Towards Better Visualisation of Alpine Quaternary Landform Features on High-Resolution Digital Elevation Models

Alpine topography is formed by a complex series of geomorphological processes that result in a vast number of different landforms. The youngest and most diverse landforms are various Quaternary sedimentary bodies, each characterised by its unique landform features. The formation of Quaternary sedimentary bodies and their features derive from the dominant building sedimentary processes. In recent years, studies of Quaternary sedimentary bodies and processes have been greatly aided by the use of digital elevation models (DEMs) derived by airborne laser scanning (ALS). High-resolution DEMs allow detailed mapping of sedimentary bodies, detection of surface changes, and recognition of the building sedimentary processes. DEMs are often displayed as hillshaded reliefs, the most common visualisation technique, which suffers from the limitation of a single illumination source. As a result, features can be barely visible or even invisible to the viewer if they are parallel to the light source or hidden in the shadow. These limitations become challenging when representing landforms and subtle landscape features in a diverse alpine topography. In this study, we focus on eleven visualisations of Quaternary sedimentary bodies and their sedimentary and morphological features on a 0.5 m resolution DEM. We qualitatively compare analytical hillshading with a set of visualisation techniques contained in the Raster Visualisation Toolbox software, primarily hillshading from multiple directions RGB, 8-bit sky view factor and 8-bit slope. The aim is to determine which visualisation technique is best suited for visual recognition of sedimentary bodies and sedimentation processes in complex alpine landscapes. Detailed visual examination of previously documented Pleistocene moraine and lacustrine deposits, Holocene alluvial fans, scree deposits, debris flow and fluvial deposits on the created visualisations revealed several small-scale morphological and sedimentary features that were previously difficult or impossible to detect on analytical hillshading and aerial photographs. Hillshading from multiple directions resulted in a visualisation that could be universally applied across the mountainous and hilly terrains. In contrast, 8-bit sky view factor and 8-bit slope visualisations created better visibility and facilitated interpretation of subtle and small-scale (less than ten metres) sedimentary and morphological features.


Introduction
The large-scale alpine topography is characterised by sharp peaks and deep valleys dominated by steep rock faces. This heterogeneous landscape consists of diverse landforms formed by the combined effect of tectonics, bedrock geology and Quaternary sedimentary processes. In particular, the large-scale alpine topography is characterised by sharp peaks and deep valleys dominated by steep rock faces. This heterogeneous landscape consists of diverse landforms formed by the combined effect of tectonics, bedrock geology and Quaternary sedimentary processes. In particular, the latter form sedimentary bodies with a distinctive shape, architecture and spatial distribution related to past or recent solely relies on inspection of an ALS derived hillshaded DEM. Each type of sedimentary deposit in the valley has a distinct morphology derived from its predominant sedimentary transport and depositional mechanism. The oldest sedimentary bodies are glacial moraines and lacustrine deposits which are of Pleistocene age and cover the valley floor. After the retreat of the Pleistocene glacier, the valley began to fill with a variety of Holocene sediments [47].

Characteristics of Quaternary Sedimentary Bodies
Following the most recent geomorphological study by [47], Quaternary sedimentary bodies can be grouped into six landform types, each with typical morphological and sedimentary features.

Pleistocene Moraines
The valley floor is covered by a large moraine (approximately 2.5 km long and up to 500 m wide) that forms a distinct, undulating topographic ridge 30 m high (Figure 2A). It is composed of poorly sorted sediment whose grain size varies from silt to several tens of cubic metres of boulders. Studies by [48,49] also identified Pleistocene moraines located at the valley entrance and on the high slopes above the valley floor. (F) A sedimentary deposit (note red A4 sized map to the left corner) with a similar sedimentary texture to sieve deposits (approx. three meters high) defined as "mega sieve deposits" in this study.

Data Acquisition and Visualisation Creation
The ALS classified point cloud data were obtained from the openly available Airborne laser scanning dataset of Slovenia [53]. Airborne laser scanning was performed between July 2014 and January 2015 using the LMS-Q780 scanner with an average scanning density of 5 points/m 2 . The classified point cloud was filtered in ESRI ArcGIS to manmade objects and bare ground, without of vegetation (point classes 0, 2 and 6) using ESRI ArcGIS raster conversion tools [55,56]. The generated DEM has a resolution of 0.5 m and (F) A sedimentary deposit (note red A4 sized map to the left corner) with a similar sedimentary texture to sieve deposits (approx. three meters high) defined as "mega sieve deposits" in this study.

