Surface morphological types and spatial distribution of fan-shaped landforms in the periglacial high-Arctic environment of central Spitsbergen, Svalbard

ABSTRACT A 1:19,500 map of the Petunibukta region documents the spatial distribution of almost 300 colluvial and alluvial fans (together with their catchments) in the periglacial high-Arctic environment of Svalbard. Fan-shaped landforms were mapped using an orthophoto and digital elevation model generated from 2009 aerial photographs and a 2013 high-resolution satellite image using Geographic Information Systems. Four additional maps at a scale of 1:40,000 provide details about the morphometric characteristics of the studied fans: long- and cross-profiles, slope and aspect. Additionally, parameters such as fan and catchment area, relief, length, width, elevation (the lowest point at the fan toe, apex, and the highest point of the catchment boundary), fan mean plan and profile curvatures, fan relation to neighbouring fans, are also presented. Fans were classified according to the dominant processes shaping their surface: colluvial fans (n = 229), alluvial fans dominated by debris flows (n = 49), and alluvial fans dominated by fluvial flows (n = 19).

To achieve the aforesaid objective, we created a map at a scale of 1:19,500 which includes the spatial distribution of alluvial and colluvial fans, and four other maps at a scale of 1:40,000, presenting the spatial pattern of morphometric characteristics of the studied fans: long-and cross-profile, slope and aspect. Additionally, based on the established boundary of fans, it was possible to calculate parameters such as: fan and catchment area, relief, length, width, elevation (the lowest point at the fan toe, apex, and the highest point of the catchment boundary), fan mean plan and profile curvatures, and fan relation to neighbouring fans. The presented data provide a context for further geomorphological examination of the evolution of fans in the Arctic, which will be published elsewhere.

Study setting
The study was carried out on Spitsbergen Island, which is part of the Svalbard archipelago, situated in the high-Arctic ( Figure 1). We focused on the Petuniabukta region in the central part of the island. This area is representative of central Spitsbergen as it is characterized by relatively small (compared to the coastal areas) ice cover. In addition, data availability is good, including high-resolution satellite imagery, time-series of aerial photographs, and several years of field-based observations. In terms of geology (Figure 2), the study area is located within the Billefjorden fault zone (with a north-south orientation), which accounts for the abundance of rock types being exposed at the surface, including: Precambrian (various crystalline), Devonian (mainly sandstones and mudstones), Carboniferous-Permian (conglomerates, sandstones, mudstones, limestones, coal, gypsum, anhydrites and dolomites), and Quaternary (alluvium, colluvium and landslide debris, as well as glacial, fluvioglacial, marine, intertidal and beach deposits) (Harland et al., 1974;Lamar & Douglass, 1995).
The contemporary geomorphology of the Petuniabukta area is dominated by mountain ridges that reach up to 1024 m a.s.l. (summit of De Geerfjellet) and several large U-shaped post-glacial valleys. Along the mountain crests, plenty of colluvial and alluvial fans have developed as a result of paraglacial adjustment of the topography (Rachlewicz, 2010). Fluvial transport in both, glacierized and non-glacierized catchments, is responsible for delivering large quantities of sediment into the fjord (e.g. Szpikowski, Szpikowska, Zwoliński, Rachlewicz, et al., 2014). The morphological effects of large sediment loads reaching the fjord can be observed, for example, in the dynamic development of coastal landforms (Strzelecki, Long, & Lloyd, 2015;Strzelecki, Małecki, & Zagórski, 2015). There is a continuous permafrost on Svalbard, with a thickness of 100 m in the valley bottoms and near the coast to as much as 400-500 m in inland mountains (cf. Etzelmüller & Hagen, 2005;Humlum, Instanes, & Sollid, 2003). The active layer's thickness near Petuniabukta is up to 1.2 m deep next to the seashore, and from 0.5 to 2.5 deep inland (Gibas et al., 2005;Rachlewicz & Szczuciński, 2008), which limits the amount of debris that is potentially possible to move via mass movement processes on the fans' surface. The terrain exposed by the retreat of glaciers since the end of the LIA was rapidly colonized by plants, as it took only ∼100 years to reach an almost mature stage of succession (cf. Prach & Rachlewicz, 2012). The latest succession stages have assembled a species composition of tundra which had existed in front of moraines during the time of the LIA. The first species appeared about 5 years after deglaciation. The Saxifraga oppositifolia was the most common first colonizer. The number of species increased steadily during the first ∼50 years, after which growth stabilized (Prach & Rachlewicz, 2012).
The vegetation in the vicinity of Petuniabukta is relatively rich in species. Prach, Klimešová, Košnar, Redčenko, and Hais (2012) distinguished seven vegetation units: dryas octopetala community, saxifraga oppositifolia community, vegetation dominated by mosses, deschampsia community, papaver dahlianum community, species rich communities under bird nesting sites and sea shore vegetation. However, the vegetation cover around Petuniabukta is very sparse or even missing on steep slopes, exposed summits and crests, near streams, and in these parts of glacial forelands which were not yet colonized by plants.
The climate in central Spitsbergen is characterized by relatively high annual air temperature and the low amount of precipitation. The yearly mean precipitation in Petuniabukta reaches about 150-200 mm Rachlewicz, 2003). Permanent snow cover lasts from the beginning of October to the first half of June (Láska, Witoszová, & Prošek, 2012). However, our 10 years of field-based observations indicated that snow patches can survive until late summer in sheltered depressions. The mean average air temperature in 2009 reached −4.5°C, with the monthly mean temperature ranged from −17.0°C (April) to 7.2°C (July) (Láska et al., 2012). The melting season (characterized by a temperature above 0°C) in Petuniabukta lasts usually from mid-June until the beginning of September (Szpikowski, Szpikowska, Zwoliński, Rachlewicz, et al., 2014). The highest temperatures are recorded in July and at the beginning of August (Láska et al., 2012;Rachlewicz, 2003;Rachlewicz & Styszyńska, 2007) and in this period, the highest melting rates of glacier and permafrost should be expected.

