Introduction

Rugged mountains exert a controlling influence in triggering rock avalanches and bedrock landslides all over the world (Hewitt 2002). Reconnaissance-level landslide susceptibility maps are those carried out without ground checking for verification, using remote sensing techniques and morphometric analogs of common landforms associated with recent landsliding. These products are intended to serve as preliminary evaluations, carried out prior to any significant development (Guzzetti et al. 1999; Balteanu et al. 2010; Ahmed and Rogers 2014). Landslide inventory maps can be prepared for any region based on anomalous topographic signatures (Varnes 1978; Hansen 1984; Carrara et al. 1991; Hutchinson 1968; Dikau et al. 1996; Cruden and Varnes 1996; Van Den Eeckhaut et al. 2009, 2011; Ahmed and Rogers 2014). Mass wasting events in excess of five contour interval lengths along the slope usually leave noticeable changes in surface topography (Ahmed and Rogers 2014). These patterns can be recorded, characterized, and mapped with the aid of remote-sensing imagery, such as aerial photographs, and digital elevation models (DEMs) data (Rib and Liang 1978; McCalpin 1984; Pike 1988). Various types of landslides exert unique and distinct surface expression, depending upon their respective slope movement, failure extent, depth, and size (Varnes 1978; Rogers/Pacific 1994; Cruden and Varnes 1996).

The scale of the inventory maps is influenced by the size of the study area, the resolution of available data, and financial resources. Small-scale landslide inventories (<1:100,000) can be prepared from the analysis of stereopair aerial photographs and hillshade topographic maps (Cardinali et al. 1990; Brabb 1991; Rogers/Pacific 1994; Guzzetti et al. 2006; Ahmed and Rogers 2014). Satellite imagery and shaded DEMs of varying resolutions are increasingly available world-wide, and can be utilized as significant tools in discerning the presence of past land slippage.

An understanding of these anomalous topographic features aids interpreters in identifying and reconstructing recent and prehistoric landslide features in the steep mountain valleys of the Karakoram and Himalaya ranges of Pakistan. According to Hewitt (2002), the geometry of the opposing slopes, run-out paths, and pattern of interfluves (higher ground between two rivers or former river terraces formed by earth flows and bisected by fluvial erosion; Whittow 1984) are significant variables controlling the shape and surface geomorphology of rock avalanches in this region. These mountains are renowned for the incidence of snow avalanches, rock avalanches, debris flows, rockfalls, and landslide dams (Hewitt 1982, 2002; Goudie et al. 1984; Rogers and Ahmed 2013; Ahmed et al. 2015). Roughly 5 % of the ~70,000 km2 of northern Pakistan was examined for mass wasting features (Hewitt 2002). The total area and volume of these rockslides varies from 2 to ~40 km2, and 1.5 × 106 to 20 × 106 m3, respectively (Hewitt 2002, 2011). Hewitt (2002) found more than 40 rock avalanches have impacted the main river channels in the Karakorum and Himalaya region, creating large impoundments of water. Many of these slide dams have been partially breached, while the remnant lakes continue to accumulate sediment for many kilometers upstream.

Many of the rockslide deposits in the Karakorum and Himalaya ranges are complex in nature and can be difficult to recognize in the field if they do not become significantly disaggregated with downslope translation. Environmental factors, such as weathering, erosion by glaciers and rivers, or deposition behind landslide dams, can also complicate their recognition. Most of the larger landslides are influenced by structural geologic features, such as bedding, foliation, jointing, sheared zones, and faults (Hewitt 2002; Ahmed and Rogers 2014). Many of the landslide and glacier dams, as well as glacial surges, have been documented in eastern part of the Karakoram and Nanga Perbat Haramosh Massif (NPHM) Ranges in the Upper Indus River Basin (Hewitt 1982, 2002).

The main focus of this study was to identify anomalous topographic expression (i.e., parallel and converging drainage patterns, divergent contours, crenulated contours, and isolated knobs, etc.) felt to be reflective of large-scale mass movements in the Shyok River Basin. The proposed methodology was then validated by comparing historically recorded landslides in the basin (Hewitt 1982, 1998, 2002; Korup et al. 2010; Hewitt et al. 2011) with those mapped “blind,” as part of this regional effort. In every case the historic landslides were positively identified.

