Geomorphological research of large-scale slope instability at Machu Picchu, Peru

Geomorphology 89 (2007) 241 – 257 www.elsevier.com/locate/geomorph

Geomorphological research of large-scale slope instability at Machu Picchu, Peru Vít Vilímek a,⁎, Jiří Zvelebil b , Jan Klimeš c , Zdeněk Patzelt d , Fernando Astete e , Václav Kachlík f , Filip Hartvich c a

Department of Physical Geography and Geoecology, Faculty of Science, Charles University, Albertov 6, 128 43 Prague 2, Czech Republic b Czech Geological Survey, Klárov 3, 100 00 Prague 1, Czech Republic c Institute of Rock Structure and Mechanics, Czech Academy of Sciences, V Holešovičkách 41, 180 00 Prague 8, Czech Republic d Czech Switzerland National Park Administration, Pražská St. 52, 407 46 Krásná Lípa, Czech Republic e Instituto Nacional de Cultura, Calle San Bernardo s/n., Cusco, Peru f Department of Geology and Paleontology, Faculty of Science, Charles University, Albertov 6, 128 43 Prague 2, Czech Republic Received 28 January 2006; received in revised form 12 December 2006; accepted 14 December 2006 Available online 16 December 2006

Abstract A multidisciplinary approach has been adopted to study the slope movements and landscape evolution at the archaeological site of Machu Picchu and its immediate surroundings. The basic event in the paleogeomorphological evolution of the area was the large-scale slope movement, which destroyed the originally higher ridge between Mt. Machupicchu and Mt. Huaynapicchu. Within remnants of that primary deformation, several younger generations of slope movements occurred. The laboratory analyses of granitoids revealed highly-strained zones on the slopes of Mt. Machupicchu, which strongly affect the largest slope deformation. The borders of the largest slope deformation are structurally predisposed by the existence of fault zones. The majority of various types of slope movements on the so-called Front Slope (E facing) and Back Slope (W facing) are influenced by the alignment between topography and joints. Along with slope movements, fluvial erosion and tectonic disturbance of the rocks have been affecting the evolution of the landscape. A monitoring network for dilatometric and extensometric measurements was used to detect the present-day activity of rock displacements within the archaeological site. In addition to standard mapping of surface hydrogeological phenomena, eleven express slug tests were conducted to verify the infiltration potential of precipitation. The results of these surveys indicate that recent large-scale slope movement as suggested by some previous studies is doubtful, and the detected movements can be explained by individual movements of rock blocks or several other mechanisms including sinking of archaeological structures, subsurface erosion and annual changes in the water content of the soils. © 2007 Elsevier B.V. All rights reserved. Keywords: Large-scale slope deformation; Landscape evolution; Geohazards; Machu Picchu; Peru

1. Introduction The famous archaeological site of Machu Picchu is situated in southeastern Peru, approximately 100 km to ⁎ Corresponding author. Tel.: +420 221951361. E-mail address: [email protected] (V. Vilímek). 0169-555X/$ - see front matter © 2007 Elsevier B.V. All rights reserved. doi:10.1016/j.geomorph.2006.12.004

the northwest of Cusco. The citadel of Machu Picchu was built on a steep ridge, about 500 m above the incised meanders of the Urubamba River (Figs. 1 and 2). UNESCO declared Machu Picchu a World Heritage site in 1983. This site was discovered for modern archaeology by Hiram Bingham in 1911, and since then, it has been under continuous scientific study. Despite this,

242

V. Vilímek et al. / Geomorphology 89 (2007) 241–257

Fig. 1. 3D view of the study area at the Machu Picchu meander, with draped air-photo imagery. No vertical exaggeration.

geomorphology and engineering geology had been, until recently, on the periphery of scientific interests. Carreño et al. (1996) and Carreño and Bonnard (1997) first brought attention to the problem of rock slope stability and possible catastrophic slope movement endangering the archaeological site. Since then the site has been incorporated into the UNESCO/IGS project IGSP 425 for thorough investigations of slope stability (Sassa et al., 2000, 2001; Vilímek and Zvelebil, 2002). Unfortunately, some preliminary findings and initial ideas, especially those related to damage to the site by a large rockslide, were distorted in newspaper/magazine reports. As a world-famous tourist location, Machu Picchu suddenly became the focus of worldwide public attention in relation to landslides, and the primarily purely scientific theme became biased by political and economical aspects. The media tended to stress the possibility of hazardous landslides, although some geotechnical and archaeological experts have expressed serious doubts about the existence of dangerous slope movements (McEvan and Wright, 2001). Four rock falls with volumes of thousands of cubic meters, which occurred in December 1995 and January 1996 (Carreño et al., 1996), attracted interest in the instability of the rock slopes at the archaeological site. Two of these failures, whose main scarps were at an altitude of 2380 m a.s.l. (immediately below the ruins), destroyed part of the Bingham Road, the only access from the valley bottom to the archaeological site (Fig. 3). The third rock fall occurred on Mt. Putucusi on the opposite side of the Urubamba valley (Fig. 1). Rock

blocks hit the railway tracks near the old Machu Picchu railway station (Fig. 1). The fourth rock fall took place from a large cliff along the Urubamba fault on the right bank of the Urubamba River, approximately 1.5 km NE from the archaeological site (Fig. 3). These rock falls may indicate present-day activity of much larger, deep-seated slope failures, which increase the disintegration of the rock mass and increased instability on the “Front Slope” (the East facing slope below the Machu Picchu meander where the Bingham Road is located; see Fig. 3). The large-scale deformation on the Front Slope of Machu Picchu Ridge was estimated to have activated at least 6 × 106 m3 of rock. The repeated destruction of the Bingham Road by ground sinking in the vicinity of the Machu Picchu hotel was considered to be another indicator of present-day activity of that large-scale slope movement by Sassa et al. (2000). A land displacement monitoring network, recommended by Carreño and Bonnard (1997), was established by Sassa et al. (2000). Twelve sites were instrumented with wire extensometers. Two of them were equipped with an automatic registration device. The results of direct monitoring measurements (Sassa, 2001; Sassa et al., 2001) suggest displacement rates of mm per month or higher. Sassa et al. (2000) also described seven main rockslide bodies and classified them into active and potentially active ones. According to them, the Front Slope deformation is a multigeneration one, taking place mostly along discontinuities striking W–E with dips 30–50° towards the valley. The

V. Vilímek et al. / Geomorphology 89 (2007) 241–257

243

Fig. 2. Geomorphological map of the study area based on field mapping. 1: floodplain, 2: river terrace, 3: alluvial cone, 4: talus, 5: colluvial slope (slope covered by variable thickness of weathering mantle and soils), 6: rock slope (bare rock slope or slope with less than 1 m of weathering mantle), 7: cliff, 8: ancient rock slide, 9: recent landslide (N50 m in length), 10: recent landslides (b50 m), 11: prominent ridges, 12: gullies, 13: double ridges, 14: rapids, 15: block fields, 16: fissure caves, 17: prominent peaks, 18: saddle, 19: rivers, 20: urban areas, 21: studied catchment borders. C: Inca City of Machu Picchu, MPV: Machu Picchu Village.

deformations on the Back Slope, the west facing slope below the Machu Picchu site (Fig. 3), are predominantly topples. The basic scientific questions addressed in this article are: • Is there evidence of slope movements in the vicinity of the Machu Picchu archaeological site? • Are the movements still active, and could they endanger the archaeological site? • Will a catastrophically rapid slope collapse occur in the near future (in a matter of years)? • Are there any other geodynamical processes highly unfavourable to the site? • Is there a need to expand the monitoring activity and research at this site? If so, which techniques should be implemented? 2. Study area The archaeological site of Machu Picchu lies between 2450 and 2500 m a.s.l. on the narrow ridge that connects the peaks of Mt. Huaynapicchu (2700 m a.s.l.) and Mt.