Lacustrine Deposits
Lacustrine deposits outcrop only at one location in the valley and represent a topographic depression with a flat floor. The deposits consist of clay, silt, and very fine sand particles.

Fluvial Deposits
Fluvial deposits consist of sand and gravel transported by perennial water streams. Morphologically, fluvial deposits form plains with incised torrential channels and thin splay (out-of-channel) deposits of sand and gravel. The fluvial deposits are partially anthropogenically reworked due to the use of the land as pasture.

Debris Flow Lobe
In recent years, a debris flow occurred on the southwestern slopes of the Ciprnik mountain in November 2000 [47,50]. It partially covered an older alluvial fan and formed a lobe up to five metres high ( Figure 2B), composed of clasts of clay to boulders several metres in size.

Talus Slopes
Talus slopes consist of very angular gravels accumulated beneath steep talus slopes and can be categorised into active and inactive talus slopes. Inactive talus slopes are generally located at lower elevations, are covered by vegetation, and have less active sediment deposits. Active talus slopes are located at higher elevations where sediment is actively deposited as rock-fall or as grain-flows via gullies ( Figure 2C). Their surface inclination is larger than 30 • .

Alluvial Fans
Alluvial fans are the most numerous sedimentary bodies with a surface slope of 5 to 25 • . Most of them are categorised as Type I and II following the work of [54]. There are no permanent water streams in the valley, so sediment deposition occurs only during high rainfall events. Sediment (particles ranging from sand to very coarse cobble) is deposited either in torrential channels cut into the fans or outside the channels ( Figure 2D). Sediment deposited outside the channel forms sheetflood deposits and also sieve deposits of various sizes ( Figure 2C,D).

Data Acquisition and Visualisation Creation
The ALS classified point cloud data were obtained from the openly available Airborne laser scanning dataset of Slovenia [53]. Airborne laser scanning was performed between July 2014 and January 2015 using the LMS-Q780 scanner with an average scanning density of 5 points/m 2 . The classified point cloud was filtered in ESRI ArcGIS to man-made objects and bare ground, without of vegetation (point classes 0, 2 and 6) using ESRI ArcGIS raster conversion tools [55,56]. The generated DEM has a resolution of 0.5 m and covers a total area of 42 km 2 , i.e., the area of the valley and surrounding mountain ridges. We choose this resolution to better detect small-scale features such as boulders, sieve deposits and shallow channels.
Visualisations of the DEM were performed using the Raster Visualisation Toolbox (RVT), an open-source software developed by ZRC SAZU and the University of Ljubljana [27,57]. RVT provides eleven different types of visualisations, which can be generated from the input raster file. Each visualisation can be modified by a wide range of setting parameters specific to each visualisation technique [27,30]. The parameters and algorithms for each visualisation used in this study were set to the program's default settings (Table 1) and are described in detail in [27,30,37,57]. Default settings were used since it offers best feature recognition based on previous studies [35]. A comparison of the produced visualisations was performed in QGIS 3.20 [58] where landforms and features were measured, and their terrain profiles were created using QGIS Profile Tool plugin [59].

