Nationwide aerial laser scanning reveals relict rock glaciers and protalus ramparts in Slovenia

In 2015 the nationwide aerial laser scanning (lidar) of Slovenia became publicly available. These data enable a wide range of detailed geomorphological studies, also in areas that are less accessible or covered with dense vegetation. This makes it 10 possible to identify potential rock glaciers and protalus ramparts in the Slovenian mountains. The laser scanning products, the grey-shaded terrain model and the classified point cloud were used to identify and measure these features. All the mountainous areas at elevations of approximately 1200 m above sea level (a.s.l.) were evaluated. During the Alpine Late Glacial period these were in glacial and periglacial conditions. The mountain ranges of the Julian Alps, Karavanks, KamnikSavinja Alps, Pohorje and Dinaric mountains (Trnovski Gozd and Snežnik) were evaluated. Twenty potential rock glaciers 15 and eight potential protalus ramparts were found. They are the most abundant in the Karavanks, followed by the Julian Alps, with one potential rock glacier also on the Snežnik plateau. The majority of the potential rock glaciers are probably relicts, due to the heavy vegetation cover, the low mean elevations (between 1040 m and 1850 m a.s.l.) and because their slopes are directed more towards southern directions (65 % of rock glaciers). The identified rock glaciers rarely exceed 600 m in length. The terminus slope angles of the identified objects are from 20° to 40°. Three of the identified protalus ramparts can 20 be regarded as relict, due to the total vegetation cover; the remaining five can be regarded as intact. The potential protalus ramparts are found at elevations between 1220 m and 1950 m a.s.l. All the identified protalus ramparts are directed towards southern directions, with terminus slope angles from 30° to 40°. The spatial distribution of the discussed permafrost objects in Slovenia with regards to the bedrock composition presented on the geological map of Slovenia (scale 1:250,000) reveals that 75 % of all objects can be found in thick-bedded Dachstein limestones with transitions to dolomite, while almost all the 25 remaining objects are found in Triassic dolomite.


Introduction
The availability of very-high-resolution digital elevation models derived from aerial laser scanning has only recently enabled spatial distribution studies of glaciers and permafrost features (Abermann et al., 2010;Knoll and Kerscher, 2009;Colucci et al., 2013;Krainer and Ribis, 2012). Mountain permafrost describes the areas where the subsurface material has below-zero temperatures all year round and can be recognised in talus areas in a continuum of landforms from rock glaciers to protalus 5 ramparts. Previously, other theories have been proposed to describe the possible origins of such features, from retreating tongue glaciers buried under thick debris cover to rock or talus avalanches (Whalley and Azizi, 2003;Haeberli et al. 2006;Haeberli et al. 2010;Berthling, 2011;Benn and Evans, 2013;Scotti et al., 2013;Scapozza, 2015). Morphologically, rock glaciers are tongue-like or lobate bodies with ridges and furrows, usually resembling small glaciers; they have steep fronts, steep sides and a length greater than their width, and they mainly exist on the valley floor. The protalus ramparts have a 10 (Cannone and Gerdol, 2003;Burga et al., 2004). In the European Alps 75 % of rock glaciers are becoming relict (Cremonese et al., 2011). Intact rock glaciers are mainly exposed to the north, while the relict glaciers are exposed to the south (Krainer and Ribis, 2012;Scotti et al., 2013). Relict forms are usually located at lower elevations than the intact forms (Scotti et al, 2013). The intact rock glaciers are believed to have been formed about 3100 years ago, the relict rock glaciers during the Alpine Late Glacial and they decayed between the end of the Alpine Late Glacial and the beginning of the Holocene around 5 11.6 ka ago (Frauenfelder and Kääb, 2000;Harris et al., 2009;Böhlert et al., 2011aBöhlert et al., , 2011bScotti et al., 2013).