Machupicchu (3051 m a.s.l.), and it is surrounded by the Urubamba River meander (Figs. 1 and 2). The river under the ridge flows at an altitude of ca. 2000 m a.s.l. under the NNE slope of the Machupicchu ridge. The ridge is a part of the Cordillera Oriental of the Andean mountain system. The general direction of the valley of the Urubamba River, in which the ridge is located, is NW−SE. This part of the valley is located between two glaciated ranges: the Nevado Verónica (5750 m a.s.l.) on the NE side of the valley and the Cordillera Vilcabamba (Nevado Salcantay, 6246 m a.s.l.) on the SW side. Bedrock is composed of various heterogeneously deformed leucogranites and tonalites of the Machu Picchu pluton of Permian age, dated to 240 ± 10 Ma based on the Rb–Sr method (Ponce et al., 1999). The Machu Picchu pluton intruded into NW−SE striking Early Palaeozoic sediments. Both the Palaeozoic metasediments and the Late Palaeozoic (Permian) intrusives underwent major deformation during the Laramide orogenesis, which occurred close to the Cretaceous/Palaeogene boundary. A NW−SE directed thrust dipping to the NE and also the regional scale folds in metasediments originated during this event.

244

V. Vilímek et al. / Geomorphology 89 (2007) 241–257

Fig. 3. Detailed map in and around the Machu Picchu site, showing structural conditions. 1–4: rosette diagrams; 1: overall diagram (576 measurements), 2: archaeological site, E of the central depression at the main square (211 measurements), 3: archaeological site, W of the central depression at the main square (197 measurements), 4: Front Slope at the Bingham Road (168 measurements).

The Machu Picchu archaeological site is located to the west of a NW−SE striking shear zone, parallel to the Urubamba fault (Fig. 3). This newly discovered ductile-to-brittle shear zone is accompanied by several other brittle faults with the same direction, and cut by less important NE–SW and E−W striking fault zones. The geometry of these faults has been described in different ways (Kalafatovich, 1963; Ponce et al., 1999; Sassa et al., 2000; Caillaux et al., 2001). The relationship between these faults and the ductile shear zone of NW−SE direction, dipping about 80° to the NE, is even less understood. The Urubamba River, which eroded through a granitic pluton, formed a deep, narrow valley in the surroundings of the Machu Picchu site (Fig. 1), called the Torontoy canyon. The canyon formation was facilitated by the relatively high strength of crystalline rocks, which give high and steep valley sides (generally between 50° and 80°). Quaternary deposits in the study area are mostly coarse (sandy to bouldery) fluvial and mass flow sediments deposited in river terraces and alluvial cones, along with colluvial sediments of talus deposits. Most of the sediments are stored in the bottom of the valleys. Slopes may be covered by a mantle of highly

fractured and weathered bedrock with a thin layer of soil, or consist of bare rock. Glacial deposits were described by Carlotto et al. (2001) in the headwaters of the Aguas Calientes basin (a right-side tributary of the Urubamba River) and by Kalafatovich (1957) in the upper Urubamba valley close to Urubamba City (60 km NE of the Machu Picchu site). The climate in the vicinity of Machu Picchu is humid and subtropical. The total mean annual precipitation is 1950 mm and mean annual temperature varies between 12 °C and 15 °C (Wright et al., 1997). Rainfall is most abundant from January to March (300 mm month− 1 on average with a peak in February), whilst the period between May and July is the driest (40 mm month− 1 on average). September is characterized by an average rainfall of 100 mm month− 1, which is important when interpreting the hydrological survey conducted in September 2004. Monthly air temperatures range from a minimum of 6.8 °C to a maximum of 23.4 °C (SENAMHi in Wright et al., 2000). Not only natural rainfall but also additional water is brought to the site through an artificial pipeline from the adjacent catchment. The main aquifer in the area is formed in the accumulations of slope sediments, predominantly bouldery to stony talus deposits with a variable portion of

V. Vilímek et al. / Geomorphology 89 (2007) 241–257

Fig. 4. Location of infiltration test sites (circles 1–11) and of rock sampling sites for petrological analysis (arrows 1–5).

loamy to sandy fill. The valleys surrounding the Machu Picchu site correspond to a sub-Andean ecosystem characterized by forest cover even on high-altitude steep slopes.

245

tions were made from these samples, which were studied by classical petrographic methods under a polarising microscope to ascertain the basic rock forming minerals, and to study the deformation textures in rocks originating under conditions of different intensities of deformation and stress regimes. Detailed field investigation as well as aerial photo interpretations were conducted to collect information about the strike and dip of joints and to identify the most important tectonic lines/fault lines of the study area. Dip direction and dip of slickensides were recorded when possible. In order to assess the significance of recent landscape evolution, we performed geomorphological and engineering-geological field surveys and mapping in the broader surroundings of the archaeological site (Fig. 2). Interpretation of aerial photos taken in 1963, 1991 and 2000 were the basis for the field survey, because the majority of slopes off the trails are inaccessible and the dense vegetation greatly limited visibility. Infiltration rates of the surface soils were assessed through indicative express tests (see Fig. 4 for the location of test sites), during which the time necessary for water to soak away on a circular plane of 110 mm diameter was measured. The occurrence, character and

3. Methods In order to understand the mechanisms, distribution and activity of the mass movement processes in the study area, we exploited numerous methods from geomorphology, structural geology, hydrogeology and engineering geology. The field survey began in 2001, when the primary mapping of slope failures along with assessment of rock movement monitoring within the archaeological site were performed. As a first step, geological, morphological and photogrammetric analysis of the area adjacent to the Machu Picchu archaeological site was carried out. It included the revision of the study of rock outcrops around the Bingham Road and in the area between the Huaynapicchu and Machupicchu peaks (Figs. 1 and 3). During the field study, the orientation, dip, striation on fault planes (where visible) and relationship of small-scale faults and joints were measured. The succession of the origin of fault planes was defined by crosscutting relationships. We also collected various rock samples ranging from relatively undeformed granites to highly sheared or cataclastically deformed granites. Thin sec-

Fig. 5. Picture of the archaeological site with locations of measurements. Arrows show dilatometric measurement sites: A = Acllawasi (A1), C = Cave (C1), I = Intiwatana (I1, I2, I3), M = Wairana o Mirador (M1, M2, M3, M4), P = Plaza (P1), Q = Qhata (Q1), R = Rodadero (R1, R2, R3), T = Principal Temple (T1, T2, T3), W = Wayna Picchu (W1, W2, W3). Hatching of the arrows indicates prevailing movement: Horizontal hatch = extension, vertical hatch = compression, sloping hatch = tilting of the block, honeycomb fill = no movement, blank = unclear movement. Dotted lines with circles are extensometer measurement lines: ES1 = Plaza south 1, ES2 = Plaza south 2 (interrupted in July 2005), EN = Plaza north 1.

