Topographic controls of landslides in Rio de Janeiro: field evidence and modeling

Please download to get full document.

View again

of 19
All materials on our website are shared by users. If you have any questions about copyright issues, please report us to resolve them. We are always happy to assist you.
Information Report



Views: 0 | Pages: 19

Extension: PDF | Download: 0

Related documents
Catena 55 (2004) Topographic controls of landslides in Rio de Janeiro: field evidence and modeling Nelson F. Fernandes a, *, Renato F. Guimarães b, Roberto A.T. Gomes
Catena 55 (2004) Topographic controls of landslides in Rio de Janeiro: field evidence and modeling Nelson F. Fernandes a, *, Renato F. Guimarães b, Roberto A.T. Gomes a, Bianca C. Vieira a, David R. Montgomery c, Harvey Greenberg c a Departamento de Geografia, Universidade Federal do Rio de Janeiro, Rio de Janeiro, , Brazil b Departamento de Geografia, Universidade de Brasília, Brasília, , Brazil c Earth and Space Sciences, University of Washington, Seattle, WA, USA Abstract Landslides are common features in the Serra do Mar, located along the southeastern Brazilian coast, most of them associated with intense summer storms, specially on the soil-mantled steep hillslopes around Rio de Janeiro city, where the favelas (slums) proliferated during the last few decades. On February 1996, hundreds of landslides took place in city of Rio de Janeiro triggered by intense rainstorms. Since then, many studies have been carried out in two experimental river basins in order to investigate the role played by the topographic attributes in controlling the spatial distribution of landslides inside them. Landslide scars and vegetation cover were mapped using aerial photographs and field observations. A detailed digital terrain model (4 m 2 resolution) of the basins was generated from which the main topographic attributes were analyzed, producing maps for slope, hillslope form, contributing area and hillslope orientation. By comparing these maps with the spatial distribution of the landslide scars for the 1996 event, a landslide potential index (LPI) for the many classes of the different topographic attributes was defined. At the same time, field experiments with the Guelph permeameter were carried out and a variety of scenarios were simulated with the SHALSTAB model, a process-based mathematical model for the topographic control on shallow landslides. The results suggest that most of the landslides triggered in the studied basins were strongly influenced by topography, while vegetation cover did affect landslide distribution. Between the topographic attributes, hillslope form and contributing area played a major role in controlling the spatial distribution of landslides. Therefore, any procedure to be used in this environment towards the definition of landslide hazards need to incorporate these topographic attributes. D 2003 Elsevier B.V. All rights reserved. Keywords: Landslide hazard; Hillslope hydrology; Mathematical modeling; Tropical soils * Corresponding author. Tel.: ; fax: address: (N.F. Fernandes) /$ - see front matter D 2003 Elsevier B.V. All rights reserved. doi: /s (03) 164 N.F. Fernandes et al. / Catena 55 (2004) Introduction Landslides are common processes in the Serra do Mar, a long coast range that follows almost all the way from south to southeastern Brazil, well-known by its steep scarps that lead to the generation of fantastic landforms like the sugar-loaves. In the major cities, particularly the city of Rio de Janeiro, these processes triggered by intense summer rainstorms, have become more frequent since the 1960s (Meis and Silva, 1968a,b; Barata, 1969; Costa Nunes, 1969; Jones, 1973), when the occupation quickly spread towards the steep hillslopes inside and around the city. Nowadays, many of the soil-mantled steep hillslopes are densely occupied (Fig. 1), especially by the slums (favelas), affecting slope stability by the extensive usage of cuts, landfills, deforestation, changes in drainage conditions, accumulation of trash deposits on hillslopes, among others, adding new relationships to the natural conditioning factors related to geology and geomorphology (e.g., Brunsden and Prior, 1984; Sidle et al., 1985; Crozier, 1986). However, along the steep slopes of the Serra do Mar escarpments in Southeastern Brazil, it is evident that landslides play a major role in controlling landscape evolution in the long term (e.g., Bigarella et al., 1965; Meis and Silva, 1968a; Modenesi, 1988; Lopes, 1997; Furian et al., 1999; Coelho Netto, 1999; Cruz, 2000). A recent landslide inventory of Rio de Janeiro city (Amaral and Palmeiro, 1997b), although covering just the period from 1962 to 1992, attests an increase in slope instability during the years of 1967, 1986 and 1988, directly related to intense summer rainstorms registered in those years. A future update in this inventory will certainly include 1996 in this list. From 1986 to 1996, landslide disasters killed 123 people and destroyed 414 houses in the city (Amaral, 1997). It is evident that the development of procedures able to carry out effective landslide prediction in such areas, although urged, is not easily achieved. Fig. 