Different regional permafrost inventories around the world have been made (Shakesby et al., 1987;Frauenfelder and Kääb 2000;Dramis et al., 2003;Millar and Westfall, 2008;Krainer and Ribis, 2012;Colucci et al., 2013;Stotti et al., 2013;Scapozza, 2015;Schmid et al., 2015). Additionally, a permafrost inventory for the whole of the European Alps has been seek to prove that stand-alone aerial laser scanning data is enough to discover larger potential relict permafrost features. 20 10 made, with only Monaco, Liechtenstein and Slovenia not being included (Cremonese et al., 2011). The first nationwide aerial laser scanning of Slovenia, made during 2014-2015, allows us to fill this gap in the permafrost inventory of European Alps with data from Slovenia. In Slovenia, 6 % of its area is at elevations higher than 1200 m (Fridl et al., 1998) (Fig. 1), covering the highest peaks of the Alps (i.e., Julian Alps, Karavanks, Kamnik-Savinja Alps) and the Dinaric Mountains (i.e., Trnovski Gozd and Snežnik). Those areas were, in the time of the Alpine Late Glacial, in glacial or periglacial conditions 15 (Colucci et al., 2014;Ferk et al., 2015). Due to the fact that no active rock glaciers are known to exist in Slovenia, the laser scanning data enables us, for the first time, to study the relict permafrost features hidden under the vegetation on a general scale. By compiling an inventory of potential relict rock glaciers and protalus ramparts for Slovenia from the aerial laser scanning data, we seek to study: (i) the average slope exposition of the relict permafrost features, (ii) the spatial distribution with regard to the bedrock composition, and (iii) the average elevation and the size of the recognized features. Finally, we The Cryosphere Discuss., doi:10.5194/tc-2016Discuss., doi:10.5194/tc- -86, 2016 Manuscript under review for journal The Cryosphere Published: 13 June 2016 c Author(s) 2016. CC-BY 3.0 License. Figure 1. The potential areas of permafrost features in Slovenia at elevations higher than 1200 m and the division into aerial laser scanning densities (5 points/m 2 for the majority of Slovenia, 2 points/m 2 only in marked areas) and the years when the data were acquired.

5
The source for the rock glacier and protalus rampart study was the nationwide aerial laser scanning data for Slovenia that was mainly acquired in 2014 and 2015 ( Fig. 1). For the first time these data made it possible to gain a detailed look under the forests that cover approximately 60 % of the Slovenian territory (Hladnik and Žižek, 2012). The previously existing nationwide digital terrain models were based on aerial photogrammetry, where a smaller spatial resolution (5 m × 5 m) and vertical accuracy were achieved, and therefore they were unable to present the geomorphologic features under the forests.
10 The aerial laser scanning of Slovenia enabled the production of a digital terrain model with a 1 m × 1 m spatial resolution.
The horizontal accuracy of the laser scanning data was 30 cm and the vertical accuracy was 15 cm. The laser scanning was (ii) the terrain point cloud in LAS format (just points classified as ground); 10 (iii) digital terrain model in ASCI format with a grid size of 1 m × 1 m; (iv) grey-shaded terrain model stored in a georeferenced TIFF format with a pixel size of 1 m × 1 m produced from the digital terrain model.
The classification of the point cloud on the ground, buildings and vegetation was made using gLidar software, where the algorithms described in  and  were used. The elevation data are stored in 15 heights above sea level and in two horizontal coordinate systems: the new D96/TM (geodetic datum 1996, Transverse Mercator projectionthis is the Slovenian realisation of ETRS89) and the old D48/GK (geodetic datum 1948, Gauss-Krüger projection). For details about the transformations between the two systems see Berk and Komadina (2013). The data are stored in files covering 1 km 2 or 5 km 2 for the grey-shaded terrain model. The nomenclature for 1 km 2 is defined by its lower-left corner; therefore, the file GK_410_145 is described with the coordinates 412000 m, 148000 m in the D48/GK 20 coordinate system. The data are freely available (http://gis.arso.gov.si/evode/profile.aspx?id=atlas_voda_Lidar%40Arso).
