nesje et al., 1994), trimlines (Anderson et al., 1998; Ballantyne, 1997) and, relevant to the present study, rock glaciers (Frauenfelder et al., 2005; Haeberli et al., 2003; Kellerer-Pirklbauer, 2008a; Kellerer-Pirklbauer et al., 2008).
The Schmidt-hammer is a light and portable instrument originally used for testing the surface strength of concrete by recording a rebound value (R-value) of a spring-loaded bolt made to impact the surface. The resulting R-value is proportional to the compressive strength of the concrete surface and gives a relative measure of the surface hardness. Therefore, for rock surfaces, the R-value can provide information on the degree of weathering and the time since exposure. High R-values are indicative of younger and low ones of older landforms. Measured R-values are also affected by lithology. As a result, only surfaces of the same,or at least similar lithology should be compared (Haeberli et al., 2003; McCarroll, 1989; Shakesby et al., 2006).
In this study, an analogue N-type Schmidt-hammer (the most used version by geomorphologists; Goudie, 2006) manufactured by Uniteam was used (Figure 3). On each of the five rock glaciers, four to six SHD measurement sites aligned along a longitudinal transect (=former central flow line) between the root zone and the frontal ridge were measured. The transect length varied between 200 (SC) and 790 m (BR). The distance between two different SHD sites was between 50 and 270 m. R-values were taken from as small as possible an area at each site (<10 m2). Measurements on rock glaciers were made on angular boulders (with frost-shattered or broken surfaces) located on ridge crests and elevated convex locations to minimise the possible protective influence of late-lying snow on weathering rates. Otherwise the degree of weathering over time on the rocks could not increase regular and this consequently affects the measured R-values (Ballantyne et al., 1989; Ericson, 2004; Kellerer-Pirklbauer, 2008a; Laustela et al., 2003).

(a) Schmidt-hammer (N-Typ), manufactured by Uniteam used in this study. (b) Field work at the GA-rock glacier at measurement point 5. Photographs by M Rode.
Figure 1c shows the location of the SHD measurement sites at the five rock glaciers and at the recent rockfall deposits close to SU. The rockfall event took place in June 2008, and was observed by the authors. SHD measurements carried out on fresh boulder surfaces formed during the rockfall were expected to yield the highest R-values and consequently their measurements were intended to help to convert R-values into absolute ages by acting as an age control point. In total, the 27 measurement sites of coarse angular boulders in the Schöderkogel-Eisenhut area are located between 1961 and 2475 m a.s.l.
Complementary measurements were carried out at the Lateglacial moraine in the Eselsberg Valley 8 km to the E. This site is located substantially lower than the other sites at an elevation of 1535 m a.s.l. On the generally well vegetated moraine, only larger rocks not covered by soil and vegetation were used for SHD measurements. Importantly, the rocks sampled on the moraine had in general a smooth surface texture resulting from glacial transport. The smoother surface texture and the substantially lower elevation are major limitations of this site as an age control point. However, for completeness the data from this site are also presented here.
Sampled stable boulders in each of the five study sites were selected on the basis of comparable lithology (compact mica-schist and gneiss) depending on the local lithology in the source headwall area – see Table 1) and surface texture. The surface texture of the measured mica-schist or gneiss boulders in the study area was comparable (cf. Williams and Robinson, 1983). As explained above, the high metamorphic mica-schist and the gneiss are similar in texture and weathering behaviour and are difficult to distinguish in the field. Therefore, it is considered acceptable that the SDH results from the two lithologies, mica-schist and gneiss prevailing in the study area are comparable because of similar rock surface hardness, resistance to weathering and surface roughness (McCarroll, 1989).