246

V. Vilímek et al. / Geomorphology 89 (2007) 241–257

yield of the springs were also mapped. Spring yield measurements were carried out on September 16, 2004. The monitoring network was established in 2001 to detect possible movements of rock blocks inside the archaeological site, which might be connected with deep-seated rock slide activity. The network consists of 20 dilatometric measuring sites placed over nine localities, and of two sets of profiles to be measured by a portable extensometric tape. The monitoring strategy was designed to be as environmentally friendly as possible, taking into consideration the special conservation needs and the great historical and cultural value of the archaeological site. At the same time, a highly precise method was needed because we anticipated a small magnitude of slope deformation within the archaeological area (e.g. Vlčko, 2004). Measurement using a portable rod dilatometer is accurate enough to detect irreversible movements in the order of 0.1 mm yr− 1 within a minimum 3-year time period (e.g. Zvelebil and Stemberk, 2000). Dilatometer requirements on permanent installations are restricted to a pair of tiny metal bolts fixed opposite each other across the measured joint. With respect to the requirements of the archaeologists, the metal bolts were only glued to the rock surface during the first installation in 2001. Unfortunately, this method did not work well and most of the bolts were lost during the first year. In the following year, all bolts were drilled into the rock. Thus, a 3-year period of continuous measurement was completed in early 2006. Extensometric profiles were chosen to span the wideopen space (25–35 m) of the main square of the Machu Picchu site called “Plaza” (Fig. 5). It is the place of the hypothetical, double ridge tension-sliding zone of the deep-seated slope failure. The profiles were installed during autumn 2002. Under fully favourable measuring conditions, this method is able to detect irreversible movements with a trend of 0.5 mm yr− 1 within a minimum 3-year time period.

4. Results and discussion 4.1. Petrology of Machu Picchu granitoids The granitoids of the Machu Picchu pluton in the area studied are fine-grained equigranular to slightly porphyritic leucocratic muscovite–biotite granodiorite, monzogranite accompanied by more leucocratic varieties. In most samples plagioclase prevails over microcline. Quartz content is quite varied, but its concentrations are generally exceeded by K-feldspar and plagioclases. Muscovite and biotite are quite scarce. The mineralogical composition of the samples studied is shown in Table 1 and sampling localities in Fig. 4. According to analyses of the samples (Fig. 6), the granitoids of the Machu Picchu pluton were sheared heterogeneously in high strained zones under green schist facies conditions. The Permian granites were deformed in a ductile to brittle regime into various types of fine-grained to ultra fine-grained rocks — ranging from cataclastic granites (Fig. 6A) with a slightly developed foliation system through cataclastic granites and mylonitized granites (Fig. 6B,C) to well foliated ultramylonitic rocks with mylonites and phyllonites (Fig. 6D) with strong anisotropic dynamo fluidal fabric. The former minerals were mostly recrystallized into a fine-grained mica rich matrix in which only rare feldspar and quartz porphyroclasts were preserved in analysed rock samples. 4.2. Structural analysis of macroscopic and mesoscopic ductile and brittle structures The main fault and deformation zones in the Machu Picchu area have been described by different authors (e.g. Kalafatovich, 1963; Ponce et al., 1999; Sassa et al., 2000; Caillaux et al., 2001). The relationships of these faults and ductile structures (e.g. joints) to mesoscopic structures are less well understood. Our general idea of

Table 1 Petrological characteristic of granitoid rocks from the Machu Picchu archeological site Sample Petrological No classification

Texture

1 2 3 4 5

Phyllonitic + Mylonitic, dynamofluidal + Cataclastic, mylonitic + Hypidiomorphic, equigranular, granitic + Hypidiomorphic, equigranular, granitic +

Phyllonite Mylonite Cataclastic granodiorite Leucomonzogranite Lecomonzogranite

Rock forming minerals Qtz Plg Kf Ms + + + + +

− − + + +

Qtz: quartz, Plg: plagioclase, Kf: microcline, Ms: muscovite, Bt: biotite, Zr: zirconium.

+ + (acces.) + −

Bt Accessories − − − + −

Secondary minerals

Zr, ep, ore Ms (sericite) Zr Ms (sericite) Zr, ore Ms (sericite) Tourmaline, zircon Ilmenite, zircon

V. Vilímek et al. / Geomorphology 89 (2007) 241–257

247

Fig. 6. Microphotographs of thin sections of rock samples (bar scale = 1 mm). (A) Fine-grained equigranular leucogranodiorite without preferential orientation of feldspars. Several brittle cracks filled with fine-grained quartz are only preserved in the SW corner. Crossed nicols. (B) Strongly cataclastic to mylonitized granodiorite. More rigid K-feldspars, some fragmented grains of plagioclases and quartz are scattered in a fine-grained mylonitic matrix composed of fine-grained sericite and quartz. (C) Mylonite with scarcely preserved relics of partly recrystallized quartz grains. Due to strong fluid flux, feldspars are mostly decomposed to fine-grained sericite. Mylonitic matrix shows strong dynamofluidal texture. Crossed nicols. (D) Ultramylonite of original granitoid rocks with strong dynamofluidal texture and scarce relics of quartz sericitized feldspar porphyroclasts and mica fishes. Crossed nicols.

the local structural conditions in the study area is shown in Fig. 3 (compare e.g. with Kalafatovich, 1963; Ponce et al., 1999; Sassa et al., 2000; Caillaux et al., 2001). The figure shows three main fault zones and their strikes based on aerial photo interpretation and field structural measurements. The measured data from individual areas were plotted on the Lambert lower hemisphere projection plane for particular areas and, after that, for the whole area (Fig. 7). The most pronounced structure in the area is a NW−SE striking ductile shear zone proved after analysis (sampling sites 1 and 2 in Fig. 4), hereafter called “The Sun Gate shear zone”. The most deformed ultramylonites to phyllonites were found on the NE slope of Mt. Machupicchu. According to field measurements, this zone strikes around 300° and dips 60°–80° to the NE. The intensity of deformation rapidly decreases towards its borders, where the mylonites and mylonitized rocks pass into less

deformed cataclastic granitoids. The Sun Gate shear zone and associated small fractures are sub-parallel to the assumed Urubamba fault, which, according to an older interpretation (Carreño and Bonnard, 1997) runs across the lower part of the Front Slope. The ductile structures are accompanied by a system of densely spaced, steeply dipping joints in the same direction. Two sets of joints may be distinguished according to their dips: 1) a steep one (with dip of 70–90°) mostly dipping to the SSE, less to the NE; and 2) a moderately dipping joint system (with dip of 30–50°) with dip direction to the SW and NE. The latter system prevails in the less deformed domain of the so-called “quarry” (Fig. 4) and in the area to the NE of the archaeological site. Joint planes of this system are morphologically well pronounced on Mt. Huaynapicchu, as well as on individual outcrops situated in the upper and middle parts of the Front Slope. The steeply dipping joints and small faults are interpreted as following the strike of the

248

V. Vilímek et al. / Geomorphology 89 (2007) 241–257

tions and processes influencing this evolution may be put in the following hierarchical order:

Fig. 7. Density diagram of poles of joint and small fault planes in the area of Machu Picchu archaeological site (Lambert equal area projection on lower hemisphere). Multiples of random distribution: 4.00, 5.00, 6.00, 0.00, 1.00, 2.00, 3.00. 155 measurements. Contours at min. dens. = 0, max. dens. = 6.38 (at 338/24), contour interval = 1.

faults and joints that originated during the main Laramide compression at the Cretaceous/Palaeogene boundary. The moderately dipping joints are probably younger and represent a joint system associated with updoming of the area during the Quaternary period. The second major structures in the study area consist of the Huaynapicchu and Machupicchu faults (Fig. 3; names according to Kalafatovich, 1963). Both of them are brittle fault zones running NE−SW. These faults are coeval or slightly younger than the Sun Gate shear zone. They may also have originated during the Laramide orogenesis. Fault zones approximately perpendicular to the NW–SE thrust compensated for the unequivocal shortening in the major SW facing thrust zones and have been repeatedly active (in a brittle manner) till recent times. These fault zones are accompanied by relatively closely spaced fractures striking 40°–70° (Fig. 7), mostly dipping steeply to the SE, more rarely to the NW. This system prevails in the outcrops of the S and SE part of the archaeological site. 4.3. Landforms and their development Interpretation of aerial photographs and field investigation indicated that the evolution of the landscape in the Machu Picchu area reflect tectonic influence, slope movements and intensive fluvial erosion. The condi-

i) The formation of the Machu Picchu pluton followed by subsequent cycles of orogenesis (the Hercynian and later Andean ages) defined the main mechanical and structural properties of the rocks, and placed the Machu Picchu site almost into the middle of the regional scale structural feature called the “deflection de Abancay” (Carlotto et al., 1999). ii) Neotectonic uplift of the Central Andes area (Gregory-Wodzicki, 2000) increased the relief energy slope and erosion processes. iii) Exogenous processes including deep fluvial incision of the rivers, intensive slope movements of various types, and climatically influenced weathering (with infiltration water as an important agent diminishing the strength of the rock fractures and enhancing weathering; Jaboyedoff et al., 2004) formed the current topography. Precipitation, supplying both ground and surface water bodies, reaches an average of 1950 mm yr− 1 at present (Wright et al., 1997), but during certain periods in the Holocene it was even higher. Ramirez et al. (2003) conclude from their ice core analyses that the Andean climate during the Little Ice Age period was 20% more humid. Aside from that, during the Pleistocene/Holocene transition period, the water coming from melting glaciers of Nevado Verónica and Nevado Salcantay was yet another factor enhancing river erosion. All the processes and conditions mentioned above mutually contributed to the evolution of the suggested ancient (pre-settlement) slope deformations located between Mt. Machupicchu and Mt. Huaynapicchu (described below) as well as recent, more superficial landslides, rockslides, rock falls and debris flows, which are abundant in the study area (Fig. 2). The geomorphological map (Fig. 2) clearly reflects the tectonic influence on the evolution of the studied area. All the faults and joints mentioned above have made the rock mass highly tectonically disturbed, and therefore influenced development of landforms of various sizes. The main landscape features, including prominent ridges, the river network and position of rapids, largely follow the general structural plan of the area. Clearly identified borders of ancient large-scale slope deformations also correspond to these faults (Fig. 6). The effect of weathering on the formation of the micro- and meso-forms should not be omitted. It also contributed to a lowering of the stability of tectonically

V. Vilímek et al. / Geomorphology 89 (2007) 241–257

249

Fig. 8. Identified landslide zones in relation to the landscape development of the Front Slope. 1: Scarp of 1st generation of ancient slope deformation, 2: Body of the 1st generation of ancient slope deformation, 3: scarp of ancient rockfall of uncertain age, 4: accumulation of ancient rockfall debris of uncertain age, 5: scarp of the 2nd generation of ancient rock slope deformation, 6: body of the 2nd generation of ancient slope deformation, 7: accumulation of 2nd generation of ancient slope deformation, 8: recent slope deformation. 9: recent landslides (small for the scale). Dotted lines indicate slope profiles shown in Figs. 9 and 10.

loosened zones by alteration of the most susceptible minerals, leaving a weakened, quartz-dominated residue (Viles, 2003). Lack of fluvial sediments in the studied part of the Urubamba River valley illustrates intensive river erosion. The only important sediment storage areas are formed either by the youngest river terraces or alluvial cones, where material is deposited through mixed slope and fluvial processes of catastrophic character (e.g. debris flows). If more fluvial sediments were preserved, more detailed reconstruction of river incision and valley-side development including chronostratigraphical dating would be possible. An important feature in the immediate neighbourhood of the archaeological site is a large-scale slope deformation of unknown age shown in Figs. 8–10 (1st generation rockslide). This landslide (or landslides) could be caused by stream down-cutting as well as tectonic, structural and lithological conditions (i.e., weakening of granite rocks due to tectonic alternation). Its morphological manifestation is clearly visible in aerial photographs and also on the interpretation maps (Fig. 8) and derived profiles (Figs. 9 and 10). Its head scarp was situated above the present-day ridge between

the Machu Picchu site and Mt. Huaynapicchu. In the present relief we can find only remnants of this scarp. Its lateral walls are defined by the Machupicchu and Huaynapicchu faults shown in Fig. 3 (Vilímek et al., 2005). Another morphological manifestation of this huge slope deformation could be the double ridge, which was subsequently anthropogenously remodelled during the building of the Machu Picchu site. The Incas filled the central depression with sediments and soil, and converted it into the main square and on the ridges they built temples and residential/industrial buildings. The potential for further opening of this double ridge is being examined by ongoing extensometric measurements. We assume two possible triggers of a slope failure of this size — long-term climatic oscillations or an earthquake. Pinpointing the time of the earthquake is difficult, but in the case of the climatic triggers Trauth et al. (2003) have identified two phases of significant clustering of the landslide events in the southern Central Andes: 40–25 kyr and ca. 5 kyr BP (14C). They concluded that it is particularly the major fluctuations of yearly precipitation and temperatures that particularly contributed to the lowering of the stability of the mountain slopes.

250

V. Vilímek et al. / Geomorphology 89 (2007) 241–257

Fig. 9. Slope longitudinal profiles of the Front Slope at X–x, Y–y and Z–z, based on topographic data provided by SAN Peru. Their locations are shown in Fig. 8. The legend is the same as that in Fig. 8.