Datasets and data processing
Analysis of the fans' surface morphology was carried out based on two main sets of remote sensing data and field-based ground verification: (1) Colour-digital aerial photographs taken in 2009 (mean flight altitude 7350 m) were purchased from the Norsk Polar Institute and photogrammetrically processed with the use of ground control points (surveyed with differential global positioning system receiver in 2013 and 2014 and post-processed). An orthophotomap with a cell size 0.4 m and a digital elevation model (DEM) with a cell size of 5 m were produced in the UTM 33N coordinate system. The DEM was used to calculate additional data such as: (1) slope, (2) aspect, and (3) curvature.
(2) A high-resolution satellite images taken on 17 July 2013 (images catalog ID: 1030010025985E00 and 10300100238C8700) by the WorldView-2 satellite were purchased from Digital Globe in the form of a panchromatic band (cell size 0.5 m) and four multispectral bands (cell size 2.0 m). The satellite image was orthorectified using a DEM and pansharpened to generate a 4-band multispectral image with a 0.5 m cell size.

Mapping of fans and catchments
The geomorphological features of the surface morphology of fans were identified based on analysis of the orthophotos combined with visualizations of the DEM (shaded relief with different azimuth angles of the light source) and maps of slope gradient, aspect and curvature. Additional field geomorphological mapping and ground checking was carried out during fieldwork in 2013, 2014 and 2016. Further guidance was also provided by a 1:40,000 geomorphological map of the Petuniabukta area (Karczewski et al., 1990). Fans and their catchment areas were delineated visually in Esri ArcMap 10.3 from the orthophotos and DEM. Delineation was guided by the watershed, flow direction and flow accumulation rasters, which were calculated using the rho8 algorithm in Withebox Gat, following the approach proposed by Sattler (2014). The main morphometric and morphological features of the fans and their catchments have been collected, as in previous research (cf. de Haas et al., 2015;de Scally & Owens, 2004de Scally, Owens, & Louis, 2010;de Scally, Slaymaker, & Owens, 2001;Giles, 2010;Saito & Oguchi, 2005) the catchment properties had been indicated as affecting the form of the fans. Morphometric analyses were performed in Arc-Map 10.3, and the morphometric variables (definitions, type, and range of values are provided in Table 1) were extracted from DEM within the geographical information system (GIS), based on the spatial location of landforms.