Visualisation Analysis and Identification of Landscape Features
The visualisations contained in the RVT toolbox were developed for use in a large variety of terrains: from very low-gradient flat areas to rugged reliefs. For this reason, some visualisations are more suited to certain terrain types and perform less optimally in other settings [37,57]. In mountainous terrain slope inclination varies greatly resulting in a complex topography with generally low-to moderately-sloping sedimentary bodies and high-gradient relief of rock outcrops. We firstly performed a general visual interpretation of the eleven visualisations. The initial requirement for each visualisation was to provide a clear, unambiguous and intuitive distinction of the main features: valley floor, slopes, summits, ridgelines, topographic depressions, and break lines. If the produced visualisation provided a poor and unintuitive distinction of basic mountainous features, the visualisation was not used for further detailed analysis. In the second stage, we selected those visualisations that were most suitable for detailed analysis of the studied area following the criteria of [27] (e.g., clear visibility of small-scale features, intuitive visualisation, no artificially produced artifacts, etc.) and compared them with analytical hillshading. Subsequent more detailed analyses were based on qualitative visual analysis of Quaternary sedimentary bodies documented in previous studies and conducted field work. We selected specific sedimentary bodies where geomorphological and sedimentary features specific to the type of sedimentary body are evident. We validated whether these landform features are also visible on derived visualisations and if so, we evaluated which visualisation provided better feature recognition based on a qualitative visualisation assessment. To accomplish this, we compiled a list of distinct morphological and sedimentary features we aimed to identify on each type of sedimentary body. The locations and figures of the investigated Quaternary sedimentary bodies are shown in Figure 1, and their distinctive surface features that we were searching for are listed in Table 2.

Comparison of All Eleven Visualisation Types
Based on the comparison of the eleven produced visualisations, we selected hillshading from multiple directions (RGB, number of directions (D) 16

Alluvial Fans
Several sedimentary and geomorphologic features that occur on alluvial fans are readily visible on the analytical hillshade visualisation, but their recognition is vastly improved on other visualisation. Visual recognition of torrential channels is much more evident on the 8-bit slope and hillshading from multiple directions RGB visualisations ( Figure 4A-C). Relatively shallow channels cut into fans with a maximum depth of one metre stand out clearly from the generally flat surfaces ( Figure 4B). In addition, distributary channels in catchments of fans are better visible on the 8-bit slope visualisation ( Figure 4C). Visibility of features inside channels, such as braided river channel systems and channel bars, is also improved on hillshading from multiple directions RGB visualisation than on slope or analytical hillshade ( Figure 4A-C).
features on low-slope sedimentary bodies are less pronounced ( Figure 3F). Local dominance, simple local relief model, sky illumination model, uncompressed slope visualisation, positive and negative openness provide clear recognition between prominent break line in surface inclinations such as rock walls and valley floor ( Figure 3I,J). However, these visualisations did not provide sufficient visibility of features on moderate to low-gradient Quaternary sedimentary bodies, where small-scale features are very poorly or not at all visible and challenging to delineate despite the high resolution of DEM.    Deposits outside the channel are virtually invisible on the analytical hillshade visualisation ( Figure 4D). However, due to their subtly undulating morphology, they are more noticeable on 8-bit slope and hillshading from multiple directions RGB visualisations ( Figure 4E,F). Their general spatial extent can be inferred from the surface morphology.
Larger boulders (greater than ten metres) were detected on Type I alluvial fan c.f. [54] along with distributary channels and grain flow deposits ( Figure 4G-I). Boulders are clearly visible on 8-bit slope visualisation, while areas of grain flow deposits appear as a smooth surface with slight undulation ( Figure 4I).
Small-scale sieve deposits (5 m long, 2 m wide, and 0.5 m high, an example is shown in Figure 2F), which are relatively common at the study site, were not recognised on any of the visualisations derived from DEM. However, larger sieve deposits (5 by 10 m, Figure 5), which are less common in the study site, were recognised. These deposits are much less apparent in the analytical hillshading ( Figure 5A), but they are much more apparent in hillshading from multiple directions and in the 8-bit slope visualisations. They can be seen in 8-bit the sky view factor visualisation, although the edges do not appear as sharp as in other visualisation types ( Figure 5C,D). Features similar to sieve deposits in terms of sedimentary texture and morphology were discovered during field work ( Figure 2G

Glacial and Lacustrine Deposits
Lacustrine deposits are not particularly prominent on any visualisation type. On all types, they are seen as a flat surface with no prominent morphological features ( Figure 6). The features visible as undulating terrain on 8-bit sky view factor and 8-bit slope visualisation are out-of-channel deposits from a nearby active alluvial fan that is gradually covering the lacustrine deposits [47]. The largest moraine covering the valley floor can be easily outlined from other sedimentary bodies on all four visualisation techniques due to its undulating terrain morphology, large boulders, and ridge-like topography ( Figure 6). However, when comparing the three visualisation techniques, the terrain morphology is less pronounced in the analytical hillshading visualisation. In comparison, the 8-bit slope and 8-bit sky view factor visualisations significantly improve the visibility of undulating terrain and boulders larger than 5 m (Figure 6B,C).