Subsequent stream down-cutting probably enabled the evolution of the 2nd generation of ancient rockslides (or a single rockslide) which involved only the central part of the Front Slope. The clear tectonic fragmentation of that part of the slope may have resulted in more favourable structural settings for rockslide evolution. We also suggest that a large rockslide or rock fall from the southwest slope of Mt. Putucusi might have forced the stream to move against the Front Slope, in similar fashion to what is now happening from the Mt. Machupicchu side. The 3rd generation of very recent slope processes is represented by debris-flows-like failures that carved deep gullies into older slopes and involved weathered or highly fractured surface materials. Other types of failures include rock falls (occurring mostly on the Back Slope), rockslides and soil slips. Occurrence of these movements may be partly explained by the steepening effect of the Bingham Road cut. They are also caused by excessive pore water pressures in slope sediments or cracks during the rainy season. Their repeated occurrence was proved by interpreting a time series of aerial photographs, literature (Carreño and Bonnard, 1997) and field mapping, and often contributed to talus deposition. The longitudinal profile of the Urubamba River clearly shows that, besides the source region, the steepest gradient of the stream is found in the neighbourhood of Machu Picchu (Fig. 11A). This significant increase of gradient may be explained, among other possibilities, by

a neotectonic activity (Hartvich, 2005). A more detailed longitudinal profile of the Machu Picchu meander (Fig. 11B) shows that most of the river sections with significantly steeper gradients are associated with mapped fault zones crossing the Urubamba valley. The first of the three most pronounced gradient anomalies (Fig. 11B) is located under Machu Picchu Village (beginning of the meander, see Fig. 1) and the second is at the place where the Huaynapicchu fault crosses the Urubamba valley. Here a significant rock fall from Mt. Putucusi has contributed to the creation of rapids. The third rapids are formed where the Machupicchu fault crosses the Urubamba valley under the Back Slope. According to the general geological map of this area (Ponce et al., 1999), lithological influence for such morphological steps in the longitudinal profile of the Urubamba River can be excluded. The presumed impact of tectonics and prehistoric seismic activity is supported by many observations of tectonic rock “polishing” with striations (e.g., at the main entrance of the Machu Picchu site and below the summit of Mt. Huaynapicchu), which were found on granite outcrops where the influence of slope movements can be excluded. The majority of the slickensides have a subhorizontal dip, which excludes their gravitational origin. Chaotically distributed huge boulders and rock blocks (some larger than 5 m, Fig. 2) on the top part of the ridge between the mountains of Machupicchu and

V. Vilímek et al. / Geomorphology 89 (2007) 241–257

251

Fig. 10. Slope transversal profiles of the Front Slope at A–a, B–b, C–c, D–d, based on topographic data provided by SAN Peru. Their locations are shown in Fig. 8. The legend is the same as that in Fig. 8.

Huaynapicchu and on the summit of Huaynapicchu are often separated by open fissures, which in several cases, formed caves with an estimated depth of more than 8 m (Fig. 2). These rock blocks may be evidence of prehistoric seismic events or may be remnants of the 1st generation rockslide described above. 4.4. Dilatometric and extensometric monitoring Positions of all the monitored localities are depicted in Fig. 5, and examples of two time series plotted into

charts are shown in Fig. 12 (dilatometer) and Fig. 13 (extensometer). Results of the most important measurements are given in Tables 2 and 3. Because of the limited time of the measurements, only indicators of the existence of irreversible displacements are given there. The indicator scaling from “very weak” to “strong” shows a certainty of assessment of presence, as well as the rate of those movements with regards to noise, which obscures them within a raw monitoring signal. Raw monitoring time series represent a mixture of displacements of different origins. The irreversible

252

V. Vilímek et al. / Geomorphology 89 (2007) 241–257

Fig. 12. Example of a graph of a dilatometric measurement. Site Huaynapicchu, measurement W3. See Fig. 5 for location. Black filled squares indicate control measurements (taken with more accurate device).

Fig. 11. Longitudinal profiles of the Urubamba River. (A) Vertically exaggerated (×100) profile along whole Urubamba River. Dotted rectangle indicates the position of the detailed section in (B) (modified after Peñaherrera and Ibérico, 1986). (B) Vertically exaggerated (× 25) profile along a 30 km long part of the Urubamba River near the Machu Picchu site. Steeper segments connected to passage of the faults (dashed lines).

displacement by slope movement is therefore obscured by noise from all the other displacement items within the signal. Reversible changes of volumes of rock blocks following circadial, seasonal, or even higher quasicycles of air temperature changes are the most common and the most powerful noise producing phenomena. These problems, however, do not look serious in our measurements of total displacement over 2 to 3 years. A phenomenological model by Zvelebil (1995, 1996), related to the influence of main obscuring phenomena on movement patterns of slope movements with differing magnitudes, was used to assess slope movement activity within the monitoring time series and the sensitivity and probable error range of each monitoring method. In general, due to the presence of climaticallydriven seasonal quasi-cycles of rock volume changes, longer time series provides better information. Then irreversible displacement resulted from the accumulation of many subtle events can be detected.

Regarding the deep-seated slope movement hypothesis (Sassa et al., 2000), the measuring sites on flanks or even across the double ridge depression zone are of the main interest. T1-3 sites (Fig. 5, Table 2), just behind the main Temple, exhibited well-pronounced indicators of irreversible movements ranging from 2.0 to 4.7 mm during 2.7 years. Also at the opposite side of the tension double-ridge zone, a displacement of − 3.5 mm (distance reduction) over 2.7 years was indicated at the P site. Also the southern extensometric profile (ES1 and ES2) gave a medium indication of irreversible increase of the distance, 4 mm over 2.2 years, whilst the northern EN1 profile showed only a weaker irreversible decrease of − 2.0 mm during 2.2 years. To assess the possible association of movements from individual measuring sites, and localities with the hypothetical, deep-seated deformation, mutual coupling

Fig. 13. Extensometric measurement EN1 crossing the northern part of the main square (Plaza). See Fig. 5 for location. Black filled dots indicate control measurements (taken with more accurate device).

V. Vilímek et al. / Geomorphology 89 (2007) 241–257 Table 2 Magnitudes and indicators of irreversible trends of movements detected on dilatometric and extensometric measurement sites on flanks and across the double ridge zone in the Machu Picchu archeological site Site

Indicator

Irreversible displacement (mm)

Years of monitoring

Level of noise in the signal

I1 I2 I3 T1 T2 T3 P A1 ES 1 + 2 EN 1

Medium Very weak Weak Strong Strong Strong Strong Nill Medium Weak

−1 − 0.25 − 0.5 − 4.75 −2 −3 − 3.5 0 4 −2

2.5 2.5 2.5 2.75 2.75 2.75 2.75 2.75 2.2 2.2

Medium High High Low Low Low Low High Medium Medium

For explanation of site abbreviations see Fig. 5.

of those movements were also studied using the graphical method of correlograms (e.g. Zvelebil et al., 2006). We propose that if the hypothetical landslide is deeper and larger, a higher degree of coupling should be shown in signals from different measuring sites, even for those substantially spaced from each other. The following sequence was obtained according to the relative coupling degree. As the standards for maximal coupling, the correlograms of time series from the sites placed on the same crack could be taken (e.g. I1–I2 and T2–T3). The medium to weak signs of coupling were shown by measurements from boundary cracks of the same rock block (e.g. T1–T3 and T1–T2). Time series obtained at different cracks but within the same rock mass of the Intiwatana hill (I in Fig. 5) did not show the medium signs of mutual dependence. Nevertheless, for the sub-horizontally oriented measurements at I1–I3, there were still some observable regular patterns. Comparison of time series from different localities, representing different parts of the slope rock mass, provided even fewer correlation patterns. Occurrence of very weak signs of inverse proportion could be speculated for the horizontal displacements within the neighbouring rock outcrops at sites I1 and T2. Those outcrops are supposed to be placed along the same zone of structural weakening of the slope rock mass. However, with the exception of a few micro-events of signal coupling probably due to the same responses of rock mass to temperature changes, there were no correlation patterns between dilatometric measurements from rock mass at the opposite flanks of the tension zone of the double-ridge (e.g. T1, T2 or T3 vs. P). At the Back Slope, irreversible movements with magnitudes of 3.0 mm during 2.5 years and 1.25 mm