Overview of the study area

The Shyok River is one of the main tributaries of the Indus River in northern Pakistan. This river originates from the Rimo Glaciers and a portion of the Siachen Glaciers (Negi 1991), then flows through the Kashmir region (Ladakh area) and enters northern Pakistan near the Ganche District. The Shyok River is fed by three principal tributaries, the Nubra (in Kashmir), Saltoro, and Hushe Rivers, and encompasses a watershed of 10,500 km2 (Fig. 1).The Shyok River flows approximately 550 km before it merges into the Indus River near Keris (Khaplo), east of Skardu.

Fig. 1
figure 1

Map of Shyok River Basin in the upper Indus River watershed of northern Pakistan. Note the structural control of many linear tributaries, trending north-northwest to south-southeast

The majority of hill slopes in northern Pakistan exceed 45°, with at least 500 m of elevation gain. The severity of the slopes aids the frequency of landslides (Hewitt 2002). The slopes within the Shyok watershed have spawned a dense array of bedrock landslides, which appear to have occurred more or less uninterrupted throughout the Holocene. The river channel exhibits slight meandering or braided flow where the valleys spread out and widen (e.g., at Nubra and Khaplo). This tendency increases after the river passes through steep-sided gorges. Devastating flooding has been reported more than 1200 km downstream of the river, due to the catastrophic failures of natural dams, between 1926 and 1932 (Hewitt 1982, 2002). The valley bottoms are filled with fine, well-sorted, silty and sandy sediments accumulated in reservoirs trapped behind temporary landslide dams. Breached landslide dams have been identified at Haldi, Hum Bluk Hushe, Litak Hushe, Kunis-Ghah, and Shyok. Each of the outbreak floods associated with these breaches have deposited elevated, but discontinuous, outbreak flood terraces (Shroder and Bishop 1998; Hewitt 2002).

Geologic and tectonic settings

The Shyok River watershed is a part of the Karakorum-Himalaya crystalline and thrust zone (Kazmi and Jan 1997). The main part of the study area lies within the Karakoram metamorphic complex (Searle 1991). The dominant rock units in this complex include; gneisses, metasediments, marbles, and amphibolites, etc. The subordinate units include predominantly plutonic metasedimetary rocks of the Kohistan-Ladakh batholith (Kazmi and Jan 1997; Ahmad et al. 2003). The principal formations mapped within the study area are shown in Fig. 2 and their respective descriptions are presented in the legend.

Fig. 2
figure 2

Regional geologic map of the Shyok River watershed (Geological Survey of Pakistan at a scale of 1:1,000,000)

Significant rockslides have been observed across the entire basin, without any particular preference to bedrock type or formation (Hewitt 2002). Most of the largest rockslides are believed to be seismically triggered, based on historic records and descriptions (Cockerill 1902; Mason 1914; Keefer 1984; Hewitt et al. 2011). A number of mega-rockslide avalanches (area >10 km2) have been documented along the main channel and its tributaries, including those at Masher-brum, Saltoro, and Khaplo (Hewitt 1998, 2002).

Methodology

Reprocessed advanced spaceborne thermal emission and reflection radiometer (ASTER) global digital elevation model 2 (GDEM2) tiles with 30-m pixel resolution of the Shyok River watershed were downloaded from an open source website (http://www.jspacesystems.or.jp/ersdac/GDEM/E/index.html). These GDEM tiles were then georeferenced to UTM Zone 43 N, using ENVI 4.8 (Environment for Visual Information Solutions software). The ASTER GDEM2 tiles appear to offer a better option for a regional level of studies over the older shuttle radar topography mission (SRTM) data (90 m, 3 arc s). This is likely ascribable to its higher resolution, which allows more accurate measurements in steep mountainous terrain due to higher radar reflectivity (Tachikawa et al. 2009; Tachikawa et al. 2011). The Shyok watershed was extracted using the hydrology module in ArcGIS10. The shape file of the Shyok River was then used to clip the hillshade map of the study area using spatial analyst tools.

