Elsevier

Geomorphology

Volumes 169–170, 1 October 2012, Pages 30-44
Geomorphology

Mapping of risk and susceptibility of shallow-landslide in the city of São Paulo, Brazil

https://doi.org/10.1016/j.geomorph.2012.01.010Get rights and content

Abstract

In the city of São Paulo, where about 11 million people live, landslides and flooding occur frequently, especially during the summer. These landslides cause the destruction of houses and urban equipment, economic damage, and the loss of lives. The number of areas threatened by landslides has been increasing each year. The objective of this article is to analyze the probability of risk and susceptibility to shallow landslides in the Limoeiro River basin, which is located at the head of the Aricanduva River basin, one of the main hydrographic basins in the city of São Paulo. To map areas of risk, we created a cadastral survey form to evaluate landslide risk in the field. Risk was categorized into four levels based on natural and anthropogenic factors: R1 (low risk), R2 (average risk), R3 (high risk), and R4 (very high risk). To analyze susceptibility to shallow landslides, we used the SHALSTAB (Shallow Landsliding Stability) mathematical model and calculated the Distribution Frequency (DF) of the susceptibility classes for the entire basin. Finally, we performed a joint analysis of the average Risk Concentration (RC) and Risk Potential (RP). We mapped 14 risk sectors containing approximately 685 at-risk homes, more than half of which presented a high (R3) or very high (R4) probability of risk to the population. In the susceptibility map, 41% of the area was classified as stable and 20% as unconditionally unstable. Although the latter category accounted a smaller proportion of the total area, it contained a concentration (RC) of 41% of the mapped risk areas with a risk potential (RP) of 12%. We found that the locations of areas predicted to be unstable by the model coincided with the risk areas mapped in the field. This combination of methods can be applied to evaluate the risk of shallow landslides in densely populated areas and can assist public managers in defining areas that are unstable and inappropriate for occupation.

Introduction

Since 2007, half of the world's population has lived in urban areas, and the number of people living in urban areas may double in the next thirty years (Véron, 2007). Growing populations increase the vulnerability of cities, making landslide-risk management more complex, especially in developing countries, where the number of people who inhabit risk areas has grown by seventy to eighty million per year (UN, 2005).

The accelerated urbanization in developing countries has contributed to the construction of housing on steep hillsides, often using inadequate building standards, intensifying the occurrence of landslides and resulting in the expansion of risk areas. Additional factors have contributed to this expansion, including economic and social crises with long-term effects, historically inefficient policies for low-income housing, inefficient soil-use management, and a lack of appropriate legislation for the most susceptible areas and technical support for local populations (Ayala, 2002, Carvalho et al., 2007).

Wijkman and Timberlake (1985), Sidle et al. (1985), Anderson and Decker (1992), Alexander (1993), Amaral (1997), Ayala (2002), and others have verified that in developing countries, the impact of landslides is a major cause of loss of human life in densely populated urban areas, whereas in developed countries, it is primarily associated with economic losses. This pattern can be explained by greater prevention initiatives in developed countries. In the principal cities of poorer countries, landslides assume catastrophic proportions due to widespread cuts in embankments, garbage deposits, deforestation, changes in drainage networks, and other anthropogenic pressures without preventive planning (Brunsden and Prior, 1984, Sidle et al., 1985, Crozier, 1986, Fernandes et al., 2004).

In various languages, the concept of risk includes the possibility that a landslide (or another natural event) will cause significant social or economic damage to a particular population. Thus, risk areas are the places likely to be reached by natural or induced landslides. Susceptibility, in turn, is determined by the set of natural factors that may contribute to triggering these events. In other words, hillsides have a natural predisposition to landslides as a function of their geologic, topographic and climatic characteristics, among others. When sites with naturally high susceptibility are occupied in an inadequately managed manner (for example, by making cuts in the land and vegetation), there may be substantial risks to the population. Thus, landslide-associated accidents that cause significant social damage frequently occur in urban slums, forcing public managers to identify, analyze and manage risk areas to prevent accidents and plan for emergency situations (Arnould, 1976, Varnes, 1985, Augusto Filho, 1994, Wu et al., 1996, Cerri and Amaral, 1998, Fell et al., 1998, Guzzeti et al., 1999, Van Westen et al., 2003, Macedo et al., 2004, Remondo et al., 2008, Korup et al., 2010).