The measured surfaces were dry, flat, clean and lichen-free surface and lacked visual joints and cracks (Haeberli et al., 2003; McCarroll, 1989; Shakesby et al., 2006). Measurements where taken horizontally because Schmidt hammer readings vary non-linearly with the angle at which the instrument is held (e.g. Day and Goudie, 1977). The means of 50 individual readings obtained from at 25 boulders per site (four impacts per boulder; only the two middle values being noted) and 95% confidence intervals were defined for all sites (Kellerer-Pirklbauer, 2008a; Matthews and Shakesby, 1984; Shakesby et al., 2006). Reasons for excluding the highest and lowest values at one boulder were to remove the outliers (as suggested by Selby, 1980) by following comparable sample designs as applied by e.g. Matthews and Shakesby (1984), Boelhouwers et al. (1999), Aoyama (2005) or Kellerer-Pirklbauer (2008a, b). The means are representative of the effective hardness of the analysed surface. The 95% confidence interval is indicative of the standard error and statistically significant age differences between measurement sites (Kellerer-Pirklbauer, 2008a; Matthews and Shakesby, 1984; Shakesby et al., 2006). Measurement sites were mapped using a Garmin eTrex VistaC GPS device.
Results
The SHD results of all 26 measurement sites on the rock glaciers in the study area as well as of the two additional sites are summarised in Figure 4 and Table 2. The frequency distributions for all measurement sites are unimodal, and skewness is generally low to moderate (Table 3). Negative skewness suggests a somewhat lower mean R-value (hence higher age) than observed; positive skewness suggests a somewhat higher mean R-value (and hence lower age) than indicated by the results (Kellerer-Pirklbauer, 2008a).

Results of the 28 Schmidt-hammer measurement sites. For the rock glaciers, the R-values are plotted against the distance from the rock glacier front. The uppermost measurement site is always located in the rooting zone of the rock glacier and therefore outside the rock glacier body. The best-fit line of the arithmetic mean R-values and 95% confidence limits at each study point are indicated. Numerals in the graph refer to the measurement sites in Figure 1. R2=coefficient of determination, p=statistical significance.
A summary of Schmidt-hammer results for all studied landforms. 50 individual Schmidt-hammer readings were carried out at each measurement point.
Schmidt-hammer results for all 28 measurement sites and information regarding elevation. Mean R-values with 95% confidence limits and skewness (SK) are indicated.
At GA, the mean R-values at the six measurement sites range from 28.1 at the front to 41.2 at root zone. The six sample sites cover an R-value range of 13.1. The 95% confidence intervals are less than ±1.22. The mean R-value significantly increases from the front to the root zone (p<0.05) despite the R-value outlier at measurement site GA4. At this site, the results indicate older surfaces compared with sites GA2 and GA3. The distance between lowest and the highest measurement site along the transect is about 660 m. Despite the fact that the dominant lithology at this rock glacier is gneiss, the measured R-values are similar to those measured at the BR rock glacier, which consists almost entirely of mica-schist.
At EI the mean R-values at the four measurement sites range from 26.0 at the front to 40.1 at the root zone, covering an R-value range of 14.1. The 95% confidence intervals are below ±1.26. The results show significant differences (p<0.01) between clasts at the front compared with those at the root zone indicating substantial age differences (Figure 4). The distance between the lowest and highest measurement sites is only 260 m.
At SU measurements by SHD were carried out at six sites along an almost 500 m long transect. Mean R-values range from 32.0 at the front to 45.6 at the root zone. The low 95% confidence intervals again show highly significant age differences along the transect (p<0.01). No outlier was measured at this rock glacier.
At BR, the only mica-schist dominated rock glacier, the horizontal distance between the lowest and highest measurement point was about 800 m. The mean R-values at the six measurement sites range from 26.1 at the front to 41.3 at the root zone covering a R-value range of 15.2. These values are very similar to those at GA and EI. Mean R-values of BR decrease again statistically significant (p<0.01) towards the root zone of the rock glacier. One exception is site BR4, which yielded an older age compared with the lower one (i.e. BR3). The 95% confidence intervals are less than ±1.18 indicating again robust results.