Map production and design
A map depicting the surface morphological types of alluvial and colluvial fans was produced at a scale of 1:19,500, additionally accompanied by four further maps at a scale of 1:40,000. We implemented uniform map layout to create the maps for all fan-related features. Firstly, each map contains a system of fans and their catchments. Furthermore, basic information about the area are provided, that is, the distribution of glaciers and extent of the fjord are given as polygons,

Results
Fan-shaped landforms developed along the valley sides and fjord margins of the Petuniabukta region. They were unevenly distributed over the study area, with the highest concentration to the east and north of Petuniabukta (Main Map 1). In total, 297 fans were mapped and classified according to the dominant processes shaping their surface (Main Map 1, Table 1): (1) Colluvial fans (n = 229) with surfaces modified by rockfall and snow avalanches.  (Table 2). They were short (median value of fan length was 208 m) and steep (median value of fan slope = 31°). Fan apex was often located high (median value was almost 300 m a.s.l.); however, the variability was large (from less than 70 m to more than 650 m a.s.l.). Elevation of the lowest parts (fan toe) ranged from 2 to almost 570 m a.s.l. (median value = 168 m a.s.l.). Such variations led to a sizable differentiation in fan relief: from 29 to 314 m.

Debris-flow-dominated fans
Debris-flow-dominated fans were located mainly in Ebbadalen and along the south-eastern part of Ragnardalen. The second biggest concentration was observed on the slopes of the Pyramiden mountain, where most fans terminated in the fjord.
The area of debris-flow-dominated fans ranged from less than 3000 m 2 to more than 314,000 m 2 , with a median value of 57,550 m 2 ( Table 2). The median elevation of the fan apex location was 161 m a.s.l., while the median elevation of the toes was 26 m a.s.l., resulting in relief ranging from 40 to 253 m. The length varied from less than 150 m to almost 1250 m (median value = 501 m). As some of the debris-flow-dominated fans terminated at the sea level, they probably also extended offshore, such that their total length and relief might be different. The median slope of the subaerial parts of fans was 11°.
In terms of the general shape of the debris-flowdominated fans, most were characterized by a concave longitudinal profile (only 26% of them had a straight long profile) and a plano-convex cross profile (more than 60% of the total number of debris-flow-dominated fans) (Table 3). However, when measuring the local values of curvature, the mean profile curvature was concave, whereas no dominant group could be indicated for mean planar curvature. About 40% of the debris-flowdominated fans are solitary landforms, with the remaining forms coalescing on one or both sides.
Catchments of the debris-flow-dominated fans had an area from 5500 m 2 to more than 772,000 m 2 (median value of almost 200,000 m 2 ), with slopes of 19-41°(median slope value of 34°) ( Table 1). The median elevation of the highest point of the catchments was 712 m, and the median relief was 473 m ( Table 2). The lithology of catchments of debris-flow-dominated fans consisted mainly of: (1) carbonate rocks, evaporates (∼20 catchments) and (2) dolomite, sandstone gypsum (∼20 catchments) ( Figure 6).