The topographic map sheets employing 40-m contour intervals at 1:200,000 scale were obtained from an online source, "Topographic Maps of the World" (www.mapstor.com), and overlaid with the hillshade map to create a seamless shaded topographic map of the region. The study area was then subdivided into small sections using ArcGIS 10 and printed on large sheets (122 × 92 cm) to allow easy viewing from any azimuth, which helps delineate landslide features. The mapped landslides and related features were digitized in ArcGIS 10 and geo-referenced with the original hillshade map.

The most common landslide features are those that are generally easiest to identify by their anomalous topographic expression (Doyle and Rogers 2005). These include translational block slides, earthflows, retrogressive slump complexes, and debris flows (WP/WLI 1993), shown in Fig. 3. The guidelines provided in the literature enable the recognition of different types of landslide-related features (Rogers 1980; Crosta 2001; Glade 2001; Crozier 2010; Van Den Eeckhaut et al. 2009, 2011; Hsiegh et al. 2012; Ahmed and Rogers 2014).

Fig. 3
figure 3

Examples of anomalous topographic expressions that can develop by mass wasting and subsequent surface erosion; a shows diagnostic topographic features helpful in mapping landsides, b translational slides are often characterized by the formation of isolated knobs by active erosion along lateral scarps; smaller landslide are commonly developed upon the displaced parent mass, and are usually much easier to discern. This example includes coalescing earthflows and rotational slumps. Other adjacent slides may form by upslope erosion of the parent slide mass by the adjacent stream channels (Rogers/Pacific 1994), c depositional lobes typical of coalescing earthflows, d geomorphic indicators, representative of rotational slumps (Rogers/Pacific 1994)

A comparative examination of slopes is of the utmost importance in identifying evidence of mass wasting. Such features are laterally restricted, because landslides tend to translate downslope, disrupting topographic patterns controlled by bedrock structure, such as benches and cliffs (Rogers 1980; Rogers/Pacific 1994; Doyle and Rogers 2005). The screening process begins with an examination of topographic patterns, searching for anomalies and inconsistencies which do not appear on adjacent slopes or changes which do not appear to be typical of underlying lithologic and structural contacts (Terzaghi 1950; Rogers 1980; Cruden and Varnes 1996; Doyle and Rogers 2005; Crozier 2010; Hart et al. 2012). In some cases, deranged or parallel drainage patterns allow large features, such as detachment complexes and landslides, to be easily discerned (Doyle and Rogers 2005).

The most common topographic anomalies associated with landslides are briefly described below (also see Fig. 3).

Divergent contours

Contours that curve upslope adjacent to contours that curve downslope. These often suggest the removal of material (deflation) from the upper portion of a landslide and deposition (inflation) in the lower portion of the slope.

Crenulated contours

Contours that exhibit waviness or scalloping, not otherwise associated with the underlying geologic material and/or structure. Crenulated contours are often diagnostic of terrain underlain by repeated sequences of superposed land slippage, especially flowage of disaggregated debris.

Arcuate headscarp evacuation areas

Arcuate headscarp evacuation areas are steep-sided, curvilinear features at the upslope boundary of a landslide, formed by translation of sliding material downslope.

Isolated topographic benches

Isolated topographic benches are relatively broad, flat areas which often form below the headscarp evacuation area formed by back-rotation of slumped masses or infilling of pull-apart grabens.

Extended topographic ridges or isolated topographic knobs

These features are often formed by deep-seated translational movement of slump blocks or translational slides causing separation of a ridge’s caprock.

Sudden up- or down-slope turns in hillside contours

Sudden up- or down-slope turns in hillside contours result from sharp changes in topography that do not appear related to the underlying geologic structure or material changes. They are often caused by downslope movement of an isolated portion of the hillside, parallel to the slope’s natural fall-line.

In steep mountainous areas, the ASTER-derived GDEM2 tiles with 30 m resolution are generally suitable for the identification of landslides greater than five times the contour interval (Ahmed and Rogers 2014). For example, if the contour interval is 40 m, the slide feature would need to perturb at least five consecutive contours to be discerned. If a translational slide occurred on a natural slope inclined about 26° that would necessitate an elevation differential of at least 200 m, which would have a minimum length of about 500 long on a map (Ahmed and Rogers 2014).

