Geographical Information Systems in Assessing Natural Hazards

The geographical and temporal patterns of disasters are first described and then considered in terms of the underpinnings and causes of human vulnerability. These include population increase, marginalization, the militarization of vulnerable societies, the politicization of aid, the accumulation of capital goods, and the dual role of technology as a source of both vulnerability and mitigation. Some of the bases of theory in hazards studies are reviewed and considered in the light of the development gap in mitigation — the wide gulf between the vulnerability of industrialized and least developed countries. The phenomenon of disaster is considered theoretically in terms of its fundamental dimensions: time, space, magnitude and intensity. Finally, the various disciplinary contributions to disaster studies are assessed and compared. Reasons are given for practitioners’ reluctance to undertake interdisciplinary work.

This paper considers the use and development of geographical information systems (GIS) in helping to reduce the impact of natural hazards. After a brief survey of the diversity of such hazards, an attempt is made to review what has been written in the past, a task made difficult by the wide range of interests involved. The review shows that, within the GIS field proper, relatively little has been published and that, within the disciplines studying natural hazards, few papers describe operational systems that are applied routinely, four examples of which are summarised. The limitations of existing GIS are then considered, with particular reference to the availability and deficiencies of the data on which hazard mitigation must depend, the limited functionality of current GIS, the failure to consider adequately the needs of inexpert users, and infrastructural deficiencies. Ways in which these limitations could be removed are then discussed, with particular reference to the needs of developing countries.

Geographical Information Systems (GIS) are becoming a fundamental tool in many spatial data applications, such as land planning and natural hazards assessment. A procedure for generating and handling hydrological basins data was developed, based on a set of modules integrated with a commercial GIS. The GIS and the other software modules make a complete system for processing the spatial entities and their attributes. Several types of spatial data are involved: elevation data, digitized polygons, which represent land attributes, and automatically generated drainage-divide networks and watersheds. The paper brings spatial data representation into focus and highlights the need for efficient and versatile conversion tools between raster and vector representations, in order to free users to handle their spatial data in different formats.

Deterministic models are based on physical laws of conservation of mass, energy or momentum. In the case of deterministic landslide hazard zonation, distributed hydrological and slope stability programs are used to calculate the spatial distribution of groundwater levels, pore pressures and safety factors. This paper is concentrated on the integration of two-dimensional, raster-based, geographic information systems (GIS) and deterministic models, with emphasis on deterministic hydrological models. Three examples of deterministic landslide hazard zonation are presented; one from Costa Rica and two from Colombia. In the example from Costa Rica, a one- dimensional external hydrological model is used to calculate the height of perched water tables in the upper metre of the soil for different soil types and different rainstorms. In the first example from Colombia, an external two-dimensional hydrological model is used to calculate the maximum groundwater level, for a 20 year period, in different slopes with a sequence of volcanic ashes overlying impermeable residual soils. In the second example from Colombia, a three-dimensional hydrologic model is used in a GIS to simulate groundwater fluctuations during one rainy season. In examples 1 and 2 the results of the hydrologic calculations are used in stability calculations to obtain maps which give the spatial distribution of safety factors and the probability of failure, with the use of distribution functions of the input parameters. In example 3 the calculated groundwater levels are exported to an external slope stability model to calculate the safety factor along slope profiles.

In the context of studies about the sediment supply into mountain torrents a Geographical Information System was used as a tool for the simulation of rapidly moving gravity-driven slope processes. It was found that procedures which use regular polygon-cascading as a ‘trajectory model’, sometimes do not calculate realistic paths for slope processes. The reason for this was found in the predecessor — successor relation these procedures establish between two adjacent polygons (triangles of a TIN or grid-cells). These relations turned out to be not transitive, when they do not take into account, that material transported into the considered polygon from a higher polygon is not always homogeniously distributed over the whole area of the considered polygon. Consequently, a new ‘trajectory model’, the ‘vector tree’ model, was developed and tested. In this model the paths are formed by’ sequences of vectors’.’ sequences of vectors’ are established by linking together vectors drawn parallel to the flow direction of the slope. This model has produced realistic results for the areas evaluated.

We describe two different GIS-based approaches to the delineation of debris-flow hazard. The first is empirically based, and uses logistic regression to predict sites of rainfall-induced shallow landslides that initiate debris flows in San Mateo County, California. The second is both empirically and process based, and uses multiple physically-based simulations of debris flows to evaluate hazard downslope from initiation sites in Honolulu, Hawaii. The two approaches use fundamentally different data to delineate hazard in fundamentally different ways, and are described here as contrasting examples of approaches to hazard delineation.