Several studies concerned with preventing and reducing risks associated with landslides have been performed. Some of the most important are the ZERMOS program (Zones Exposed to Soil Movement Risks), a risk-mapping system in France that aims to supply details about potential soil movement and slope instability in particular areas (Antoine, 1977, Humbert, 1977); the methodology developed by the Geotechnical Control Office (CGC), which develops maps to evaluate of land instability in the Hong Kong region (an area of high susceptibility to landslides) (Brand, 1988); and the International Decade for Natural Disaster Reduction (IDNDR; 1990–1999), which was instituted by the United Nations General Assembly in response to the increasing number of accident victims and economic damage due to natural disasters throughout the world (WP/WLI, 1990; UN, 2005). Additional methods have been proposed by Varnes (1985); Einstein, 1988, Einstein, 1997; Terlien et al. (1995); IUGS (1997); Wilson and Jaiko (1997); Hartlén and Viberg (1988); Jibson et al. (1998); Campbell et al. (1999).

In Brazil, the mapping of risk areas began around 1965 in the city of Rio de Janeiro, where slums had been located on elevated hillsides since the end of the nineteenth century. Since then, numerous landslide accidents have occurred. Various risk methodologies have subsequently been developed, implemented, and applied by research institutes and universities in other urban areas in Brazil. These projects include the use of risk maps in urban planning, the implementation of corrective and mitigating infrastructure in risk areas, the creation of emergency-response systems, the improvement of legislation dealing with land-use issues, and the public dissemination of information and training.

However, despite the innumerable studies that have been performed, Brazil currently faces a complex challenge. The number of risk areas has grown quickly, and public agencies generally do not possess the technical resources to address this problem. One of the principal management tools, an accurate map of risk areas, is difficult and costly to produce because of the many activities (such as field inspections) necessary to create it. Here, we propose a methodological tool that can be applied to the management of the landslide-risk areas. Traditional risk-mapping approaches could utilize prior knowledge of topographic instability through mathematical models based on physical processes (e.g., the SHALSTAB model). Most of these models evaluate susceptibility through a combination of stability models based on hydrological and infinite-slope theories, independent of previous landslide events. Such models have been widely disseminated in the international literature, especially since the 1990s, when the development and improvement of Geographic Information Systems (GIS) promoted the emergence of new approaches to identifying and evaluating unstable areas, minimizing costs, and facilitating the management of risk areas (Dietrich et al., 1992, Montgomery and Dietrich, 1994, Carrara et al., 1995, Wu and Sidle, 1995, Guzzeti et al., 1999, Iverson, 2000, Morrisey et al., 2001, Pack et al., 2001, Dhakal and Sidle, 2003, Van Westen et al., 2003, Van Westen et al., 2006, Calcaterra et al., 2004, Van Westen, 2004).

Existing methods must be strengthened because landslides are the natural events that cause the most casualties in urban areas in Brazil (Fig. 1). Landslides also cause major economic damage and block highways, especially during the summer, when the largest amounts of rainfall occur. Precariously situated urban areas are the result of a housing deficit of more than seven million homes throughout the country. A framework of social exclusion causes thousands of people to occupy inappropriate sites such as steep hillsides and floodplains of rivers, resulting in the presence of numerous slums. The intensive development of these slums began in Brazil during the 1960s, and they have expanded rapidly. The slums are characterized as illegal occupation of public or private land with inappropriately and densely arranged houses, encompassing areas that lack infrastructure and essential public services.