Glacial and Lacustrine Deposits
Lacustrine deposits are not particularly prominent on any visualisation type. On all types, they are seen as a flat surface with no prominent morphological features ( Figure 6). The features visible as undulating terrain on 8-bit sky view factor and 8-bit slope visualisation are out-of-channel deposits from a nearby active alluvial fan that is gradually covering the lacustrine deposits [47]. The largest moraine covering the valley floor can be easily outlined from other sedimentary bodies on all four visualisation techniques due to its undulating terrain morphology, large boulders, and ridge-like topography ( Figure 6).
However, when comparing the three visualisation techniques, the terrain morphology is less pronounced in the analytical hillshading visualisation. In comparison, the 8-bit slope and 8-bit sky view factor visualisations significantly improve the visibility of undulating terrain and boulders larger than 5 m ( Figure 6B,C). A previous study by [48] describes boulders and ridge-shape morphology of the terminal moraines at the valley entrance oriented in NW-SE direction (Figure 7). The terminal moraines are less pronounced in analytical hillshading but clearly visible in other visualisations, especially in the hillshading from multiple directions visualisation (Figure 7  B). The moraine in the Zadnja Ponca cirque (Figure 8), described by [49], represents a ridge approximately 220 m long, up to 5 m high and up to 40 m wide. This moraine is more visible in hillshading from multiple directions rather than on analytical hillshading. However, no large boulders are visible on any visualisation type. Under Mount Mojstrovka (Figure 9), we recognised a moraine more than 600 m long and up to 200 m wide. Its undulating morphology derives from huge glacial boulders with a maximum size of 30 m. Similarly, to the largest moraine covering the valley floor, the visibility of the undulating topography of the moraine is greatly enhanced by hillshading from multiple directions, 8-bit sky view factor and 8-bit slope visualisation. A previous study by [48] describes boulders and ridge-shape morphology of the terminal moraines at the valley entrance oriented in NW-SE direction (Figure 7). The terminal moraines are less pronounced in analytical hillshading but clearly visible in other visualisations, especially in the hillshading from multiple directions visualisation (Figure 7 B). The moraine in the Zadnja Ponca cirque (Figure 8), described by [49], represents a ridge approximately 220 m long, up to 5 m high and up to 40 m wide. This moraine is more visible in hillshading from multiple directions rather than on analytical hillshading. However, no large boulders are visible on any visualisation type. Under Mount Mojstrovka (Figure 9), we recognised a moraine more than 600 m long and up to 200 m wide. Its undulating morphology derives from huge glacial boulders with a maximum size of 30 m. Similarly, to the largest moraine covering the valley floor, the visibility of the undulating topography of the moraine is greatly enhanced by hillshading from multiple directions, 8-bit sky view factor and 8-bit slope visualisation.

Scree Deposits
The most distinctive sedimentary and morphological features on scree deposits (Sc) are grain flow lobes ( Figure 8) and gullies cut into the deposited sediment ( Figure 9). They can be seen on scree slopes in the Zadnja Ponca and on the scree slope below Mojstrovka Mountain (Figure 9). In both cases, these features are poorly visible on aerial photographs because of sunlight direction ( Figure 8A). The sediment consists of light grey limestone and dolomite clasts that appear white in direct sunlight, completely obscuring the geomorpho-logical features. However, on the 8-bit sky view factor, hillshading from multiple directions RGB, and the 8-bit slope visualisation, the grain flow deposits, and fluvial gullies are spectacularly pronounced, not only compared to aerial photographs but also compared to the analytical hillshade visualisation ( Figure 8A-E). Grain flows on scree deposits represent morphologically very subtle sedimentary deposits, rarely exceeding one metre in height ( Figure 8F). They, therefore, do not stand out from the local relief topography. Nevertheless, their shape and microphotography are distinctly pronounced on hillshading from multiple directions visualisation ( Figure 8C,F), and less so in other visualisations.