1. Spreading of occupation towards the hillslopes around Lagoa Rodrigo de Freitas, Rio de Janeiro city. N.F. Fernandes et al. / Catena 55 (2004) Intense summer rainstorms are the main landslide-triggering factor. Unfortunately, rainfall amounts in the order of mm in just 1 2 days are not rare in the city (Brandão, 1997). For example, in January 1966, a summer storm dumped about 480 mm rainfall over downtown Rio de Janeiro (about sea level) in just 3 days, while at stations placed at higher elevations (about 500 m), rainfall values got closer to 700 mm during this short period (Jones, 1973). More recently, in February 1988, about 384 mm rainfall were accumulated in the city during 4 days, with half of it registered during just one night (Brandão, 1997). Geological features associated with the Pre-Cambrian bedrocks (granites, biotite and plagioclase gneisses, migmatites, basic and alkaline dikes, etc.), including metamorphic foliation, unloading and tectonic fractures, also play a major role in landsliding within the city limits (e.g., Barata, 1969; Jones, 1973; Barros et al., 1988; Amaral et al., 1992; Barroso, 1992; Amaral and Palmeiro, 1997a). Unloading fractures, for example, control the downward migration of the weathering processes (differential weathering), resulting in zones of similar weathering inside the thick weathering mantles, attested in many places by abrupt soil-bedrock boundaries. Frequently, these fractures can be clearly observed within the city, representing important mechanical and hydrological discontinuities. Slope deposits are also directly associated with landslides in Rio de Janeiro. On the foothills of the escarpments, as well as in most of the topographic hollows, highly heterogeneous talus deposits can be observed. In some hillslopes, usually associated with saturated soils, these deposits can move very fast, attaining rates of 12.7 mm/day, as observed by Lacerda (1997) in the Tijuca massif. Colluvial deposits in the city, although thinner than the ones observed in other places in southeastern Brazil, may locally be important in defining critical places, particularly when these deposits are layered (Fernandes et al., 1994). The colluvial mantle along the hillslopes of the city has a high clay content, usually above 40%, overlaying a thick sandy-silt saprolite. As already suggested by Jones (1973), and later confirmed by many others, the hydraulic conductivity in this environment tends to be greater in the saprolite than in the colluvial mantle (e.g., Wolle and Hachich, 1989; Vieira and Fernandes, 2003). Such behaviour has important implications for defining the rupture mechanisms in these type of hillslopes, particularly those concerned either with the development of critical, positive pore-water pressures inside saturated soils or with the loss of soil suction and decrease in apparent cohesion inside unsaturated soils (e.g., Vargas et al., 1986; Wolle and Hachich, 1989; Campos et al., 1994). By comparing the location of landslide scars, associated with one single event or with a sequence of events, with the spatial distribution of the topographic attributes observed in the field, one can estimate the effect of each parameter on landsliding, as well as the triggering mechanisms (e.g., Gao, 1993; Larsen and Torres-Sanchez, 1998; Montgomery et al., 2000; Zhou et al., 2002). Besides, they improve our prediction abilities. These analyses have become more widely accessible due to increasing availability of highquality digital elevation models. In general terms, previous studies concerned with the role played by geomorphology in landsliding in southeastern Brazil have focused their attention mainly to the steepness parameter. For example, Cruz (1974) estimated threshold slope angles of about 22j (40%) 166 N.F. Fernandes et al. / Catena 55 (2004) for the Serra do Mar escarpments in Caraguatatuba, south of Rio de Janeiro in São Paulo State, by overlaying the spatial distribution of landslide scars with the map for slope angles. Similar values (from 20j to 29j) were later obtained by field experiments with a portable shear device in the same area (Cruz and Colangelo, 2000). Threshold angles have also been estimated in other places around the world. For example, Gao (1993) observed that the landslide potential increased rapidly for hillslopes above 31j in Virginia (USA), while Zhou et al. (2002) observed that most of the landslides in Hong Kong, following the 1993 event, took place on slope angles above 25j 30j. Although slope is a major topographic attribute, affecting both the hydrological conditions and stability analysis, its importance seems to have been overestimated in landslide hazard mapping procedures, particularly in Rio de Janeiro. Consequently, gentle hillslopes initially considered as having low landslide susceptibility were affected by landslides, especially debris-flows, during the 1996 rainstorms. This attests that other topographic parameters must be taken into consideration. In Southeastern Brazil, few studies have tried to consider the contribution of other morphological attributes on landsliding. Hillslope form, for example, although earlier suggested as an important parameter (Meis and Silva, 1968a,b), was not incorporated into stability analysis and landslide hazard mapping procedures. More recently, the role played by concave forms (hollows) has been intensively investigated, including their effects on surface and subsurface hillslope hydrology (e.g., Coelho Netto, 1985; Fernandes et al., 1994), on gullying (e.g., Coelho Netto et al., 1988; Coelho Netto and Fernandes, 1990; Coelho Netto, 1999), as well as on landsliding (e.g.; Lacerda, 1997; Guimarães et al., 1999; Guimarães, 2000; Fernandes et al., 2001). Another important topographic attribute is contributing area (per unit contour) since it defines the location of the convergent segments in a landscape. These portions are directly associated with the concentration of surface and subsurface flows, contributing to the development of soil saturation (e.g., Beven and Kirkby, 1979; O Loughlin, 1986) and many studies have attested the importance of contributing area in defining critical areas for landsliding (e.g., Dietrich et al., 1993, 1995; Montgomery and Dietrich, 1994; Wieczorek et al., 1997; Montgomery et al., 2000). The effective use of mathematical models for landslide prediction implies the understanding of the complex interactions between triggering mechanisms and conditioning factors. This knowledge requires, on the other hand, detailed field data concerning soil properties (cohesion, friction angle, specific weight, hydraulic conductivity, etc.) and hillslope hydrology, particularly the spatial and temporal variations in pore-water pressures (e.g., Anderson and Burt, 1978; Coelho Netto, 1985; Wilson and Dietrich, 1987; Harp et al., 1990; Fernandes et al., 1994; Montgomery et al., 1997; Gerscovich et al., 1997). In February 1996, hundreds of landslides were triggered in the city of Rio de Janeiro after a series of intense rainstorms, especially inside the Tijuca Massif (Fig. 2). In some parts of the city, during just 2 days, more than 350 mm of rain was recorded (GEORIO, 1996; Brandão, 1997; Coelho Netto, 1999). Most of the mass movements were shallow translation slides that converged towards the main valleys leading to catastrophic debris flows, which were able to move huge blocks and generated a great amount of destruction, including the loss of 222 houses and the death of 44 people (Amaral, 1997). N.F. Fernandes et al. / Catena 55 (2004) Fig. 2. Shallow landslides triggered by intense rainstorms on February 1996 on the southwestern flanks of the Tijuca Massif. The most catastrophic debris-flows triggered in this event, which alone destroyed about 150 houses (Amaral, 1997), were located at Quitite and Papagaio river basins, draining the southwestern flanks of the Tijuca Massif towards the Jacarepaguá lowlands (Fig. 3). In these two basins (5 km 2 ), around 100 landslide scars were mapped in the 1996 event and since then a series of studies have been carried out inside their limits (e.g., Vieira et al., 1997; Guimarães et al., 1999, 2003; Gomes, 2002; Vieira and Fernandes, 2003). In this study, we combined field evidences with a process-based mathematical model (Shallow Stability SHALSTAB) in order to investigate the spatial relationships between landslides and their conditioning factors in Rio de Janeiro city. The study was applied to the Quitite and Papagaio basins and emphasis was given to the role played by topography, vegetation cover, and land-use in controlling the spatial distribution of landslide scars during this event. The ability of the model to predict landsliding in these basins was evaluated by comparing the location of the predicted unstable areas with the location of the actual scars for the 1996 event. 2. Study area The two studied basins, Quitite and Papagaio, draining side by side the west flank of the Tijuca massif, have areas of 2.13 and 2.22 km 2, respectively, and about 100 landslide scars were mapped in the 1996 event (Fig. 4). Since then, a series of studies has been carried out in these basins (e.g., GEORIO, 1996; Vieira et al., 1997; Macias et al., 1997; Guimarães et al., 1999, 2003; Gomes, 2001; Vieira and Fernandes, 2003). Elevations decrease from 975 to 20 m in about 4 km, where the rivers start to flow within densely occupied lowlands of Jacarepaguá. In the upper portions of the basins, steep forested slopes prevail and thin soils are usually less than 2.0 m thick, while in the 168 N.F. Fernandes et al. / Catena 55 (2004) Fig. 3. Upper A and lower B portions of the Quitite debris-flow (Tijuca Massif) which destroyed many houses and other buildings. N.F. Fernandes et al. / Catena 55 (2004) Fig. 4. Location of two studied basins (Quitite and Papagaio) in the southwestern flank of the Tijuca Massif. This aerial photograph shows the landslide scars from the 1996 event (white) in the upper parts of the basins and the debris-flows in the lower ends. About 150 houses were partially destroyed in the area where the two debris-flows converge. middle portions part of the forest has already been substituted by grasslands. At that point in the basins, slopes become less steep and soil thickness frequently attains about m. In the lower portions of the basins, densely occupied nowadays, the channels have been