Finally, at the intact SC rock glacier, the mean R-values at the four measurement sites range from 24.4 at the front to 34.3 at the root zone covering a R-value range of only 9.9. The 95% confidence intervals are less than ±1.31. Also at the fifth studied rock glacier the results show statistically significantly lower R-values (p<0.05) at the front compared with higher R-values at the root zone. SC is mainly composed of gneissic rocks as at GA, EI and SU. However, the R-values are generally lower compared with the other three rock glaciers. SC yielded the lowest R-values of all five studied rock glaciers at the front as well as at the root zone. Site SC1 was also the site with the lowest mean R-value of all 28 SHD measurement sites (Table 3). This seems to indicate that SC1 is the oldest surface of all measurement sites if the other factors affecting R-value are excluded as discussed below.
The SHD measurement site at the rockfall deposits consisting of mica-schist yielded the third highest mean R-value and hence indicated a rather young surface age, as expected. However, it is important to note that the two neighbouring sites, SU5 and SU6, seem to be younger compared with the recent rockfall deposits to judge from the mean R-values. This is possibly a result of the boulders at sites SU5 and SU6 being composed of slightly harder gneiss.
The measurement site at the moraine, located 8 km to the east of the study area, yielded a mean R-value of 33.1. Disregarding the substantially lower elevation of the moraine site (influencing weathering history during the Lateglacial and Holocene periods) and differences in surface texture (smoothed surfaces versus broken), this R-value result would suggest that the dated Gschnitz moraines are not the oldest of all the 28 sites.
Discussion
Relative-ages
When analysing the Schmidt-hammer literature, it becomes evident that R-value decrease over time is different in different alpine climates and in lithologies (e.g. Evans et al., 1999; Winkler, 2000, 2005; Winkler and Shakesby, 1995). Mean R-value differences of >10 suggest periods of thousands ranging up to more than 10,000 years (e.g. Frauenfelder et al., 2005; Kellerer-Pirklbauer et al., 2008).
The confidence limits at all 28 measurement sites are relatively low (0.73–1.31) indicating reliable mean values. At all five rock glaciers, significantly lower R-values at the highest (youngest) compared with the lowest (oldest) measurement points were recorded indicating different rock exposure ages. For two rock glaciers (GA, BR), the measured R-values along the longitudinal transect are not constant but affected by outliers. This can be explained by three main effects. First, relevant lithological differences affected the SHD results. Second, relatively fresh debris could have been transported from the valley/cirque sides by debris-charged snow avalanches onto the central part of the rock glacier surface and scattered over a part of it. This process would cause a mixture of older and younger rock surfaces which would further lead to a mixed SHD signal mirrored by a high confidence limit and a non-unimodal distribution of the single values. Third, differences in weathering efficiency along the transect due to protection from weathering by perennial or late-lying snow cover (Ballantyne et al., 1989) might have occurred at some time in the Lateglacial and/or Holocene periods. Related to this effect, SHD sites now on upstanding parts of the rock glacier surface, might have been in topographic depressions at some time of the rock glacier development. The later location would also favour snow cover preservation leading to a perennial or late-lying snow cover which would have influenced weathering. At the measurement sites of GA and BR, the first and the third effects are more likely to have occurred because the confidence limits at both measurement sites are low (1.01–1.15) and frequency distribution is unimodal.
At the four relict rock glaciers, the R-value ranges between the youngest (root zone of rock glacier) and the oldest (rock glacier front) measurement site are consistent (13.1–15.2). It is important that the lowest and highest R-values as well as the R-value range at the three rock glaciers consisting predominantly of gneiss (GA, EI and SU) are more or less identical to those of the mica-schist dominated rock glacier BR. The intact SC consists mainly of gneiss but differs from the other four rock glaciers in terms of absolute R-values and R-value range. At SC, despite the lowest and highest R-values at both ends of the longitudinal transect being measured, the R-value range is only 9.9. This suggests a shorter formation period by a factor of about 0.65–0.75 compared with those of the other four rock glaciers. Furthermore, if only the absolute R-values are considered, the results indicate that SC is the oldest of the five studied rock glaciers.