Scree Deposits
The most distinctive sedimentary and morphological features on scree deposits are grain flow lobes ( Figure 8) and gullies cut into the deposited sediment ( Figure 9). T morphological features. However, on the 8-bit sky view factor, hillshading from multiple directions RGB, and the 8-bit slope visualisation, the grain flow deposits, and fluvial gullies are spectacularly pronounced, not only compared to aerial photographs but also compared to the analytical hillshade visualisation (Figure 8A-E). Grain flows on scree deposits represent morphologically very subtle sedimentary deposits, rarely exceeding one metre in height ( Figure 8F). They, therefore, do not stand out from the local relief topography. Nevertheless, their shape and microphotography are distinctly pronounced on hillshading from multiple directions visualisation ( Figure 8C,F), and less so in other visualisations.

Debris Flow Deposits
The debris flow deposited under the slopes of Mountain Ciprnik is relatively well visible in aerial photographs, especially the surface of the rapture, which has not (yet) been overgrown with vegetation ( Figure 10A). However, the deposited debris flow lobe (DfL) itself is difficult to outline from the surroundings because of the vegetation cover and the colouring of the sediment. However, the lobe is visible on the analytical hillshade visualisation ( Figure 10B), but its visibility is substantially improved on the 8-bit sky view

Debris Flow Deposits
The debris flow deposited under the slopes of Mountain Ciprnik is relatively well visible in aerial photographs, especially the surface of the rapture, which has not (yet) been overgrown with vegetation ( Figure 10A). However, the deposited debris flow lobe (DfL) itself is difficult to outline from the surroundings because of the vegetation cover and the colouring of the sediment. However, the lobe is visible on the analytical hillshade visualisation ( Figure 10B), but its visibility is substantially improved on the 8-bit sky view factor ( Figure 10C) and 8-bit slope visualisation ( Figure 10D). In all three visualisation types, the morphology of the rupture surface is well pronounced with the recognisable V-shaped gullies. Despite the high resolution of DEM, the two-metre sized boulders transported by debris flow are not visible on any of visualisation. In the produced DEM, a two-metre sized object is represented by four pixels, which seems to be too small for visual recognition.
factor ( Figure 10C) and 8-bit slope visualisation ( Figure 10D). In all three visualisation types, the morphology of the rupture surface is well pronounced with the recognisable Vshaped gullies. Despite the high resolution of DEM, the two-metre sized boulders transported by debris flow are not visible on any of visualisation. In the produced DEM, a twometre sized object is represented by four pixels, which seems to be too small for visual recognition.

Fluvial deposits
Fluvial deposits are present in areas with a relatively flat surface. A common morphological feature present on these deposits are torrential channels, which have a width of a few tens of metres, a depth of up to one metre and have a relatively short runout. Although the channels are visible on analytical hillshading, they are not strongly pronounced due to the orientation of the incoming light. They are much better pronounced on the 8-bit sky view factor, hillshading from multiple directions RGB, and 8-bit slope visualisations ( Figure 11). In addition, features such as channel bars can also be also recognised, most notably in hillshading from multiple directions RGB and 8-bit sky view factor ( Figure 11B,C). Sediments deposited by ephemeral torrents (marked with circles in Figure 11) are placed out of channels on the flat surface, similarly to out-of-channel deposits on alluvial fans. However, due to their low gradient and lack of distinct

Fluvial Deposits
Fluvial deposits are present in areas with a relatively flat surface. A common morphological feature present on these deposits are torrential channels, which have a width of a few tens of metres, a depth of up to one metre and have a relatively short runout. Although the channels are visible on analytical hillshading, they are not strongly pronounced due to the orientation of the incoming light. They are much better pronounced on the 8-bit sky view factor, hillshading from multiple directions RGB, and 8-bit slope visualisations ( Figure 11). In addition, features such as channel bars can also be also recognised, most notably in hillshading from multiple directions RGB and 8-bit sky view factor ( Figure 11B,C). Sediments deposited by ephemeral torrents (marked with circles in Figure 11) are placed out of channels on the flat surface, similarly to out-of-channel deposits on alluvial fans. However, due to their low gradient and lack of distinct morphological features, these deposits are barely visible on both analytical hillshading and hillshading from multiple directions. To some extent, they are visible on 8-bit sky view factor and 8-bit slope visualisations, where their spatial extent can be tentatively outlined ( Figure 4D-F).
Remote Sens. 2021, 13, x FOR PEER REVIEW 17 of 23 morphological features, these deposits are barely visible on both analytical hillshading and hillshading from multiple directions. To some extent, they are visible on 8-bit sky view factor and 8-bit slope visualisations, where their spatial extent can be tentatively outlined ( Figure 4D-F).