253

during 2.75 years were indicated at the Cave and Quata sites, respectively (Fig. 5; Table 3). The slope movement activity at the Back Slope is important for our considerations of rock movement kinematics at the extensometric EN1 profile. Indications of irreversible movements within the block system at the edge of a high, nearly vertical rock wall near the summit of Mt. Huaynapicchu (W1–3 in Table 3) should be cautiously considered with regards to the safety of tourists and preservation of the Inca ruins. Their magnitudes of − 2.75 mm during 2.75 years, and 1.2 mm during 1 year (quite consistent in its magnitude with the former) are too large to be neglected with regards to the overall stability of that block system. The magnitudes of all the irreversible displacements indicated by dilatometric monitoring range from 0.5 to 3.5 mm over 2.5–3.7 years. Compared to characteristic magnitudes of surface movements recorded at the sites of large, deep-seated slope movements (e.g. Moser et al., 2002; Moser, 2003), the magnitudes at Machu Picchu are at least one-order smaller and do not prove the existence of deep-seated slope movement at the site. After inspecting the monitoring sites established by Sassa et al. (2000) and re-evaluating their rather alarming results, we suspect that the deformations they observed include significant “noise” from more superficial movements occurring in the upper part of highly disaggregated and disturbed colluvial deposits. 4.5. Hydrogeological surveys The hydrogeological research aimed at investigating the potential contribution of hydrogeological conditions Table 3 Magnitudes and indicators of irreversible trends of movements detected on dilatometric and extensometric measurement sites away from the double ridge zone in the Machu Picchu archeological site Site Indicator

Irreversible Years of Level of noise displacement (mm) monitoring in the signal

C W1 W2 W3 R1 R2 R3 M1 M2 M3 M4 Q1

−3 0.75 1.2 − 2.75 −6 – 0.5 4.5 − 3.5 0 − 0.5 1.25

Strong Medium Weak Strong Very weak Not included Weak Very weak Very weak Very weak Weak Medium

2.5 2.5 1 2.75 1.2 1.2 2.7 2 2.5 2 2.5 2.75

For explanation of site abbreviations see Fig. 5.

Low Medium Low Low High High High High High High Medium

254

V. Vilímek et al. / Geomorphology 89 (2007) 241–257

to the instability of the slopes. Aside from the “Inca spring” (yield of 0.4 l s − 1 , Fig. 4), only weak manifestations of runoff of underground water of shallow circulation were observed at the Bingham Road cut (with a cumulative flow rate up to 1 l s− 1 at the time of the survey). The yield of individual talus springs varied from 0.1 to 0.2 l s− 1 with only two cases yielding up to 0.4 l s− 1. The survey was conducted in a relatively dry period (September). According to local information, the yield of the springs is five times greater in the rainy season (December to March), suggesting that groundwater circulation in the slope sediments is fairly fast, with a residence time of several days to several weeks. The deeper ground water circulation looks insignificant, because only slight water leakage was apparent in the railway tunnel at the foothill of Mt. Putucusi and in a large open crack below the Machu Picchu site. Based on these findings we assume that infiltration into ground water amounts to up to 10% of the total rainfall (cf. 18% according to Carreño and Kalafatovich, 2006), while the rest is drained by surface runoff or lost by evapotranspiration. It is also suggested that the vast majority of the ground water is drained by concealed discharge into the Urubamba River through slope and fluvial sediments. As noted, water supplied to the historic channel system within the Machu Picchu site is artificially

brought from the neighbouring Huinay Huayna basin. At the time of our survey, its flow rate into the basin was 1 l s− 1, from which only 0.15 l s− 1 was discharged into the historic water system. It was found that 80% of the supply (0.12 l s− 1) was lost on its way across the centre of the historical monument. This means 12 960 l of water leakage per day, which has the potential of saturating 4000 m3 of soil in a month if effective porosity is assumed to be 10%. This amount of water may substantially alter the water budget of the historical monument during the dry season. The water leakage may be responsible for terrace deformation in the lowest part of the archaeological site. A relationship between the wall damage and the above-average ground water supply is suggested by vividly green grass found only in that place. Our hydrological tests suggest a low to medium permeability of the soil, with the coefficient of filtration ranging from 10− 3 to 10− 6 m s− 1. It is also evident from the tests that the forest soil covered with humus is the most permeable, whereas anthropogenous loams with grass cover, forming the surface of the squares and terraces of the archaeological site, have low permeability. Weathered granite underlying the surface cover is highly permeable. Stony colluvial deposits also constitute the main infiltration body due to its loose structure as well as high porosity and permeability (Fig. 14).

Fig. 14. Map showing the thickness of colluvium and hydrological conditions of the Front Slope. 1: thin colluvium, 2: continuous colluvium, 3: deep colluvium (more than 3 m), 4: permanent streams, 5: preferred superficial flows, 6: gully, 7: river, 8: contour lines (50 m interval). IC: Inca City of Machu Picchu.

V. Vilímek et al. / Geomorphology 89 (2007) 241–257

4.6. Evaluation of landslide risk Now we shall discuss the significance of the obtained results in relation to our research aim including the five questions posed in Introduction. Different types of slope deformations that vary in relative age, size and type of movement were described through interpretation of aerial photos and field investigation. The oldest, largescale deformation identified on the Front Slope is evident from the morphology because it occurred within the limits of the Machupicchu and Huaynapicchu faults. Its age is uncertain, but it may be related to the sliding periods defined by Trauth et al. (2003): 40–25 kyr and 5 kyr BP. The longitudinal profile of the Urubamba River may also be used as a relative age indicator. The large-scale 1st generation slope deformation on the Front Slope (Figs. 8 and 9) did not affect the gradient of the Urubamba River, in contrast to the rock fall from Mt. Putucusi that contributed to the gradient increase and the formation of the rapids. Since fluvial erosion has already removed most of the supplied sediment, it may be concluded that the 1st generation Front Slope deformation is old. The 2nd generation of landslides evolved in the area of the Bingham Road before the construction of the archaeological site, and is influenced by structural conditions and river down-cutting. The corresponding landslide deposits are partly deposited at the toe, on the left riverbank of the Urubamba River. Recent slope movement is associated with shallow landslides (debris flows, rock falls, rockslides or soil slips) affecting weathered or highly fractured surface materials of the Front and Back Slopes. These landslides may endanger the archaeological site, by damaging the retaining walls of the terraces. These terrace walls are the most vulnerable to slope movement and also the most important for the stability of the terraces themselves. Such possible damage is being prevented by recent management practices of the responsible institution, which regularly checks the most vulnerable parts of the Machu Picchu site and performs mitigation work when needed. Even though the vicinity of Machu Picchu is not among the most active seismic zones (based on data from the last 30 years), there are records of recent seismicity in its neighbourhood (USGS, 2006). Within 100 km of the Machu Picchu site, the catalogue shows one earthquake of magnitude 7; eight of magnitude 5–6, and 19 of magnitude 4–5. The strongest earthquake occurred on July 6, 1991, and its epicentre was situated approximately 40 km to the east of the site. Furthermore, indirect signs of recent tectonic activity are apparent in the landscape: intensive erosion