A more plausible explanation for differences in R-values might be that the lower R-values at SC are linked to more extreme weathering conditions than at the four other rock glaciers because of a substantially higher elevation of this landform. The front of SC1 is located at 2422 m a.s.l. whereas the measurement sites at the root zones are, respectively, 121 m lower at BR and 238 m lower at GA.
In general, the increase in mean R-values from the fronts to the root zones at all five studied rock glaciers in gneiss and mica-schist is similar. Therefore, the four relict rock glaciers were presumably formed synchronously whereas SC started to form some thousands of years later. On normalised distance-from-rock-glacier-front versus R-value diagrams, the measured R-values of the northernmost rock glacier (SC) compared to the southernmost one (GA) are much lower (R-value differences from 3.7 to 6.9) and the calculated linear trend at SC is slightly less steep (Figure 5).
Reasons for the steeper trend in R-values at GA are the more distinct and steep cirque headwalls which would have delivered debris over a longer time. In contrast, the rock walls of the Schöderkogel cirque containing SC are eroded and the remaining cirque headwalls are low. These circumstances would have reduced the debris production of the rock wall and, additionally, limited the necessary debris accumulations for the further development of SC (Rode and Kellerer-Pirklbauer, 2009) Moreover, when downslope movement of the rock glacier SC was prevented by a distinct ridge of sediments and bedrock at its front, there the rock glacier would have become inactive (cf. Barsch, 1996). Taking all this into account, the two rock glaciers, GA and SC, were formed under quite different circumstances and had different periods of formation.
In comparison with GA and SC, the R-values of the clasts on the three rock glaciers in between revealed the following pattern (Figure 5). The R-values of the SU are higher by a factor of 1.2, than those on BR and EI, but they are comparable with GA. The steepness of the linear trends of SU and BR are almost identical with the one for GA. The steepest trend line calculated for EI was possibly related to faster rock glacier formation. The less steeply inclined trend at SC indicates slower rock glacier formation. However, in general all five rock glaciers have comparable linear trends indicating roughly equal rock glacier development.

Mean R-values and 95% confidence limits plotted against relative distance from the rock glacier front for all five rock glaciers (a). BR consists predominantly of mica-schist, whereas the other four rock glaciers are composed of gneiss. Best-fit lines are indicated at all rock glaciers for comparison. (b) Arithmetic mean R-values of all 28 measurement sites plotted against elevations and differentiated into the different lithologies and landforms.
Figure 5B depicts the relationship between altitude, landform studied, bedrock type and mean R-value. Results show that there is no specific trend and thus there are no homogeneous weathering processes determinable, regardless of whether the rocks are of the same lithology or at the same elevation. A trend in different degree of weathering is only detected on the specific surfaces of the five different rock glaciers – the increase of the R-values from the front to the root zone indicating rock glacier-specific differences.
The results from the rockfall deposits and the Lateglacial moraine are more difficult to interpret. There are several possible explanations for the relatively high R-values and hence suggested younger age of the Lateglacial moraine. One important explanation is surely the previously mentioned smoother initial surface texture of the measured clasts at this site yielding higher readings. A second possible explanation is the big elevation difference between the moraine (1535 m a.s.l.) and the lowest rock glacier front (SU1, 1969 m a.s.l.) of 426 m (Table 3, Figure 5b) affecting rock weathering intensity. The rocks at the low-lying moraine ridge of Lateglacial age are eventually less weathered compared with the rocks belonging to the must be younger rock glaciers at higher elevations. A third explanation might be progressive coverage of the measured clasts by soil and vegetation during the Lateglacial and Holocene period at the moraine site. For a long period of time the measured clasts were possibly covered by a slowly developing soil layer and vegetation which formed after moraine stabilisation. This soil and vegetation cover protected the clasts against weathering. Since this area was used as alpine pasture, soil erosion occurred through grazing practices and some rock material was consequently exposed and later measured by us. For these reasons, the results from the moraine site cannot be used to convert the relative into absolute dates by establishing age-calibration curves for rock glaciers (cf. Kellerer-Pirklbauer, 2008a).