An additional source for an easier identification of the talus material behind the inspected objects was the orthophoto images made as part of the Cyclical aerial surveying of Slovenia (CAS) from large-format aerial photogrammetric images.
Stereoscopic images for orthophoto production were acquired in a different year than the aerial laser scanning. The orthophoto is produced on a scale of 1:5,000 with a pixel size of 0.5 m for the mountainous areas. The CAS project started in 25 1975; the permanent 3-year cycle for imaging the whole of Slovenia was introduced in 1985; and the orthophoto is regarded as its standard product from 2001 onwards.
The study of the bedrock composition of the identified objects is based on the Geological map of Slovenia on a scale of 1:250,000 (Buser, 2010). This map was compiled from the basic geological map on a scale of 1:100,000, which was later updated using the facts gained from extensive fieldwork lasting more than 15 years (Komac, 2005). 30 The Cryosphere Discuss., doi:10.5194/tc-2016-86, 2016 Manuscript under review for journal The Cryosphere Published: 13 June 2016 c Author(s) 2016. CC-BY 3.0 License.

Method
The inventory of potential rock glaciers and protalus ramparts was compiled via a visual inspection of the grey-shaded terrain model with a 1 m × 1 m pixel size. Due to the nicely described relief features in the grey-shaded terrain model the identification is similar to identifying objects from Google Maps (Schmid et al., 2015). The rock glaciers and protalus characteristics of such potential objects in Slovenia.
The objects were classified on the basis of their morphology (Scotti et al., 2013), where rock glaciers can be distinguished depending on the source of the sedimentary material that is transported downslope on the talus and debris types. A talus rock glacier is located at the base of a talus slope. A debris rock glacier forms downslope from the end moraines of small glaciers, mainly located in cirques, and is composed of reworked glacier debris. Only the objects that have some kind of terminal 15 edge, with a steep front, were included, so as not to confuse them with glacier moraines. Due to the fact that all the examples of protalus ramparts presented in this study are found on talus slopes we will treat them as a relict embryonic stage of talus rock glaciers, as proposed by Scapozza (2015), and not as pronival ramparts. It would be hard to distinguish relict pronival ramparts from relict terminal moraines based only on a grey-shaded terrain model identification, especially in cirques that are no longer filled with snowfields. Therefore, the ramps of debris formed at the end of small cirques were not accounted 20 like the slope aspect, the geographical location (the name of the nearest mountain) and the existence of nearby watercourses 25 were defined from the grey-shaded terrain model and additional detailed topographic maps. The elevations, terminus slope angles and vegetation coverage were measured in a classified point cloud. The bedrock composition is defined based on a general geological map of Slovenia. 5 ramparts were identified on the basis of their flow patterns and structures, which include ridges and furrows and the frontal ridge appearance as well as the texture differences of such features compared to the surrounding slopes (Fig. 2). When an object was found, detailed measurements were made in a classified point cloud. This classified point cloud enables measurements of the heights, slope angles, areas and a general study of the vegetation coverage. Fig. 2 shows a protalus rampart (no. PR_1) and a rock glacier (no. RG_1), both under coniferous forest. No detailed fieldwork was conducted yet, 10 due to the fact that the intention of this research was to obtain a general overview of the spatial distribution and the common for in our inventory as we treat them as glacial terminal moraines.
The landform attributes include geographical coordinates (described with aerial laser scanning 5 km 2 in which the object is presented), mountain sectors, elevation (minimum of terminus, maximum and mean), height of the terminus, terminus slope angle, length and width, slope aspects, vegetation coverage, and bedrock composition ( mean that it has the potential to be an intact rock glacier. This rock glacier is located on the Austrian side of the border.
RG_10 has the largest terminus angle of 40° and sparse vegetation; therefore, this object could also potentially be intact.