The state of São Paulo, where this study was performed, has the largest housing deficit in Brazil. While the state has only about 4 million houses, the state capital (the city of São Paulo) has about 11 million inhabitants in an area of a little more than 1500 km². From the 1960s to the 1990s, the city of São Paulo experienced a territorial growth of 40%, resulting in the removal of 31% of the vegetative cover. Forests were replaced with inadequately planned roads that cut into the deep valleys and hillsides, and many slums were built (Meyer et al., 2004). In 2004, 562 landslide-risk areas were identified in the city of São Paulo, of which more than 50% were evaluated as having high or very high rates of risk for the population (SVMA and IPT, 2004). Consequently, deaths and damage triggered by precipitation occur on unstable hillsides every year during the rainy period. In January 2010, for example, about 600 mm of rainfall was recorded in 27 days, triggering landslides that killed 60 people and left more than 20,000 homeless. This was the greatest amount of rainfall measured during the corresponding period in São Paulo since January 1947, when about 480 mm was recorded (IAG, 2010).

One of the most critical and largest basins in the city of São Paulo is that of the Aricanduva River, which is located in the eastern portion of the city and has an area of about 100 km². Significant flooding problems occur along the primary river course, which crosses an important avenue in that part of the city. The area is characterized by intense urbanization and the associated reduction in soil permeability. For years, catastrophic flooding events have been recorded during the summer due to rapid drainage of superficial rainwater to watercourses. The overflowing water volume reaches innumerable houses. Between December 2002 and March 2003, more than ten floods were recorded, caused by rainfall lasting between 30 and 120 minutes and amounts from 60 mm to 80 mm. These floods caused several economic and social damages (Canholi, 2005, IPT (Technology Research Institute of the São Paulo State), 2005).

The Aricanduva River basin has 22 tributaries, and numerous slums are built on steep hillsides or very close to drainage channels, especially in the upstream areas of the basin. Poorly executed embankments, road cuts, deforestation, and other anthropogenic pressures in this portion of the basin contribute to the production of sediments that are not retained in the alluvial plain because of the absence of a riparian zone. These sediments are transported downstream, aggravating the occurrence of silting and flooding. In other words, the drainage channels are blocked by debris and silting caused by erosion and landslides, resulting in the accumulation of garbage that obstructs the passage of water, further reducing outflow capacity.

We selected the Limoeiro River basin as a study area because of the frequent occurrence of landslides in the slums built on dissected topography there and the resulting socio-economic losses. The Limoeiro River is one of the headwater tributaries of the Aricanduva River basin and is located on the outskirts of the city of São Paulo. In January 2010, the Limoeiro River basin received heavy rainfall (103 mm in 24 hours), triggering shallow landslides that caused six deaths and considerable economic damage, destroying slums and forcing about 300 local residents to evacuate their homes. Moreover, the operation of public transportation was compromised, especially the circulation of trains and automobiles; water and electricity distribution were interrupted; and several parts of the basin were flooded. The general objective of this article is to analyze the probability of risk and susceptibility to shallow landslides in the Limoeiro River basin. Although the events described above were triggered by an extreme rainfall event in January 2010, the damaging soil-use practices and natural conditions in the basin may exacerbate landslides in this area.

Section snippets

Study area

The Limoeiro River basin has an area of 9 km² (Fig. 2) and runs through a rocky plateau called the Paulistano Plateau, which is located more than 700 m above the Serra do Mar escarpment along the coast of the state of São Paulo. The Paulistano Plateau contains a diverse set of metamorphic and igneous rocks from the Precambrian and Eopaleozoic ages, with various forms of topography.

In the study area, the most common types of mass movements are shallow landslides and creep, in addition to fluvial

Materials and methods

To analyze the probability of risk and susceptibility to shallow landslides, we used two distinct methodologies. In a direct field survey, we evaluated landslide risk based on natural and anthropogenic indicators. In an indirect approach, we verified susceptibility through the application of a deterministic mathematical model based on physical forces (Fig. 4).