Reasons for the low R-values of the recent rockfall could be a combination of differences in the lithology and the delay between the moment of rock fracture and actual rockfall. This delay could take between several hundreds and thousands of years allowing a long period of rock weakening by chemical and physical weathering in the crack (e.g. Kellerer-Pirklbauer, 2008b; Matthews and Shakesby, 2004;). The mean R-value of the recent rockfall is 42.0. Therefore, the values obtained at this site are 1.3 to 2.6 R-values lower than the two uppermost measurement sites at the neighbouring SU. This would indicate that the new rock deposits of the fall from June 2007 are older than the two uppermost measurement sites at SU which is clearly not possible.
Absolute-age dating estimates
Based on the lengths of the five rock glaciers (Table 1), assuming similar surface creep velocities compared with presently active and morphologically comparable rock glaciers in the nearby Tauern Range (Dösen Rock Glacier movement 1954–2005: 13.4–37.4 cm/yr; Kaufmann et al., 2007) and taking surface weathering as constant over time (which was certainly not the case), we estimate that the rock glaciers took about up to 500 (SC), 1950 (EI), 3550 (SU), 5150 (GA) and 5600 (BR) years to reach their present extent. Hence, if these dates are correct, rock glacier favourable conditions would have lasted for at least several thousands of years in order for the five studied landforms to have formed.
In the Schladminger Tauern Range strong glacier retreat started at the end of the Gschnitz Stadial (dated to 15.4 ka BP; Kerschner and Ivy-Ochs, 2007) and valley glaciers changed to cirque glaciers. Later, in the Senders and Daun Stadials also the cirques were also becoming ice-free and rock glaciers were able to form (Lazar et al., 1988; Lieb, 1987). As presented by Ivy-Ochs et al. (2009), the Central European Alps were also strongly affected by cold climatic conditions during the Younger Dryas (or Egesen, 12.9–11.7 ka) and the subsequent Preboreal oscillation until 10.5 ka BP were suitable for rock glacier development. According to the same authors, climatic conditions in the Alps were however less favourable for glacier expansion (and also for rapid rock glacier evolution) between 10.5 and 3.3 ka BP. This indicates that the most active phase of all five studied rock glaciers must have been the later part of the Lateglacial period and the early Holocene.
Owing to the fact that our SHD results of the oldest (Lateglacial moraine in the Eselsberg Valley) and youngest (rockfall deposit near the SU) sites are not high quality age control points and hence cannot be used for establishing an age-calibration curve as developed elsewhere (e.g. Matthews and Owen, 2011; Matthews and Winkler, 2011; Shakesby et al., 2006, 2011; Winkler, 2005), absolute ages for the five rock glaciers can only be given within approximate boundaries. Kellerer-Pirklbauer (2008a) established age-calibration curves for nearby rock glaciers consisting of mica-schist and gneiss. These rock glacier sites are located some 80 km to the west of the study area in the Hohe Tauern Range at comparable elevations (2300–2700 m a.s.l.), so that, they would have experienced a similar climatic history since Lateglacial times. The results from these two rock glaciers yielded a mean decrease of 1.33 (mica-schist rock glacier) to 1.46 (gneiss rock glacier) R-values per 1 ka. Using these values, absolute surface ages for the five studied rock glaciers would be theoretically 6.7–7.4 ka for SC, 9.0–9.8 ka for GA, 9.3–10.2 ka for SU, 9.7–10.6 ka for EI, and 10.4–11.4 ka for presumably the oldest rock glacier (BR).