Based on several layers of spatial map patterns, multivariate regression methods have been developed for the construction of landslide hazard maps. The method proposed in this paper assumes that future landslides can be predicted by the statistical relationships established between the past landslides and the spatial data set of map patterns. The application of multivariate regression techniques for delineating landslide hazard areas runs into two critical problems using GIS (geographic information systems): (i) the need to handle thematic data; and (ii) the sample unit for the observations. To overcome the first problem related to the thematic data, favourability function approaches or dummy variable techniques can be used.

This paper deals with the second problem related to the sample units. In this situation, the unique condition subareas are defined where each subarea contains a unique combination of the map patterns. Weighted least squares techniques are proposed for the zonation of landslide hazard using those unique condition subareas. The traditional pixel-based multivariate regression model becomes a special case of the proposed weighted regression model based on the unique condition subareas. This model can be directly applied to vector-based GIS data without the need of rasterization.

A case study from a region in central Colombia is used to illustrate the methodologies discussed in this paper. To evaluate the results adequately, it was pretended that the time of the study was the year 1960 and that all the spatial data available in 1960 were compiled including the distribution of the past landslides occurred prior to that year. The statistical analyses performed are based on these pre-1960 data about rapid debris avalanches. The prediction was then compared with the distribution of the landslides which occurred during the period 1960–1980.

In the recent years, the ever-increasing diffusion of GIS technology has facilitated the application of quantitative techniques in landslide hazard assessment. Today a wider spectrum of instability causal factors, mainly morphological and geological in nature, can be cost-effectively acquired, stored and analysed in digital form. In particular, by processing elevation data and its derivatives new morphometric parameters can be readily generated over wide regions, and used as predictors of landslide occurrence. Despite the potential of such technological advancements, landslide hazard mapping remains a major, largely unsolved task. The identification and mapping of past and present landslide bodies, which constitute fundamental steps for predicting future slope-failures, remain highly subjective. Likewise, many basic instability determinants cannot be acquired and mapped with adequate accuracy. Most of the current methods for manipulating instability factors and evaluating hazard levels remain error-prone or questionable.

The experience gained from the application of multivariate models in small drainage basins, located in southern and central Italy, indicates that the type of terrain-unit selected, namely: grid-cell, unique-condition unit and slope-unit, exerts a relevant influence on the reliability and feasibility of the hazard model developed. Models, based on different types of terrain-units or statistical approaches, yield responses that may be statistically comparable but dissimilar in terms of applicability. In addition, when different landslide types occur over a region, each type requires the development of a specific model.

Among the different techniques traditionally applied, multivariate approaches, although with limitations, are the most feasible and cost-effective for evaluating the landslide hazard on a regional scale. This is true if GIS techniques are fully and carefully exploited for data acquisition, processing and analysis.

An application of Geographical Information Systems (GIS) for flood mapping in flat areas is described. On this basis impacts on agricultural practice can be assessed. Flood mapping is achieved with a two dimensional hydraulic model supported by a suitable Digital Terrain Model (DTM). The GIS framework includes: dBaseIV as the database management system, MapInfo as the vector oriented support and IDRISI for the raster component. The methodology is illustrated by a case study for a part of the Basse Broye river flood plain in Switzerland.

A review of some recent applications of “hydrologically oriented” GIS in flood forecasting and related issues is presented. A regional approach to the assessment of flood hazard in orographically accentuated areas is also described on the basis of the supporting data handling capabilities of Geographical Information Systems. In particular, the development of operational methodologies aimed at the issuing of effective and timely warnings at the regional scale is presented both in a deterministic perspective — by simple enumeration of a series of pre-defined targets — and in a probabilistic one, based on the most recent advances in drainage geomorphology. The use of multisensor monitoring devices for the detection of extreme events, from the overall meteorological scenario down to the scale of the local variability of rainfall in space, is proposed on the basis of the experience of some international research projects recently activated in the field. The multi-faceted role of GIS in the collection of rainfall data from the whole set of available sensors, the identification of areas of potential occurrence of extreme meteorological events, the operational support to distributed rainfall-runoff models and the mapping of flood prone areas and landscape vulnerability is highlighted throughout the paper.

An information system named GEO and its related instrumental environment are described. The technology implemented in GEO integrates the geographic information technology with case-based and knowledge-based system technologies. An example of a GIS application to seismic hazard assessment for the region which is formed by the Lesser Caucasus and by the Eastern part of the Great Caucasus is discussed.

Vector-based GIS computer systems are used with digital geologic maps of the active volcanoes of Kilauea and Mauna Loa, Hawai‵i to assess the probability that activity from those volcanoes might adversely affect the wells, power plants, and transmission lines of the Hawaii Geothermal Project. For each of the geothermal subzones, the probability of inundation by lava flows is 64–65%. For the three possible transmission line routes, the proposed has an 86% probability, the first alternate has a 63% probability, and the second alternate has a 28% probability of lava flow inundation. On the basis of this analysis, the second alternate transmission line route is the least likely to be damaged by a lava flow.