In the direct analysis, areas of risk were mapped using a cadastral survey form containing a checklist of the triggering factors for

Map of landslide-risk areas

In mapping the areas of risk, four of the fourteen mapped probability sectors were classified as R1 (130 houses), two were classified as R2 (130 houses), two were classified as R3 (35 houses), and six were classified as R4 (390 houses). The areas of these sectors varied from 7000 m² to 75,800 m², and approximately 685 houses were counted within the total area (Fig. 9 and Table 3). Houses were counted in the field and confirmed from field and oblique aerial photographs. In some cases, the close

Discussion: prediction of susceptible areas associated with the probability of landslide risk

We verified that the SHALSTAB model, which has been widely used to predict landslides in areas with low anthropogenic intervention, produces satisfactory results that are consistent with landslide-scar maps. In dense urban areas, however, its use is still uncommon. For example, Dietrich et al. (1998) applied the model in seven hydrographic basins in the Oregon chain near Coos Bay, California (U.S.A.), where they recorded approximately 844 landslides between 1979 and 1996 and found that 94% of

Conclusions

Several Brazilian cities have areas of landslide risk due to the predominance of weak building standards and lack of infrastructure. In regions with severe rainy periods, these areas of risk are frequently subjected to accidents that result in death and economic damage. In many Brazilian cities, the weak environmental land-use legislation tends to be applied more rigorously in districts that are more economically and financially important. In contrast, areas that are farther from the city

Acknowledgements

This research and this paper (figures in print version) were supported by FAPESP (São Paulo State Research Foundation). The authors thank their colleagues for continuing support and discussion: Kátia Canil, Luis Antonio Bittar Venturi, Emerson Galvani, Cristiane Incau Pinto Pimentel, Tulius Dias Nery, Nívia Marcello, researchers of IPT (Technology Research Institute of the São Paulo State) and Civil-Defense of São Mateus District (São Paulo city). This manuscript was significantly improved by

References (79)

  • P. Antoine

    Réflexions sur la carthographie ZERMOS et bilan des expériences en cours

    Bulletin Bureau de Recherches Geologiques et Minieres

    (1977)
  • M. Arnould

    Geological hazards – insurance and legal and technical aspects

    Bulletin of the International Association of Engineering Geology

    (1976)
  • Augusto Filho, O., 1994. Cartas de risco de escorregamentos: uma proposta metodológica e sua aplicação no município de...
  • I.A. Ayala

    Geomorphology, natural hazards, vulnerability and prevention of natural disasters in developing countries

    Geomorphology

    (2002)
  • I.A. Ayala

    Hazard assessment of rainfall-induced landsliding in Mexico

    Geomorphology

    (2004)
  • K.J. Beven et al.

    A physically based, variable contributing area model of basin hydrology

    Bulletin of Hydrological Sciences

    (1979)
  • E.W. Brand

    Landslide risk assessment in Hong Kong

  • D. Calcaterra et al.

    Assessing shallow debris slide hazard in the Agnano Plain (Naples, Italy) using SINMAP, a physically based slope stability model

  • R.H. Campbell et al.

    Mapping time-dependent changes in soil-slip-debris-flow probability

  • A.P. Canholi

    Drenagem urbana e controle de enchentes

    (2005)
  • A. Carrara et al.

    Gis-Based techniques for mapping landslide hazard

  • C.S. Carvalho et al.

    Mapeamento de riscos em encostas e margem de rios

    (2007)
  • M. Casadei et al.

    Testing a model for predicting the timing and location of shallow landslide initiation in soil-mantled landscapes

    Earth Surface Processes and Landforms

    (2003)
  • L.E.S. Cerri et al.

    Riscos Geológicos

  • F. Cervi et al.

    Comparing predictive capability of statistical and deterministic methods for landslide susceptibility mapping: a case study in the northern Apennines (Reggio Emilia Province, Italy)

    Landslides, Online First

    (2010)
  • L. Claessens et al.

