Instrumentos de recolección de la información

The Geographic Information System contains a set o f procedures which facilitate data input, storage, manipulation and analysis, and which output both spatial and attribute data to support decision-making activities (Grimshaw, 1994). It provides an excellent tool for landslide hazard zonation. The first application o f GIS to landslide hazard zonation took place in the late 1970s (Newman et al., 1978). In the 1980s, the application o f GIS to landslide mapping increased rapidly because o f fast developing computer technology and commercial GIS software packages. Over the past three decades, geoscientists have developed several approaches to landslide hazard analysis. The methods presented in the literature can be broadly classified into five basic approaches: direct mapping, heuristic approach (geomorphic analysis), statistical analysis, deterministic analysis, and probabilistic analysis.

Despite a lack o f agreement on methods and the scope o f research, all o f the methods are founded upon a basic concept (Carrara et al., 1995) which includes the following aspects: mapping o f landslides over a target region; identification and mapping o f a set o f causative factors which are correlated with slope instability; estimation o f the relative contribution o f these factors in generating slope-failures; and classification o f the land surface into domains o f different degrees o f hazard by different analysis methods.

Most o f the current hazard mapping aims to predict where failures are most likely to occur instead o f when they are likely to take place.

Landslide hazard mapping and assessment requires a preliminary selection o f a suitable mapping unit that refers to a portion o f the land surface which has a set o f ground conditions that differ from adjacent units across definable boundaries (Hansen, 1984).

Various methods have been proposed to partition terrain into mapping units for the purpose o f landslide hazard zonation (Carrara et al., 1995). The commonly used mapping

units can be classified into five groups: grid cells, terrain units, unique-condition units, slope units, and geomorphic units.

Guzzetti et al. (1999) summarized the definition o f each mapping unit. Grid cell, or pixel, is preferred by raster-based GIS users. The area is divided into regular squares o f a predefined size which become the mapping unit o f reference. Each grid is assigned a value for a particular factor (terrain parameters, geological, and land use, etc.). A stack o f raster map layers, each representing a single factor o f slope instability, is obtained for the grid based landslide hazard assessment. The grid format offers many advantages due to the simplicity o f operation through matrix algebra, and has been widely used by many researchers in heuristic, statistical, and deterministic analysis. Different analysis models have been developed to perform landslide hazard assessment based on the factor map layers.

Terrain units, which are traditionally preferred by geomorphologists, can be described as natural divisions o f the terrain that can be distinguished on aerial photographs: for example, bottom and summit areas, relative flatness and steepness in slope, convergent and divergent areas. Terrain units are the basis o f the land-system or land-unit classification approach, which has found wide application in many land resources investigations (Meijerink, 1988).

Unique condition units are constructed by the overlay o f different categorical maps, so each map unit is defined by a unique combination o f attributes. It implies the classification o f each slope-instability factor into a few significant classes, which are stored in a single map or layer. By sequentially overlaying all layers, homogeneous domains (Chung et al., 1995) are singled out whose number, size and nature are strictly dependent on the criteria used in classifying the input factors.

Slope units can be obtained by partitioning the terrain into hydrological regions between drainage and divides (Carrara, 1988). It can be considered as the left or right side o f a sub-basin o f any order into which a watershed can be partitioned. It can be automatically derived from a high resolution Digital Elevation Model (DEM). According to the prevailing slope failure mode and size, slope units can be resized by partitioning a

river basin into nested subdivisions, coarser for larger landslides and finer for smaller slope failures (Montgomery and Dietrich, 1994). Slope units can be further subdivided into geomorphic units.

Methods for landslide hazard assessment and zonation can be qualitative or quantitative, and direct or indirect. Qualitative methods are subjective in defining landslide hazard using descriptive terms. Quantitative methods produce numeric estimations such as factor o f safety or failure probability. Van Westen (1993, 2000), Van Westen et al. (1996, 1997), and Soeters and Van Westen (1999) discussed the main approaches in landslide hazard zonation and the scales o f mapping. The main methods and their characteristics in landslide hazard analysis are listed in Table 2.6.

Table 2.6 The main methods used in landslide hazard zonation (after Van Westen, 1993 , 2000)

Methods M ain Characteristics decision rules based on the experience o f the earth scientist.

Statistical analysis

Indirect methods in which statistical analyses are used to obtain predictions o f the mass movement hazard from a number o f parameter maps. Usually, the bivariate, multivariate, and fuzzy logic approaches are used.

Deterministic analysis

Indirect methods in which parameter maps are combined in calculations including slope stability analysis by geotechnical engineering methods and Newmark displacem ent analysis for seismic-induced slope stability assessment.

Probability analysis

Indirect methods in which the uncertainties in geotechnical parameters are considered in slope stability analysis by a probabilistic method such as Monte-Carlo simulation and the first order second moment approach. The failure probability or reliability index is used in landslide hazard assessment.

Geomorphic analysis is generally used for very large areas such as national hazard maps. The scale o f such maps can be on the order o f 1 TOO,000 to 1:250,000. This kind of

mapping is mostly used by large regional or national planning. The statistical analysis has the most flexibility in scale and in data type. It can also be used for medium scale mapping with scale o f 1: 25,000 to 1:50,000. This scale o f mapping can be used for infrastructure and transportation route planning. The deterministic and probabilistic analyses are used generally for local, regional, and site-specified hazard analyses such as land use planning for large engineering projects like dams, nuclear power plants, highways, and slopes o f open pit mines and spoils. The scales for local regional hazard assessment can be at 1:15,000 to 1:25,000. For detailed engineering study, the scale o f hazard mapping can be at 1:5,000 to 1:15,000. Such large scale mapping needs large scale digital topographic maps and geological maps, and/or high resolution aerial or satellite images (e.g. 1 to 4 m in spatial resolution).