Rogers/Pacific (1994) and Rogers (1998) presented 13 steps for mapping common landslide features using topographic indicators. These include very large, deep-seated bedrock slides down to shallow earthflows. This simplified procedure was based on the guidelines presented by Terzaghi 1950, Varnes et al. 1984, Hutchinson 1968, Carrara et al. 1991, and Cruden and Varnes 1996. Most of the smaller slides in mountainous terrain are developed upon deep-seated “parent slides” (landslides that extend well below the weathered regolith, into underlying strata). In this criterion, each stage involved in the initiation of various types and/or scales of sliding are explained and described in a comprehensive manner, using block diagrams and topographic expressions typical of terrain that has suffered deep-seated sliding (Figs. 3, 4). By following the guidelines explained above, the geomorphic analyses (using the topographic mapping protocols) were performed across the Shyok River watershed to identify those anomalous features likely to be attributable to mass wasting (Fig. 5).

Fig. 4
figure 4

The image shows some common examples of anomalous topographic expression, which serve as useful indicators for landslide mapping. These include anomalous drainage patterns (parallel and converging), bedrock contacts and faults (if any), anomalous topographic benches, and isolated knobs

Fig. 5
figure 5

Interpretation of gross landslide features (mass wasting), using the guidelines shown in the Fig. 4 and in the “Methodology” section. Most of these features appear to be composite landslides, which exhibit different types of movement in different areas of the slide (e.g., rotational, translational, and flowage)

Results and discussions

The application of anomalous topographic protocols allowed the identification and reconstruction of many landslide features in the steep mountain valleys of the Karakoram and Himalaya. According to Hewitt (2002), the geometry of the (impact) opposite slopes, run-out paths, and pattern of interfluves, are major controlling factors shaping the surface morphology of rock avalanches in this region.

This reconnaissance-level inventory identified over 600 landslide features more than 500 m long in the Shyok River Basin (Fig. 6). The average size of the mapped slide features was found to be >1000 m long based on the scale of the hillshade map. The majority of these landslide features appear to be composite landslides (WP/WLI 1993), demonstrating characteristics of more than one slide type likely developed in the parent rock units, because of complex deformation, high rates of tectonic uplift, and severe rates of denudation. Subsequent movements are often of a different type, due to toe erosion, monsoons, and earthquakes. These stress tensors often trigger smaller slump blocks, flows, and rock avalanches.

Fig. 6
figure 6

Landslide inventory map of the Shyok River watershed. The western tip of this watershed along the Indus River channel was mapped as part of another regional study reported in Ahmed and Rogers (2014)

A comparison was then made of this inventory map with the available data on historic documented landslides (Hewitt 1982, 2002, 2011). This suggests that nearly all of the documented features overlap with those slides denoted on the landslide inventory map (see Fig. 7).

Fig. 7
figure 7

Shyok River landslide inventory map, the red dots denote locations of historically documented landslides

Some example features have been selected to describe the mapping procedures. Figure 8 presents an excerpt of the landslide inventory map from the Hushe River. The topographic map exhibited anomalous features along the both sides of the river, indicative of numerous mega slides and past landslide dams (with volumes >10 million m3). The historic rockslides in this region include interfluve overflow type, such as the Hum Bluk event, as well as extreme impact slope effects, like the Litak landslide along the Hushe River (also shown in Fig. 8). The debris generated from these rock avalanche dams had sufficient kinetic energy to run up the opposing slopes (Hewitt 2002).

Fig. 8
figure 8

Portion of the landslide inventory map illustrating a reach of the Hushe River. The red dots denote documented slides that occurred during historic time

Figure 9 shows the Khaplo area along the Shyok River, a few kilometers downstream of its junction with the Hushe River. The rocks exposed in this region are of Mesozoic to Paleozoic age and include volcanic rocks, limestone, red shale, and conglomerates. The Karakoram thrust fault passes through this area, from SE to NW (see Fig. 2). The mapped landslides and historically documented features in this region are denoted in Fig. 9a.