    DEM resolutions effects on shallow landslide hazard and soil redistribution modeling

    Earth Surface Processes and Landforms

    (2005)
  • M.J. Crozier

    Landslides: causes, consequences & environment

    (1986)
  • DAEE (Department of Water and Electrical Energy of São Paulo state)

    Plano diretor de Macrodrenagem da Bacia Hidrográfica do Alto Tietê – Bacia do Rio Aricanduva: diagnóstico geral e ações recomendadas

    (1999)
  • A.S. Dhakal et al.

    Long-term modelling of landslides for different forest managent pratices

    Earth Surface Processes and Landforms

    (2003)
  • W.E. Dietrich et al.

    SHALSTAB: A Digital Terrain Model for Mapping Shallow Landslide Potential

    NCASI (National Council of the Paper Industry for Air and Stream Improvement), Technical Report, Berkeley, February

    (1998)
  • W.E. Dietrich et al.

    Erosion thresholds and land surface morphology

    Geology

    (1992)
  • W.E. Dietrich et al.

    Analysis of erosion thresholds, channel networks and landscape morphology using a digital terrain model

    Journal of Geology

    (1993)
  • W.E. Dietrich et al.

    A process-based model for colluvial soil depth and shallow landsliding using digital elevation data

    Hydrological Processes

    (1995)
  • W.E. Dietrich et al.

    A validation study of the shallow slope stability model, SHALSTAB, in the forest lands of Northern California

    Stillwater Ecosystem

    (1998)
  • H.H. Einstein

    Landslide risk assessment procedure

  • H.H. Einstein

    Landslide risk – systematic approaches to assessment and management

  • R. Fell et al.

    Guidelines for landslide susceptibility, hazard and risk zoning for land use planning

    Engineering Geology

    (1998)
  • F. Guzzeti et al.

    Landslides hazard evaluation: a review of current techniques and their application in a multi-scale study, Central Italy

    Geomorphology

    (1999)
  • Cited by (44)

    • Evaluation of shallow landslide susceptibility and Factor of Safety variation using the TRIGRS model, Serra do Mar Mountain Range, Brazil

      2021, Journal of South American Earth Sciences
      Citation Excerpt :

      The existence of large cities located in areas susceptible to landslides induced by rain and even by earthquakes requires the conduction of investigations that seek to identify the triggering mechanisms, as well as the distribution thereof in space and time, and to fill gaps with the new studies, especially, in South America (Gariano and Guzzetti, 2016). However, the socio-economic condition of most South American countries is still a great obstacle to the evaluation of susceptibility and risk (Alcántara-Ayala, 2002; Listo and Vieira, 2012), since the real-time monitoring systems are rare, either to obtain field data or connected to warning systems. Although field data are fundamental for a more appropriate performance of the models, they are still deemed sparse for the mountainous regions in South America.

    • Urbanization, economic growth and environmental pollution: Evidence from China

      2019, Sustainable Computing: Informatics and Systems
      Citation Excerpt :

      Moomaw and Shatter found that the urbanization rate increased with the growth of per capita GDP, industrialization, and other factors, and decreased with the increase of the level of agricultural production [28]. The environmental pollution caused by urbanization is mainly due to the effects, such as the increase of urban land use, of the excessive concentration of population [29–31]. The destruction of the environment through economic growth is mainly due to excessive consumption of energy.

    • The influence of land urbanization on landslides: An empirical estimation based on Chinese provincial panel data

      2017, Science of the Total Environment
      Citation Excerpt :

      3) This paper studied whether landslide prevention measures are improved with the increase of economic development levels between different regions in a country. Other studies (Listo and Carvalho Vieira, 2012; Mandasari et al., 2016; Smyth and Royle, 2000) argued that landslide prevention measures improve with the increase in the economic development level between different countries. However, whether the improvement exists between different regions in the country requires validation because landslide prevention measures in all regions are unified by the central government to a large extent, for example, ROTPOGD is applicable to all provinces, autonomous regions and municipalities in China.

    View all citing articles on Scopus
    View full text