255

in the Torontoy canyon and its tributaries; very steep slopes along tectonic lines; the steep gradient of the segment of the longitudinal profile of the Urubamba River below Machu Picchu (the Torontoy Canyon); rapids in the lithologically homogenous riverbed of the Urubamba River correlated with faults crossing the valley; and disintegrated huge blocks on the ridge between the mountains of Machupicchu and Huaynapicchu. This last evidence may reflect a large-scale gravitational deformation causing mountain ridge spreading (Mahr and Nemcok, 1977; Radbruch-Hall et al., 1977; Dramis and Soriso-Valvo, 1994; Lotter et al., 1998), which may have been triggered by a strong earthquake. This may also have resulted from enhanced river incision in the highly fractured rocks of the Machu Picchu area, in contrast to the other parts of the Torontoy canyon with high, steep and relatively stable slopes. Based on the above geomorphological investigations, we conclude that only a future earthquake could trigger a large, rapid slope failure with possible catastrophic effects on the Machu Picchu site. Unfortunately, no relevant data regarding earthquake probabilities for the Machu Picchu area are available. Another possible danger to the site is loss of vegetation cover (e.g. by wild fire), enhancing slope processes including surface wash, gully erosion and shallow landsliding, resulting in higher sediment mobility and possible damage to roads, trails and some Inca structures. Based on spring yields and hydrological evaluation of the Machu Picchu basin, we assume that the effective infiltration is 10%. We consider this value more reasonable, since the high mean annual rainfall total of 1950 mm (Wright et al., 1997) gives ground water outflow of 6.32 l s− 1 km− 2 and a reasonable discharge rate of 22.8 l s− 1 for the observed catchment area (3.6 km2). This input into underground water circulation has a direct influence on the activity and development of landslides (Schuster, 1978; Ondrášik and Rybář, 1991), particularly in the case of shallow ones. These are represented in the area of interest by the 3rd generation soil slips and resultant debris flows. A direct link between the precipitation and the development of shallow landslides and debris flows, based on an example in 2004, is reported elsewhere (Vilímek et al., 2006). The opposite directions of movements at the extensometric profiles crossing the double ridge zone (ES1 and ES2) can be explained by several individual local slope deformations, not by deep-seated slope movements, because the correlograms indicate no coupling of these movements. The decreasing trend of deformation with distance at the EN1 profile is also

256

V. Vilímek et al. / Geomorphology 89 (2007) 241–257

consistent with “back-toppling” within the individual systems of blocks forming the upper rim of the Back Slope. Such local movements are supported by the relatively widespread occurrences of deformed landforms. Moreover, the present-day activity of this type of slope movement was also detected by dilatometric measurements at the Cave site situated in the southern part of the Back Slope. The same kinematics seem to be a plausible explanation for movements recorded within the rock block system at W1–3 sites (Fig. 5), where the tourist path goes through a fissure cave near the summit of Mt. Huaynapicchu. Petrological analyses show the strong influence of tectonics on the character of the rocks around the site. The faults form weak zones, where the weathering is the most active and the potential sliding planes or sidelimit of deep-seated movements might develop. In particular, the presence of a ductile shear zone with green schist facies, in the granites of the Machu Picchu pluton along the NE side of the Machu Picchu site, is very important. This weak zone, associated with faults and dense spaced cleavages, is a potential source of instability, especially where steep NE dipping schistosity is cut by moderately dipping faults or joint planes of similar strikes and by NE–SW transversal faults and fractures dividing the rock massif into separate blocks. These blocks may have been moved both horizontally and vertically as was suggested by Kalafatovich (1963), who described the tectonic structure of the Machu Picchu Ridge as a graben between Mt. Machupicchu and Mt. Huaynapicchu. This inference, however, is not supported by the rather sub-horizontal striations with a strike of ENE−WSW and NW−SE observed in the study area. In summary, the results of the petrological investigations do not explain the present active deformation of the slope, although they do confirm the fact that tectonic and structural conditions are suitable for the development of deep-seated landslides.

2) Detected movements may be caused by local gravitational movements of individual rock blocks (e.g. C, W, and Q in Fig. 5) or several other mechanisms including the sinking of archaeological structures, subsurface erosion and annual changes in the water content of the soils. 3) None of the irreversible deformations detected by our monitoring could be considered as an indicator of immediate danger of collapse of the Machu Picchu site. From the point of view of short-term slope stability, we did not find any significant danger threatening the archaeological site. However, if we consider the longterm stability conditions of the site, we must be aware of the fact that the whole area is located in an unstable area because of structural preconditions, the relatively high seismic activity of the broader surroundings and intensive fluvial erosion possibly enhanced by neotectonic uplift. Therefore, we suggest that the research in the area should continue, particularly in the following directions: 1) absolute and relative dating of landslides (e.g. 10Be dating and Schmidt hammer tests); 2) heavy minerals and provenance analyses of terrace and floodplain material to pinpoint the sliding events in relation to river valley development; and 3) maintaining the existing monitoring network, with the possibility of implementing new measurements such as monitoring soil water content. Acknowledgements The authors would like to thank the Ministry of Education, Youth and Sports of the Czech Republic (Projects MSM 00216 20831 and INGO, LA 157) for their financial support. Special thanks are due to INC Cusco and INRENA Cusco for scientific and personal support. References

5. Conclusion The results of the dilatometric and extensometric monitoring, hydrogeological research, geomorphological mapping, and petrologic/structural analyses in the Machu Picchu area have led to the following general conclusions: 1) Recent fast activity of large-scale deep-seated slope movement, suggested by Sassa et al. (2001), was not confirmed by our monitoring and therefore is doubtful.

Caillaux, V.C., Bustamante, J.G., Mora, J.S., Valencia, J.A., Olivera, P. M., Pérez, E.M., 2001. MS, Plan para la Mitigación de Desastres del Poblado de Machu Picchu – Aguas Calientes. Programa de las Naciones Unidas para el Desarrollo, Cusco. 178 pp. Carlotto, V., Cárdenas, J., Romero, D., Valdivia, W., Tintaya, D., 1999. Geología de los cuadrángulos de Quillabamba y Machupicchu, Hojas: 26-q y 27-q. INGEMMET, Lima. Carlotto, V., Galdos, B., Solís, M., Arias, V., Machicao, O., Mar, P., 2001. Plan para la mitigacion de desastres del poblado de Machu Picchu - Aguas Calientes. MS PROFONAPE-Programa Machu Picchu, Cuzco, Peru. Carreño, R., Bonnard, C., 1997. Rockslide at Machupicchu, Peru. Landslide News 10, 15–17.