Discussion
The availability of high-resolution ALS data and presence of several typical alpine geomorphological features assembled in the small area of the Planica valley provide an excellent testing area for comparing different DEM visualisations. Qualitative visual analysis of several visualisations revealed their advantages and disadvantages when applied to a complex alpine landscape. Of the eleven visualisations produced, seven were not suitable for detailed landform analysis in alpine landscapes, offering poorer recognition of basic mountainous landscape features on a small scale compared to analytical hillshading ( Figure 3, Table 3). This is because each visualisation is computed using different parameters [30,37]. As a result, some visualisations are better tailored for low-gradient terrains, while others are better at pronouncing high-gradient features. Topographic break lines between steep mountain slopes and relatively low and moderate surface gradients of Quaternary sedimentary bodies can be identified in the local dominance, simple local relief model, slope visualisation, sky illumination model, positive and negative openness visualisations although not always intuitively ( Figure 3, Table 3). However, despite the high resolution of DEM, these visualisations were all ineffective for an in-depth examination of small-scale features located on sedimentary bodies with low to moderate surface slopes. Features appeared either too light or too dark (Figure 3), making detailed analysis impossible. The principal component analysis resulted in visualisations in which landforms and small-scale features can be identified, but relief recognition is not always intuitive or easy to interpret. Viewers unfamiliar with the landscape type or study area could easily misinterpret or misidentify landforms. Visualisation techniques not used for analysis of low-to mid-gradient Quaternary landforms could potentially be used for analysis of steeper-gradient surfaces or other non-alpine landscapes (Table 3).
However, a qualitative visual comparison of the 8-bit slope visualisation, hillshading from multiple directions RGB (D16 in H35), and the 8-bit sky view factor leads to identifying the vast majority of sedimentary and morphological features not seen in either aerial photographs or analytical hillshading. The visualisations provide a surface analysis beyond simple delineation of individual sedimentary bodies and enable the recognition of smallscale sedimentary and morphological features characteristic of the sedimentary body type. Large, raised landscape features such as debris flow lobe, grain flow deposits, larger boulders on fans and moraines, are much more pronounced and noticeable than their representation in analytical hillshading. Similarly, all shallowly incised features such as channels and gullies appear much clearer. In addition, features inside channels such as channel bars, which are not evident in analytical hillshading, are visible in the 8-bit slope visualisation, in hillshading from multiple directions RGB (D16 in H35), and in the 8-bit sky view factor visualisation. However, despite the high resolution of the generated DEM, some subtle and small-scale sedimentary features could not be detected. Subtle elements, such as out-of-channel deposits, are not discernible and are challenging to outline, but they can be at least roughly delineated due to their weak surface undulation. Small-scale sieve deposits with an approximate extent of two metres in width, ten in length, and a maximum of half a metre could not be detected on any visualisation created. However, at least twice that size larger sieve deposits were identified. No boulders were recognised on debris flow deposits, scree deposits, and on some of the moraines. It appears that the limit for detecting boulders on the produced DEM is five metres in diameter. In comparison to grain flow deposits, which are clearly recognisable (e.g., Figure 8F), boulders have a higher elevation difference. However, boulders smaller than five metres were not identified. This might be due to the boulder's shape, which is a spherical feature, while grain flows are longitudinal features. This comparison indicates that longitudinal objects, despite being smaller, might be more successfully visually recognised.
The use and availability of high-resolution DEMs derived from ALS point clouds in topography analysis are increasing [15,37], however the ALS data is not available everywhere nor in such a high resolution. The use of presented visualisation techniques is not limited to high resolution ALS but could be used in other available DEM sources. As shown in previous studies, there is no general visualisation that could be provided as a universal surface representation across all landscape types [60]. The mountain landscape has a complex and diverse topography consisting of steep rock faces and relatively flat valley floors covered by intertwined landform features. Such topography is challenging to represent so that all landform features are equally and appropriately identifiable in a single visualisation. In our study, the best option for representing all surface aspects of Quaternary landforms was provided with hillshading from multiple directions RGB, 8-bit sky view factor, and 8-bit slope visualisation. We recommend using a combination of these visualisations as each one has different benefits and can subjected to user's personal preferences. Hillshading from multiple directions, however, proved to be a visualisation that can be universally used in mountainous or hilly terrain. The visibility of small-and large-scale features is vastly improved compared to analytical hillshading. All investigated features, from relatively small scales, such as shallow channels, sieve deposits and boulders, to very large-scale features, are easily recognised and have an excellent visual appearance. However, some features such as boulders and channels, might be better noticeable on 8-bit sky view factor and 8-bit slope visualisations than in hillshading from multiple directions RGB. Sky view factor and slope visualisations proved highly effective in detail recognition of small-scale features on a relatively flat surface with low surface inclinations, such as alluvial fans and fluvial deposits. There, topographically less expressed features, such as shallow torrential channels, small scale sieve deposits, and out-of-channel deposits, appear much more prominent. However, there are differences in applicability between the two visualisations. Slope visualisation produces an image where edges of features appear sharper than in the 8-bit sky view factor ( Figure 5C,D and Figure 11C,D). For object delineation, the 8-bit slope visualisation appears more applicable.
By comparing the produced DEM visualisations to field mapping [47] we estimate that majority of sedimentary bodies and features can be successfully identified. The identifications and mapping of subtle and small-scale features using DEM is even more exact than field mapping due to inaccessibility of certain areas and/or vegetation cover during mapping campaign. This indicates that different visualisation techniques can be used for re-evaluating the existing geomorphological maps. The remaining unrecognized areas are challenging to identify and need to be either field mapped or investigated using other remote sensing techniques (e.g., using terrestrial laser scanning, structure-frommotion technique, structured light scanning, etc.).
Both field mapping and visual analysis of DEM can be very time consuming. Recent studies in advanced machine learning indicate a great potential for automated recognition of both manmade and natural landforms [41,[61][62][63]. Automated recognition can vastly increase the mapped area size and shortens the mapping time. To achieve successful recognition, the input reference data is needed for further pattern recognition. This study shows which visualisations provides best input data for alpine Quaternary landform visual recognition and could serve as a benchmark for future geomorphological process-studies using advanced machine learning.

