Institute of Industrial Science, The University of Tokyo, 4-6-1 Komaba, Meguro-ku, Tokyo 158-8505, Japan
Disaster reduction requires the integration of information from diverse sources. Geographic Information Systems (GIS) is one new key technology often used to collect, store, analyse and display large amount of spatially distributed information layers. The core of a GIS is a set of spatially referenced maps, which are stored either as points, lines, polygons or raster data. GIS makes it easy to assign attributes to these spatial quantities and combine different layers of information.
Keywords: GIS technology; Disaster reduction
Natural disasters are the outcome of many complex geophysical characteristics and the related social circumstances that are subjected to a hazard. The hazards may be meteorological in origin such as cyclones, severe storms, droughts and blizzards, or may be earth processes such as earthquake, volcanic eruptions, tsunamis, etc., or a combination of both as in the case of floods. All these events are location dependent in the sense that a hazard is aggravated by the geological, topographical and land cover at the location of the hazard. Similarly, natural hazards turn into disasters when they affect societies. The degree of damage is dependent on the population density, infrastructure and means available for mitigation such as flood control dams and evacuation facilities. In order to grasp the impact of different disasters, it is necessary to understand the interactions and inter-relationships among these diverse and complex entities subjected to a given magnitude of the hazardous event.
The strength of GIS lies in the ability to represent the real world situation closely with layers of information (maps) that can be combined in a predetermined manner to identify the impacts of a natural hazard through the introduction of hazard dimension. In the case of floods, the hazard information is represented as water height, velocity and the flood duration distribution over the catchment. Combining this information with population distribution helps identify people at risk, with road network shows available or passable roads for evacuation and relief, with hospitals and emergency facilities in planning response and relief and with the property distribution in estimating damage. In the case of earthquakes, this information could be ground shaking intensities due to an earthquake, which again can be combined with population, housing and infrastructure information to assess disaster impact and plan response and relief strategies.
GIS has developed substantially over the past decade with the advent of large-volume data handling capabilities that facilitates synthesising information from many different data sources. It has become an indispensable tool for managing complex information related to both societal and environmental functions.
Disaster reduction discipline has benefited largely from these developments in risk map preparation, damage assessment and modelling for forecasting and planning.
The next section briefly describes the current status and use of GIS in disaster reduction
Identifying the risk from natural disasters is an important requirement for
mitigation and preparedness. The automation provided by GIS could be directly
used in microzonation, as the basic information fusion process involving
comparison, indices and overlaying in microzonation is the same for basic GIS
operations. Depending on the type of disaster, there are varying approaches for
preparing risk maps. For example, in the case of floods, the possible hazard
scenarios are more deterministic than for earthquakes or cyclone disasters. In
this case, hazard maps for different occurrence probabilities are prepared
first. This could be, for example, flooding extent due to rainfall intensities
with different return periods produced by scenarios such as embankment
overtopping, embankment collapse, etc. This information is then combined with
population and infrastructure distribution to prepare risk maps.
Another approach is the vulnerability analysis, where the hazard potential is considered to be equally distributed regionally. This approach is adopted in earthquake microzonation where each location is subjected to the same type of ground motion and vulnerability is assessed based on the geological structure of each location. In either approach, there are many uncertainties in the assumptions related to hazard scenario as well as in the regional physical characteristics and the forecasting models employed in the hazard estimation. GIS is effective in carrying out such analysis as automated processes within the GIS, and different outcomes resulting form changed input parameters, assumptions and scenarios can be easily compared with due consideration given to uncertainties in methodology and the input data.
In the case of flood risk map preparation, first the flood extent estimate is prepared, either from historical data or through numerical simulation of selected extreme events. If numerical simulation is to be carried out, data other than the meteorological input data is necessary to supply a host of information related to the locality depending on the type of hydrological and flood plain model used. At a minimum, it is necessary to have the elevation information to determine the surface gradients, and the land cover information to estimate the surface roughness of the catchment together with the physical characteristics of the river. The set up of a distributed hydrologic model is schematically shown in Figure 1. The topmost layer represents a mathematical model in which the catchment is divided into a large number of grids. While the smaller grid sizes allow higher accuracy, computational capacity generally dictates the feasible extent of a grid unit. The mathematical model computes the water flow in the catchment resulting from rainfall using equations that govern the water flow in and out of each grid, and routes the flow along slopes, to rivers and then along the rivers. In order to use the water flow equations, the properties of the location such as slope, roughness, infiltration capacity, etc., are required. These properties are assumed to be uniform within each grid and are estimated from GIS layers of elevation, land cover and soil property. Figure 2 shows the results of such a simulation where flooding occurs due to over-topping of the embankments during a heavy rainfall. Once the flood extent for a given frequency has been established, the risk can be estimated by overlaying the population distribution and infrastructure information on the inundation map. By considering rainfall corresponding to different return periods, flood extent frequency maps are prepared and the risk to people and assets subjected to flooding can be established.
Damage estimation for a potential hazard is a key parameter in designing
mitigation measures. While it is difficult to estimate intangible damage such as
injuries, or anxiety in a purely deterministic manner, there are GIS systems
that are currently available or are being developed to estimate both primary and
secondary tangible damage. The methodology widely followed is to establish
fragility functions for different types of property, such as residential and
non-residential buildings, infrastructure, crops, farms, etc., which express the
potential damage as a percentage of cost under a particular type of hazard,
given by the depth and duration of water height in case of floods or ground
shaking intensity in case of earthquakes. By overlaying the map of hazard level
onto the property distribution map, the damage estimation can be carried out
either in an external program or within the GIS depending on the complexity of
fragility functions and the damage estimation model.