Fig. 9
figure 9

a Landslides mapped in the Khaplo area of the Shyok River. The Haldi rock avalanche appears to be an interfluve overflow feature, which occurred during historic time. Note the debris train extending across the opposing ridge and moving down the far slope, with its main scarp and the debris field shown in shaded light yellow, b cross-section through the Haldi rock avalanche, showing the likely structure of the remnant debris, which overtopped a valley ridge approximately 500 m high. Slide debris was also observed on the right bank of the Shyok River, indicating the subsequent breach of the landslide dams across both channels (Vertical exaggeration 2x)

The Haldi prehistoric rockslide avalanche is another example of an overtopping interfluve type landslide. The source rock was located approximately 2000 to 2500 m above the valley floor. The rock avalanche debris initially filled in the Saltoro River valley, blocking the channel. Subsequent avalanche debris passed over the valley fill and rode up the opposing valley slopes (nearly 500 m high), overtopping it and then moving down the far slope (Hewitt 2002). Some portion of this debris also blocked the Shyok River Valley (see Fig. 9a, b).

Figure 10 shows the section of landslide inventory map in the Kharkoo area. The geologic units exposed in this region belong to the Karakoram Batholith and associated plutons intruding Miocene- to Cretaceous-age rocks. There are a number of historic landslides identified in this area, shown as red dots. The historic Kunis-Ghwah landslide is a prominent interfluve event in this area. The epigenetic gorge in this region was formed because of the emplacement of many historic rockslide deposits along the Shyok River (i.e., Kunis-Gwah rock avalanche event). Hewitt (2002) observed that the distal deposits of these types of avalanche events generally moved downward and stalled against the opposite slope, creating asymmetrically long profiles, with the main deposits emplaced against the opposing slopes, which is the salient feature of rockslide avalanches triggered in deep, narrow gorges. Due to their fault-controlled lineation (adjacent to the MKT), rock avalanche deposits have been utilized as permanent settlements for villages, infrastructure projects, and agriculture fields in the Khaplo, Skardu, and other districts of northern Pakistan (Hewitt 2002). Recurring rockslides may have catastrophic impacts in these same areas. These issues need to be addressed, at least in the areas of economic interest and/or densely populated areas.

Fig. 10
figure 10

Excerpt from the landslide inventory map showing documented rockslides (red circles) in the Kharkoo area along the Shyok River

The reliability of these sorts of inventory maps depends upon the available historic landslide records and effective use of landslide identification protocols. The procedures employed in this study to delineate landslide-related features appear to have been validated by the historic data, but these data are spread over a large geographic area which has only been subject to recorded observations for about 200 years. Most of large historic events like the Hushe R, Haldi, Hum-Bluk and Kunis-Gwah events coincide with the landslide features identified in this inventory mapping. The Shyok River watershed inventory overlapped with the regional level landslide susceptibility hazard maps generated for the Indus River watershed (Ahmed et al. 2014). Most of the mapped features (more than 80 %) overlap areas with high to very high hazards (Fig. 11). These comparisons are encouraging for a reconnaissance-level study, which is only intended to identify features believed worthy of further study and field verification.

Fig. 11
figure 11

Overlay of landslide inventory map and susceptibility map (Ahmed et al. 2014) of the Shyok River watershed

Conclusions

Topographic expression is a powerful tool in identifying slopes that have been modified by mass wasting processes. The use of topographic mapping protocols allows an inexpensive means to screen large mountainous regions for bedrock landslide features. This reconnaissance-level inventory identified over 600 bedrock landslide features more than 500 m long within the Shyok River watershed. The majority of these landslide features exhibit characteristics typical of composite landslides developed in the parent bedrock units. Most of the bedrock landslide features we identify are more than 1 km in length and tended to be structurally controlled when they were initially triggered.

The historic rockslide events identified in the Hushe and Khaplo areas of the Shyok River watershed (i.e., Haldi, Litak, and Kunis-Gwah) also exhibited anomalous topography, which served as useful analogs for mapping similar landslide-related features across the subject watershed. The documented historic events appear to validate the procedure employed here, which is intended to tentatively identify similar landslide-related features (>500 m in length). These maps are intended to serve as “guides” for more detailed analyses of specific projects sites, such as power transmission corridors, pipelines, structures, highways, tunnels, dams, and powerhouses.