The other 18 identified rock glaciers have terminus angles between 20° and 35°. The 13 rock glaciers are located more on the southern sides of the mountain peaks from which they originate, 7 are more to the north. Only 8 of them have maximum can be identified, only in two cases they do not have it.  The eight potential protalus ramparts are located at a mean average height of 1566 m a.s.l. (Table 2), which is higher than the previously mentioned potential rock glaciers. Due to the fact that they are mainly located above the timberline, they tend to 10 be covered with sparse shrubs (4 objects), although in one example there is no vegetation at all. Therefore, five potential protalus ramparts can be regarded as potentially intact forms, while three potential protalus ramparts are covered with dense coniferous forest, which means they can be regarded as relict forms. The activity status of the potential protalus ramparts should be defined in the future based on field measurements. All the protalus ramparts are located on the southern sides of mountain peaks, from which the talus feeding them originate. The ridge height on the terminus side is from 7 m to 15 m, 15 when it can be distinguished from the slope on which it is located. When the ridge cannot be distinguished from the on-going slope in Table 2 this is marked with a slope. The terminus slope angles are generally higher than those for potential rock glaciers with angles from 30° to 40°. Mainly in the vicinity of the protalus ramparts some kind of occasional water streams   Fig. 2a, Fig. S1). The rock glacier RG_1 is the most prominent rock glacier in this inventory and was the initiator of this research. Therefore, a field survey was conducted, where we found that the rock glacier is made of fine-grained material (Fig. 4). It has a very steep front angle of 35° and a height of 32 m. Next to its northern edge is RG_2, which is much smaller and not such a prominent feature, with a smaller front angle and height. The  with dense forest, with trees up to 12 m high. One, PR_4, at Kotova Špica/Monte Trmine (2351 m) has no vegetation coverage. The protalus rampart material PR_5 and PR_6 originates from massive Upper Triassic dolomite, and for PR_4, PR_7 and PR_8, from thick-bedded Dachstein limestone with transitions to dolomite.

Discussion
For their formation the rock glaciers require bedrock material that disintegrates into blocks and talus; therefore, the most 5 favourable types of rocks enabling rock glacier formation are granites and gneisses, and among the sedimentary rocks, limestones and dolomites (Žurawek, 2003;Haeberli et al., 2006;Johnson, 2007). In the Italian Alps the number of rock glaciers doubles in areas where the rock glaciers are fed by metamorphic rocks, when compared to the other areas with the same climatic conditions (Guglielmin and Smiraglia, 1998). The most frequent rock glaciers in the Niedere Tauern Range, Austria, are found in mica schist, but only due to the most abundant occurrence of this type of rock in the study area 10 (Kellerer-Pirklbauer, 2007). Additionally, Kellerer-Pirkbauer (2007) found a significant correlation between the lengths of the rock glaciers and the material from which they originate: gneiss produces relatively long rock glaciers, while mica schist produces short glaciers.
An analysis of the spatial distribution of periglacial features in Slovenia with regards to the bedrock composition reveals a high correlation between the rock glacier's appearance and the thick-bedded Dachstein limestones with transitions to 15 dolomite: 75 % (15) of all the discussed rock glaciers and 75 % (6) of the protalus ramparts are formed in this bedrock. Only 15 % (3) of the rock glaciers and 25% (2) of the protalus ramparts are found in Triassic dolomite. As limestone and dolomite are both known to be favourable types of rocks with respect to rock glacier formation the spatial distribution of Slovenian rock glaciers and protalus ramparts is primarily correlated to the glacier and periglacial extent in the Alpine Late Glacial, as the majority of Slovenian rock glaciers can be regarded as relict forms. The 65 % of identified Slovenian rock glacier slopes 20 are oriented more to the southern directions, which indicates that those objects are relict, as already shown by the example of the Tyrolean Alps in Austria by Krainer and Ribis (2012). Additionally, the rock glaciers are covered with at least sparse shrubs (30 %) and forest (60 %), which also supports the theory that they are relict (Cannone and Gerdol, 2003;Burga et al., 2004). Only two rock glaciers with steep front angles and without a lot of vegetation could potentially be intact and probably inactive (RG_10 and RG_11). Both are located at relatively high elevations in the Karavanks. In contrast, other identified 25 rock glaciers are found at relatively low mean elevations, from 1040 m to 1850 m a.s.l., which further supports the theory that they are relict periglacial features (Scotti et al., 2013).