Fluvial-flow-dominated fans
Fluvial-flow-dominated fans were not common within the study area, constituting only 6% of the total number of fans. The largest landforms were observed in Ebbadalen and Ragnardalen, with smaller fans being scattered along the fjord and the margin of Nordenskiöldbreen. Their area ranged from about 6500 m 2 to more than 450,000 m 2 , with a median area value of about 60,000 m 2 (Table 2). They were relatively flat, with a median slope of 8°(slope values ranging from 5°to 13°). This was the result of low locations for the fan apexes (median value = 112 m) and fan toes (median value = 15 m), and consequently low fan relief (median = 87 m). Fans were long (median value = 475 m) and wide (median value = 318 m). Note: 'dominated'means that the specified lithological unit occupied more than 50% of the catchment area; 'and other'means that each of the specified lithological units occupied less than 50% of the catchment area.
The longitudinal profiles of fluvial-flow-dominated fans were mostly concave, with only 6 fans (31% of the fluvial-flow-dominated fans) characterized by a straight profile (Table 3). Cross-profiles were mostly plano-convex (13 fans). More than 60% of the fluvial-flow-dominated fans (12 fans) were solitary landforms, with the remaining ones coalescing with other fans. Most of the fan surfaces had a dominant southern aspect.
Catchments of fluvial-flow-dominated fans were large, with the area ranging from about 56,600 to more than 2,050,000 m 2 (Table 2). Such large areas of catchments influenced their elevation range (median value of relief = 554 m), leading to high values of slopes that ranged from 17°to 32°(median value = 29°). The length of catchments was also large, with a median value of 1211 m. In terms of lithology, fluvial-flowdominated fans' catchments occupied areas of dolomite, sandstone, gypsum, clastic and carbonate rocks, and evaporites ( Figure 6).

Conclusions
This research documented the distribution of three types of fans (colluvial fans, debris-flow-dominated fans, and fluvial-flow-dominated fans) and their catchments in the central part of Spitsbergen Island (Petuniabukta region) (Main Map 1). Colluvial fans were the most common within the study area, followed by debris-flow-dominated fans, and a much smaller number of fluvial-flow-dominated fans.
The uneven spatial distribution of fans was probably influenced by the topography of the area. A larger number of fans, which were located in the eastern and northern part of the region, are potentially related to the existence of a larger number of post-glacial valleys, which were also longer and wider than valleys along the western coast of Petuniabukta. Therefore, the topographic conditions for fan development (i.e. accommodation space) were more favourable there.
Additional characteristics of the fans are demonstrated by a series of maps that present the spatial pattern of fans' long-and cross-profiles, slope, and aspect (Main Map 2). Further morphometric properties of fans and catchments are presented in Tables 1-3. Colluvial fans were the smallest and steepest, whereas debris-flow-dominated and fluvial-flowdominated fans were significantly larger and characterized by more gentle slopes. Moreover, the catchments of colluvial fans were much smaller in comparison to the large catchments of fluvial-flowdominated fans. Other morphometric properties of both fans and catchments also differed according to type. The presented data will be used for further in-depth analysis of the relationship between the surface morphology of fans, dominant processes, and morphometric variables.
The character of the surfaces was assessed based on various remote sensing data and verified during fieldwork. Morphological fan types were based upon the approach of de Haas et al. (2015). They provided a description of fan surface morphology for the Adventdalen region (about 60 km south of Petuniabukta), and have documented the major role of snow avalanches in the modification of colluvial fan surfaces. In contrast to this finding, the number of colluvial fans with a flattened top (an indicator of snow avalanche activity) was relatively small (only 11 fans) in the vicinity of Petuniabukta. Moreover, we did not observe any tongueshaped snow-avalanche-dominated fans. These results suggest that the role of snow avalanches in our study area was potentially less important. A possible explanation might be related to the character of the mountains in the Petuniabukta region. In contrast to the Longyearbyen surroundings, where the landscape is dominated by relatively flat plateaus, mountain ridges near Petunibukta are generally narrow (with the exception of part of the Wordiekammen). Moreover, the amount of precipitation is slightly lower than in Adventdalen. These two facts might influence the amount of snow cover that can be accumulated along the edges of the mountain ridges, and consequently limit the number of cornice-fall avalanches, which are very common in the vicinity of Longyearbyen (Eckerstorfer & Christiansen, 2011;Eckerstorfer, Christiansen, Vogel, & Rubensdotter, 2013). Further research on this aspect is needed, and a comparison in terms of the characteristics and morphometry of fans located in the vicinity of Petuniabukta with those of the Adventdalen region would be helpful in better understanding the evolution of fan surface morphology in high-Arctic periglacial settings.

Software
Fans were identified based on orthophotos, high-resolution satellite imagery, and DEMs using Esri ArcGIS 10.3. Landforms were digitized on-screen in ArcMap. Morphometric analyses of fans and catchments were performed in ArcGIS and Whitebox GAT. The layouts of the maps were produced in ArcMap.