V. Vilímek et al. / Geomorphology 89 (2007) 241–257 Carreño, R., Kalafatovich, C., 2006. The Alcamayo and Cedrobamba catastrophic debris flow (January, March and April 2004) in Machupicchu area, Peru. Landslides 3, 79–83. Carreño, R., Lopéz, R., Tupayachi, L., 1996. Informe preliminary sobre el derrube de Machupicchu del 26.12.95 y situación del cerro Putucusi. Nota técnica, MS, PROEPTI, Cusco, Peru. Dramis, F., Soriso-Valvo, M., 1994. Deep-seated gravitational slope deformations, related landslides, and tectonics. Engineering Geology 38, 231–243. Gregory-Wodzicki, K.M., 2000. Uplift history of the Central and Northern Andes: a review. GSA Bulletin 112, 1091–1105. Hartvich, F., 2005. Analysis of the longitudinal profiles of selected rivers in the Šumava Mts., Southwest Bohemia. Miscellanea Geographica 11/2005, 13–30. Jaboyedoff, M., Baillifard, F., Bardou, E., Girod, F., 2004. The effect of weathering on Alpine rock instability. Quarterly Journal of Engineering Geology and Hydrogeology 37, 95–103. Kalafatovich, C., 1957. La glaciación pleistocena en Urubamba. Revista Universitaria 46 (112), 235–252. Kalafatovich, C., 1963. Geología del conjunto arquelógico de Machupicchu y sus alrededores. Revista Universitaria UNSAAC, vol. 121. Lotter, M., Moser, M., Glawe, U., Zvelebil, J., 1998. Parameters and kinematic processes of spreading of mountain ridges. Proc. 8th Int. IAEG Congr., II. A.A. Balkema, Rotterdam, pp. 1251–1257. Mahr, T., Nemcok, A., 1977. Deep-seated creep deformations in the crystalline core of the Tatra Mountains. Bulletin of IAEG 16, 104–106. McEvan, G., Wright, K.R., 2001. Machu Picchu will endure. New Science. April 3 issue. Moser, M., 2003. Geotechnical measurements for assessment of the kinematic and long-term behaviour of unstable mountainsides. In: Natau, O., Fecker, E., Pimentel, E. (Eds.), Geotechnical Measurements and Modelling — Proceedings of the International Symposium on Geotechnical Measurements and Modelling. A.A. Balkema, Lisse, pp. 217–223. Moser, M., Meier, H., Lotter, M., Weidner, S., 2002. Geotechnical aspects of deep-seated mass movements in Austria. In: Rybář, J., Stemberk, J., Wagner, J., et al. (Eds.), Landslides-Proceedings of the 1st European Conference on Landslides. Swets & Zeitlinger, Lisse, pp. 423–430. Ondrášik, R., Rybář, J., 1991. Dynamic Engineering Geology. SPN, Bratislava. 267 pp. Peñaherrera, C., Ibérico, M., 1986. Geografía Física del Peru, Geología del Peru. Ediciones Manfer, Barcelona. Ponce, C.F., Solano, P., Suárez, G., 1999. Plan Maestro del Santuario Histórico de Machupichu. MS, Instituto Nacional de Recursos Naturales, Instituto Nacional de Cultura, Lima, Peru. Radbruch-Hall, D.H., Varnes, D.J., Colton, R.B., 1977. Gravitational spreading of steep-sided ridges (“sackung”) in Colorado. Journal of Research of the U.S. Geological Survey 5, 359–363. Ramirez, E., Holmann, G., Taupin, J.D, Francou, B., Einstein, P., Caillon, N., Ferron, F.A., Landais, A., Petit, J.R., Pouyaud, B., 2003. A new Andean deep ice core from Nevado Illimani (6350 m), Bolivia. Earth and Planetary Science Letters 212, 337–350. Sassa, K., 2001. The second investigation report on the slope instability in Inca's World Heritage in Machupicchu, Cusco,

257

Peru. MS DPRI Kyoto University — Report for INRENA, Kyoto, Japan. Sassa, K., Fukuoka, H., Shuzui, H., 2000. Field Investigation of the slope instability at Inca's World Heritage, in Machupicchu, Peru. Landslide News 13, 37–41. Sassa, K., Fukuoka, H., Kamai, T., Shuzui, H., 2001. Landslide risk at Inca's World Heritage in Machu Picchu, Peru. Proceedings of UNESCO/IGCP Symposium — Landslide Risk Mitigation and Protection of Cultural and Natural Heritage. UNESCO/IGCP, Tokyo, Japan, pp. 1–14. Schuster, R.L., 1978. Introduction to landslides: analysis and control. In: Schuster, R.L., Krizek, J.R. (Eds.), Landslides: Analysis and Control. Transportation Research Board, Washington D.C., pp. 1–10. Trauth, M.H., Bookhagen, B., Marwan, N., Strecker, M.R., 2003. Multiple landslide clusters record Quaternary climate changes in the northwestern Argentine Andes. Palaeogeography, Palaeoclimatology, Palaeoecology 194, 109–121. USGS, 2006. Online earthquake catalogue. Earthquake Hazard Program. http://neic.usgs.gov/neis/epic/epic_rect.html. Viles, H.A., 2003. Rock weathering at Machu Picchu. MS — Preliminary Report. University of Oxford, Oxford, UK. Vilímek, V., Zvelebil, J., 2002. Slope instability at Machu Picchu: ideas and questions. Acta Montana 19, 75–89. Vilímek, V., Zvelebil, J., Klimeš, J., Vlčko, J., Astete, F., 2005. Geomorphological investigations at Machu Picchu, Peru. In: Sassa, K., Fukuoka, H., Wang, F., Wang, G. (Eds.), Risk Analyses and Sustainable Disaster Management. Springer, Berlin, pp. 49–56. Vilímek, V., Klimeš, J., Vlčko, J., Carreño, R., 2006. Catastrophic debris flows near Machu Picchu village (Aguas Calientes), Peru. Environmental Geology 50, 1041–1052. Vlčko, J., 2004. Extremely slow slope movements influencing the stability of Spis Castle, UNESCO site. Landslides 1 (1), 67–71. Wright, K.R., Witt, G.D., Zegarra, A.V., 1997. Hydrogeology and paleohydrology of ancient Machu Picchu. Ground Water 35, 660–666. Wright, K.R., Zegarra, A.V., Wright, R.M., Mcevan, G., 2000. Machu Picchu, A Civil Engineering Marvel. ASCE Press, Reston, Virginia. Zvelebil, J., 1995. Determination of characteristic features of slope movements present-day activity by monitoring in thick-bedded sandstones of the Bohemian Cretaceous Basin. Acta Universitatis Carolinae. Geographica, Supplementum, Prague 79–113. Zvelebil, J., 1996. A conceptional phenomenological model to stability interpretation of dilatometric data from rock slope monitoring. In: Seneset, K. (Ed.), Landslides. Proc. VII-th Int. Symp. on Landslides. Balkema, Rotterdam, pp. 1473–1480. Zvelebil, J., Stemberk, J., 2000. Slope monitoring applied to rock fall management in NW Bohemia. In: Bromhead, E., et al. (Ed.), Landslides in Research, Theory and Practice. Proc. 8th Int. Symp. on Landslides, vol. 3. Thomas Telford, London, pp. 1659–1664. Zvelebil, J., Paluš, M., Novotná, D., 2006. Nonlinear science issues in the dynamics of unstable rock slopes: new tools for rock fall risk assessment and early warnings. In: Cello, G., Malamud, B. (Eds.), Fractal Analysis for Natural Hazards. Geological Society of London Special Publications, vol. 261, pp. 79–93.