modified and houses constructed over previous river beds and the valleys are filled with old debris-flow deposits. Drillings on these deposits showed that they can be thicker than 12.0 m (GEORIO, 1996), attesting the recurrence of debris-flow processes in this region. In general terms, bedrock is composed of a complex combination of Pre-Cambrian high-grade metamorphic rocks with granite intrusions (GEORIO, 1996; Coelho, 1997). The most frequent lithologic unit is a highly foliated banded gneiss, (Archer Gneiss). Although the foliation planes are variable, the predominant directions are between 000j/ 30j and 180j/30j (direction of the dip/dip angle). A medium texture, highly homogeneous granite intrudes the previous unit forming tabular levels with variable thickness, particularly in the upper portions of the basins. Different types of fracture sets are observed in the area, including unloading fractures and sub-vertical tectonic fractures with prevailing directions between 300j and 340j, and with a second set between 050j and 070j. Many dikes, mostly with a basic composition, are parallel to these directions and have affected landscape evolution by defining preferential sites for drainage incision. 170 N.F. Fernandes et al. / Catena 55 (2004) Methods 3.1. Field mappings and landslide scars Landslide scars were mapped in 1: aerial photographs, taken just 2 months after the 1996 event. Most of them were later visited in the field in order to define landslide types, soil characteristics, thickness of the failed material, soil depth and geology. Field mapping was also carried out to generate drainage and vegetation maps (Vieira et al., 1997, 1998). Detailed geologic mapping has also been conducted in some parts of the two basins (GEORIO, 1996; Coelho, 1997), but access to the steep and dense forested slopes in the upper portions is very difficult Topographic attributes and landslide potential indices A detailed digital elevation model, with a m grid, was generated for the two basins from the restitution of the aerial photographs described before. The spatial distribution of the investigated topographic attributes (slope, hillslope form, contributing area and hillslope orientation) was mapped using a geographic information system (GIS). These maps, together with the vegetation map, were overlaid with the landslide scar map, allowing the definition of the classes prevailing in each cell inside the landslide scars, for the four topographic variables, as well as for the vegetation cover. Frequency maps for the classes considered for the four topographic attributes were then produced. A landslide potential index (LPI) was obtained for each terrain variable by calculating the ratio between the number of cells disturbed by landslides and the total number of cells for that specific class. This index identifies the relative influence of each class on landsliding, for vegetation and for each one of the four topographic attributes investigated. More details concerning these procedures can be obtained in other studies (Vieira et al., 1998; Guimarães et al., 1999; Guimarães, 2000) Modeling landslide susceptibility We applied the model SHALSTAB (Dietrich and Montgomery, 1998) to the Quitite and Papagaio basins, simulating a variety of scenarios (Guimarães et al., 1999, 2003; Guimarães, 2000). Because this model simulates the topographic control on shallow landsliding, it was considered suitable to be used in the study area. During the numerical simulations, the topographic variables were obtained from the detailed digital elevation model of the basins while the soil properties were estimated from field investigations and from previous studies carried out in areas nearby. Although the model allows the incorporation of the spatial variability of soil properties, in this study, they were considered constant within the two basins. Other studies are being conducted in order to analyze these influences on landsliding. The SHALSTAB model is a deterministic mathematical model, which defines the relative susceptibility to shallow landsliding in a defined landscape. It has been developed since the early 1990s (Dietrich et al., 1992, 1993, 1995; Montgomery and Dietrich, 1994) and has been applied to many sites in the west coast of the United States (e.g., Montgomery et al., 1998, 2000; Dietrich and Sitar, 1997). This model assumes that although site specific properties control the size and the moment when shallow landslides are triggered, the main controlling factor defining their location is topography (Montgomery and Dietrich, 1994). Based on this assumption, SHALSTAB combines a steady state runoff model that estimates the topographically induced spatial variation in pore pressures with an infinite slope model for shallow landsliding (Dietrich and Sitar, 1997). The calculations incorporate, for each cell in the basin, variables associated with topograp
View more...
We Need Your Support
Thank you for visiting our website and your interest in our free products and services. We are nonprofit website to share and download documents. To the running of this website, we need your help to support us.

Thanks to everyone for your continued support.

No, Thanks