The majority of the identified potential rock glaciers were found in the Karavanks (12 objects) that were not covered with extensive glaciers at the time of Alpine Late Glacial. Only traces of smaller cirque glaciers were identified (Ferk et al., 2015). Therefore, one potential rock glacier (RG_10) can be regarded as a debris rock glacier that formed from already-30 available cirque glacier moraine material. Others are probably talus rock glaciers that were formed in periglacial environmental conditions. On the other hand, the Julian Alps were covered with extensive ice fields and valley glaciers flowing down to the surrounding basins. Only the highest ridges and peaks were above the ice fields (Ferk et al., 2015); the periglacial environments were probably at lower elevations. Therefore, the mean elevation of the identified rock glaciers in the Julian Alps (1212 m a.s.l.) is approximately 300 m lower in comparison to those in the Karavanks (1557 m a.s.l.). The most prominent rock glacier (RG_1) is found at an elevation of 1040 m a.s.l. 5 Glaciers were also formed in the Dinaric Mountains during the Late Glacial, on the Trnovski Gozd and Snežnik mountain, where glacial accumulation features are preserved, with one potential debris rock glacier (RG_17) on Snežnik, which was formed from already-available glacier moraine material. Minor cirque glaciers were also formed on the Pohorje mountains in the most eastern and lowest part of the Slovenian Alps, which are formed from metamorphic rocks, but no rock glaciers could be identified there so far. 10

Conclusion
In this study we mapped 20 potential rock glaciers and 8 potential protalus ramparts in Slovenia based only on national aerial laser scanning data and knowledge about the bedrock composition gained from a general geological map. The existence of such objects in Slovenia was unknown prior to the availability of this national aerial laser scanning data, due to the fact that the majority of them are hidden under dense vegetation, which hinders their identification on an orthophoto. 15 Based on the vegetation coverage, slope aspect direction and mean elevation we concluded that almost all identified rock glaciers can be regarded as relicts, except for two, which may potentially be intact. A half of the protalus ramparts might be relict due to the heavy vegetation coverage (dense coniferous forest), while a half are possibly active. The rock glacier slope angles are oriented 65 % in a southern direction and 35 % in a northern direction. All the identified protalus ramparts are oriented in a southern direction. The terminus slope angles of the potential rock glaciers are from 20° to 35° and from 30° to 20 40° for the potential protalus ramparts. They are located at mean elevations from 1040 m to 1850 m a.s.l. The identified rock glaciers rarely exceed 600 m in length.
The spatial distribution of the identified objects is in close correlation to the areas outside the Alpine Late Glacial maximum glacier extent, and more abundant in areas that were at that time influenced by periglacial environmental conditions.
Regarding the lithology, they are extensively correlated to thick-bedded Dachstein limestones with transitions to dolomite. 25 This study fulfilled its expectations in proving that the national aerial laser scanning data with a DTM grid size of 1 m × 1 m enables the identification of potential permafrost features hidden under dense vegetation based only on the morphological characteristics of these objects. This inventory of potential rock glaciers and protalus ramparts in Slovenia provides an ideal starting point for detailed studies of individual objects. The described methodology could enable similar studies of relict permafrost features in other mountainous areas where national aerial laser scanning has become available. 30 The Cryosphere Discuss., doi:10.5194/tc-2016-86, 2016 Manuscript under review for journal The Cryosphere Published: 13 June 2016 c Author(s) 2016. CC-BY 3.0 License.