A remaining question is the relationship between weathering (and hence R-value decrease) and time. At the beginning this relationship, is linear but weathering rate must decrease over time when the weathering rind is thickening (see discussions in e.g. Kellerer-Pirklbauer, 2008a; Kellerer-Pirklbauer et al., 2008; Shakesby et al., 2006, 2011). Flaking of the weathered surface of rock (partly or entirely) further complicates age estimate partly restarting the weathering process at the rock surface. A further aspect to consider is the assumption that the total age of a rock glacier might be 2–5 times greater than that obtained for the surface (Kääb, 2005). If correct, the five studied landforms would be old features that started to form soon after glacier disappearance (possibly incorporating glacier remains into the rock glacier body) in the cirques, and continued to develop in particular up to 10.5 ka BP, and than their development slowed down.
Conclusion
This study clearly shows that the SHD method is a powerful tool in rock glacier dating by providing comparable and consistent results at five neighbouring rock glaciers at least at the site specific (i.e. rock glacier) scale. However, study areas with a varying geology in the area under investigation cause complications specifically where the change in the lithology occurs smoothly rather than abruptly. In the study area, this transition occurs from high metamorphic series of mica-schist to gneiss. This also points to the general problem that mica-schist (as one rock type example) might occur in different forms (fine-grained, medium-grained, coarse-grained) causing different surface textures and roughness characteristics even over short distances. At the Schöderkogel-Eisenhut area the high metamorphic series of mica-schist and gneiss yield comparable SHD.
For absolute-age dating estimates it is necessary to have at least two age control points in a study area and to be sure that the landform (e.g. moraine ridge of known age) has experienced the same climatic history (at a similar elevation), consists of the same lithology, and sampled boulders had a similar initial surface roughness (e.g. glacial/smooth versus periglacial/fractured surface origin) as for the five investigated rock glacier. Sites possibly covered in the past by soil or vegetation should be avoided.
The significant age differences indicated by R-values for boulders at sites on longitudinal transects of the five rock glaciers demonstrate the development of the rock glaciers during their active phase and enable the establishment of statistically significant longer temporal coarse-scale relative chronologies. Reasons for the apparently good R-value–age relationship in this study are the relatively homogeneous lithologies with the occurrence of comparable rock types (gneiss and mica-schist) which do not show any systematic differences and thus the R-values are suitable for comparison.
Converting the R-values to absolute ages in the study area was not possible for various reasons. However, comparison of our results with velocity measurements and SHD results from rock glaciers in neighbouring areas suggests that the rock glaciers were formed over a period of about 500–5600 years and have surface ages of c.6.7–11.4 ka indicating that the old landforms evolved over a long period during the Lateglacial to early Holocene. The rock glaciers were not able to form before 15.4 ka BP and the most active phase of rock glacier formation would have ended around 10.5 ka BP because of warming climate conditions after the Preboreal oscillation. The degree of activity was reduced substantially afterwards. Only the one studied intact rock glacier (SC) is possibly still slowly moving downvalley.
More generally, our results suggest that many of the known 1300 relict rock glaciers in central and eastern Austria were formed over a long period during the Lateglacial and Holocene period.
Footnotes
Acknowledgements
Many thanks to Anneliese and Eduard Rode, Roland Kranabeter, as well to Claudia Stoiser for their essential support during field work. The very valuable comments and suggestions by two anonymous reviewers and the reviewer Stefan Winkler are highly appreciated. Herwig Proske (geologist at Joanneum Research, Graz) and Thomas Untersweg (geologist at the Geological Survey of Austria, Vienna) are thanked for their discussions regarding the characteristics of the mica-schist and gneiss in the study area.
This study was supported by the projects ‘ALPCHANGE – Climate Change and Impacts in Southern Austrian Alpine Regions’ financed by the Austrian Science Fund (FWF) through project no. FWF P18304-N10 and ‘PermaNET – Permafrost long-term monitoring network’. PermaNET is part of the European Territorial Cooperation and co-funded by the European Regional Development Fund (ERDF) in the scope of the Alpine Space Programme.