Conclusions
Landform features typical of alpine environments may be poorly or not visible on analytical hillshading or in aerial photographs due to inadequate illumination. This can lead to misinterpretation of landforms or overlooking of important features. We have compared a set of eleven different visualisations of a high-resolution DEM covering a relatively large research area. The principal component analysis, local dominance, simple local relief model, slope visualisation, positive and negative openness visualisations allowed us to distinguish steep mountain faces and ridges from the valley floor. However, these visualisations do not lend themselves to a more detailed analysis of landforms with low to moderate surface gradient. On the other hand, hillshading from multiple directions, 8-bit sky view factor, and 8-bit slop visualisations allowed better interpretation of multiple low-gradient, subtle, small-and large-scale morphological and sedimentary features compared to analytical hillshading or aerial photographs. Our results show that hillshading from multiple directions can be used as a general-purpose visualisation for identifying and mapping Quaternary sedimentary bodies. It provides unambiguous recognition of landforms without regard to their size, surface inclination and regardless of light orientation compared to analytical hillshading. In addition, the 8-bit sky view factor and 8-bit slop visualisation provide more unambiguous recognition of small-scale and subtle sedimentary and morphological features, such as boulders, channels, and channel bars, sieve, and grain flow deposits, on a high-resolution DEM. We recommend interlinked use of these three visualisation techniques in future mapping and analyses of mountainous and hilly terrain. Funding: The Slovenian Research Agency financially supported this work within the Young Researcher grant program (nr. 51975), within the research programs P1-0195 ("Geoenvironment and Geomaterials") and P2-0406 ("Earth Observation and Geoinformatics"), and research project J2-9251 ("M3Sat").
Institutional Review Board Statement: Not applicable.