Figure 3 schematically shows the procedure to estimate the potential flood damage by the above-mentioned procedure. At first, a hydrological model is used to estimate the inundation depth and duration at each grid for the expected scenario. If the damage due to a past flood is to be estimated, then the observed flood extent maps are converted to a GIS layer so that floodwater height and flood duration at each grid point is known. Once this hazard information is available, the damage in various categories are estimated. For example, one may distinguish between damage to business and industry, damage to residential housing, damage to crops and farmhouses, etc. For each of these categories, it is necessary to have what are called fragility functions or damage functions that relate the flood height and duration information into economic loss. These curves are prepared from the past flood data for different residential building categories as wooden, concrete, industrial building types, etc. In order to utilize all the information contained in the fragility functions, information layers that describe such property distribution in each grid are required. As most of these data are not readily available, GIS analysis is used for creating them from auxiliary data. As an example, if one knows the number of food industry complexes located in an administrative area, their distribution can be precisely made in appropriate grids, if the locations are known. However, if locations are not available, then land cover information and road network data may be used to distribute the industries within the administrative units by considering only highly urbanized land cover areas located within a certain distance to the roads. Once the hazard dimension, i.e. flood height and the property distribution, is prepared, then the total damage in each grid is estimated using a fragility function for each property type identified within each grid subjected to the water level in that grid. Integrating this information over the inundated area provides the total loss resulting from the event. Figure 4 shows the economic loss distribution estimated following such a procedure.
Mathematical simulation and modelling disaster processes is the main procedure in forecasting or warning, as well as in impact assessment. Unfortunately, present day GIS cannot handle time-varying information or dynamic updates of information required in the modelling of disasters. However, GIS is widely used to prepare the input information to mathematical models as a pre-processor. The ability of GIS to process complex spatial information as input data has in turn helped to produce more complex models capable of representing the disaster scenarios in more detail.
4.1. GIS in Planning
GIS has become a prominent tool for city and infrastructure planning in general. In the disaster mitigation field, several uses in land management, response and reconstruction are reported. In more specific examples of usage, GIS is a tool in securing lifelines such as gas, water and electricity. Up-to-date GIS systems specifying the state of lifelines is a must for identifying and isolating network components that have been rendered inoperable during disasters. GIS has been heavily used in the recovery process in the aftermath of a disaster. GIS had been used to identify locations for evacuation and resettling soon after the disaster. Demolition and reconstruction have been greatly assisted by GIS technology in the aftermath of the Northridge and Kobe earthquakes. GIS has also been used in selecting locations for resettlement after a disaster.
4.2. GIS in Response
For disaster response, it is necessary to identify passable roads, locations
of emergency services, refugee camps, feasible transportation routes and a host
of other information which can only be derived by combining the most recent
disaster area status information with other static information related to the
In the past, GIS-based response systems have been successful in the firefighting industry, where the spread of fires and fire conditions can be generalised and the systems are used continuously. During major disasters, the chaos and the rapidly changing situations have prevented the use of GIS in response, although many response systems have been developed or are under development. The recent rapid advancement and easy access to GPS and mobile communication are now leading the development of real-time innovative response systems. As a typical example to illustrate the methodologies, consider an accident where a chemical is released into the air from an industrial complex. With the information on the quantity and time of the chemical release, the diffusion process can be modelled using the most recent information on wind velocities in the area coupled with the information on buildings and other infrastructure in the vicinity. Once the spread and timing is known from the modelling study, disaster managers are alerted to the situation with information on expected risk and occurrence time. Their response can be made more effective by supplying information on people at risk in the target area such as homes for elderly, schools, etc. In order to make such systems operational, it is necessary to have easily accessible infrastructure and demographic databases in GIS available to a range of responsible organizations. Ideally, the development of such systems should aim at integrating diverse information from more than one source, with a forecasting system to arrive at possible consequences. Then this information has to be disseminated to persons responsible for responding to the situation. Hence, the coupling of GIS with communication and modelling tools are a necessity in using GIS for disaster response.
4.3. Overview of GIS Utilisation
GIS technology is helpful in integrating data from various sources in all phases of a disaster cycle. GIS information, especially, can be easily combined with detailed land cover information obtainable from remote sensing, thereby updating the dynamic component of information. GIS technology is not inexpensive, even though hardware prices have been steadily declining. On the other hand, software as well as training requires a significant amount of resources. Before the benefits of GIS become available, a considerable investment in time and resources is required in system development, data preparation and training.
At the global level
It is important to realize that GIS is only a tool for overlaying maps, though it makes such overlay and integration easy. The results of GIS processing are determined by the analyses carried out and the quality of data. The most expensive part of the GIS use lies in the data preparation. It is important to note that more and more regional and global data of topography, land cover, soil characteristics, etc., prepared under various international and regional collaborative programs are becoming freely available. This makes it possible to start on GIS programs with base data sets that can be upgraded with more detailed data if the need arises.
At the national level
In preparing data for GIS, it is extremely important that data--especially the static data that does not change often--are produced as collaborative efforts between various governmental and other interested organisations. Experience shows that unnecessary duplication of this expensive endeavour is delaying the widespread use of GIS in many countries. It is important for governments to have a national policy on the development and sharing of digital data related to disaster mitigation efforts.
Figure 1. Use of different GIS layers in a distributed hydrological model
Figure 2. Example of flood inundation simulation during heavy rainfall using a distributed hydrological model
Figure 3. Schematic diagram showing methodology for flood damage estimation using GIS, Remote Sensing and distributed hydrological model
Figure4. Flood damage estimation from flood information (top), and property distribution (middle), utilizing damage functions for different property categories