We present two applications of GIS technology to groundwater studies aimed at the assessment of pollution risk. The first study were carried out over the entire Po Plain, an highly developed area that depends entirely on groundwater for water supply; the second regarded a small portion of it. The research aimed at producing maps of the factors controlling groundwater-pollution risk to support decision-making for an effective land planning and management. The proposed risk-assessment procedure considers factors related to natural vulnerability, hazard due to human activities, and costs resulting from pollution. To evaluate existing hazard, a census was carried out over a test area. Industrial, agricultural and quarrying sites were mapped, and industrial activities were divided into four groups according to national waste type regulations. A land information system was used to store and manage the vast amount of spatial data and to prepare thematic output maps.

This paper reviews some of the factors relevant to the implementation of various GIS facilities with respect to terrain-related hazards prevailing in Hong Kong, with particular reference to landslides. Examples of the current state of practice are described, together with some considerations for future developments with regard to landslide hazard assessment and emergency response. Although Hong Kong faces the combined hazards of landslides, flooding, tidal surge and typhoon winds, the emphasis of this paper is placed on the approach to reducing landslide risks to the public. The approach is discussed within the framework of hazard mitigation (prevention and preparedness), disaster response and to a lesser extent, recovery. The key programmes within the Geotechnical Engineering Office of the Hong Kong Government are given as case studies. These systems continue to be developed to be more responsive to the pressing needs of urban development and to an increasing community awareness of the hazards of slope failure.

San Mateo County, California adjoining San Francisco has several hazardous geologic processes, including movement of the San Andreas fault and associated ground failures, landsliding, flooding, and coastal erosion. In the early 1970’s, county planners requested help from the U.S. Geological Survey in determining the location and severity of the most hazardous processes, and in preparing maps that would help them deal with the problems. Maps were prepared showing the location of fault zones and the most recently active fault traces, areas subject to flooding, and an inventory of past and recently active landslides. Geographic Information Systems (GIS) techniques were used to make a map of landslide susceptibility, but the analysis was done by hand.

San Mateo County used these hazard maps to enact ordinances that limit the density of development permitted in geologically hazardous areas to as little as one dwelling unit per 16 hectares, and that require geologic reports and review by the County Geologist before any development is permitted. During the past two decades, these ordinances have been expanded in area and strengthened in application.

The availability of this geologic-hazard data base, the innovative use of this information by the County, and the development of computer systems capable of manipulating large data sets and preparing color-separation negatives quickly, inexpensively, and with correct registration prompted a small group of U.S. Geological Survey scientists, engineers, cartographers, geographers, and computer specialists to consider a new set of hazard maps using computer-based GIS technology. Several new maps were prepared including slope and shaded relief maps prepared from a Digital Elevation Model; a map showing the direction and amount of bedding dip; a map showing the seismic-shaking intensities in a repeat of the 1906 San Francisco earthquake; maps showing cumulative damage potential to wood-frame, tilt-up concrete and steel buildings from earthquake ground shaking; a map showing where earthquake-triggered landslides will impact the county during a repeat of the 1906 earthquake; maps showing debris-flow probability; and a map of liquefaction susceptibility. These maps have been used by the county for general planning, to evaluate environmental impact reports, for the design of public facilities, for water supply and quality, and for land capability.

The National Weather Service (NWS) maintains the National Operational Hydrologic Remote Sensing Center, based in Minneapolis, to generate operational hydrologic products from in situ and remotely sensed snow cover data sets. The Center maintains an Airborne Snow Survey Program, a Satellite Hydrology Program, and a Snow Estimation and Updating Program. In all three programs, the Center relies heavily on the use of multiple Geographic Information Systems (GIS) to process, analyze, and distribute spatial snow cover data sets. Real-time, airborne snow water equivalent data and satellite areal extent of snow cover data are used operationally by the National Weather Service, the U.S. Army Corps of Engineers and other Federal, state, and private agencies when issuing spring flood outlooks, water supply outlooks, river and flood forecasts, and reservoir inflow forecasts. The remotely sensed and interpolated, gridded, snow water equivalent data products are generated by hydrologists in the Minneapolis office and distributed electronically, in near real-time, to NWS and non-NWS end-users in both alphanumeric and graphic format. The reliable, real-time, snow water equivalent information is critical to water managers and disaster emergency services officials who are required to make decisions with regard to snow melt flooding, reservoir regulation, and water supply allocation.