Floodplain Management In Urban Developing Areas. Part II. GIS-Based Flood Analysis and Urban Growth Modelling PDF

Preview

Summary

Water Resources Management 13: 23–37, 1999. 23 © 1999 Kluwer Academic Publishers. Printed in the Netherlands. Floodplain Management in Urban Developing Areas. Part II. GIS-Based Flood Analysis and Urban Growth Modelling FRANCISCO NUNES CORREIA1, MARIA DA GRAÇA SARAIVA2, FERNANDO NUNES DA SILVA1 and ...

Description

Water Resources Management 13: 23–37, 1999. © 1999 Kluwer Academic Publishers. Printed in the Netherlands.

23

Floodplain Management in Urban Developing Areas. Part II. GIS-Based Flood Analysis and Urban Growth Modelling FRANCISCO NUNES CORREIA1, MARIA DA GRAÇA SARAIVA2, FERNANDO NUNES DA SILVA1 and ISABEL RAMOS3 1 Instituto Superior Técnico, Lisboa, Portugal 2 Instituto Superior de Agronomia, Lisboa, Portugal 3 Centro Nacional de Informação Geográfica, Lisboa, Portugal

(Received: 24 March 1997; in final form: 22 November 1998) Abstract. In Part I of this article the very dynamic nature of floodplain management was discussed and the need for modelling the urban growth processes and formulating scenarios of urban development was emphasised. In this second part, the use of Geographic Information Systems (GIS) for addressing those problems is presented. GIS have been recognised as a powerful means to integrate and analyse data from various sources in the context of comprehensive floodplain management. Adequate information and prediction capability is vital to evaluate alternative scenarios for flood mitigation policies and to improve decision making processes associated with flood management. A framework for the comprehensive evaluation of flood hazard management policies is also addressed in this article. This comprehensive approach to flood problems is more than an attitude or a philosophical starting point. It makes use of specific technological tools conceived to be used by different actors, some of them being nonexperts in flood analysis. These tools, based on GIS, are very appropriate for a participatory approach to flood policy formulation and floodplain management because they help communicating with the public in a scientifically correct and yet rather simple manner. Key words: flood analysis, flood hazard, floodplain management, GIS, scenarios of urban development, urban growth modelling.

1. Introduction and Problem Statement The main object of this article is to explore the possibility of using Geographic Information Systems (GIS), and complementary multimedia interactive devices, as tools for the comprehensive evaluation of floodplain management policies taking into account the dynamics of urban growth in fast developing areas. In fact, dissemination of relevant flood information using GIS, or other computer graphic devices, can help support individual and institutional decision making processes, either for preventive flood management or for dealing with emergency situations. In both cases, public involvement is a major issue for better decision making and

24

FRANCISCO NUNES CORREIA ET AL.

policy effectiveness. Therefore, GIS provides an appropriate working environment for the participatory evaluation of flood hazard management policies. Geographic Information Systems (GIS) have been recognised as a powerful means to integrate and analyse data from various sources (Correia et al., 1994 and Correia et al., 1995). Adequate information and prediction capability is vital to improve decision making processes. Therefore, emphasis is put on flood risk mapping based on hydrologic and hydraulic simulation of flood hazards, and a GISbased comparison of different scenarios for urban growth, improving the possibility of simulating the consequences of the alternative scenarios for urban development and for flood mitigation strategies. In general terms, the following steps have to be pursued in order to use GIS as a platform for integrating and crossing information that is relevant for flood policy formulation: • Refinement of hydrologic and hydraulic modelling aiming at a more accurate flood mapping for different scenarios; • Social and economic characterisation of the affected people and property in the floodplain; • Consideration of alternative scenarios for urban development and ex-post evaluation of flood land use control measures; • Preliminary evaluation of environmental impacts of flood mitigation measures. The integration of these components in the GIS, used as a technical tool by the professionals and also as a tool for public involvement, in the context of a participatory approach to floodplain management, is presented in Figure 1. An integrated and comprehensive approach is especially important in coastal areas subject to development pressures and floodplain encroachment that increase the frequency of flooding and the severity of the damages. This is the case of many areas in Southern Europe where land use practices generally do not take into account flood hazard management and mitigation. In these areas flood hazard management is closely related to land use management, requiring a comprehensive approach to floodplain management and the establishment of integrated and complex procedures for decision making. In this article, the use of GIS as a tool for integration is emphasised and its use for formulating urban growth scenarios is presented. In other related articles the use of GIS-based hydrologic and hydraulic modelling (Correia et al., 1998a), the relevance of the interface with the public (Correia et al., 1998b), and the use of multi-criteria methods for making flood management decisions (Silva et al., 1998), are discussed with more detail.

2. Conceptual Approach to GIS-Based Floodplain Management Many activities are necessary for the formulation of floodplain management policies. It is important to have a global picture of these activities in order to understand

GIS-BASED FLOOD ANALYSIS AND URBAN GROWTH MODELLING

25

Figure 1. Integration role of GIS in the decision making process.

how they relate to each other and how they contribute to the decision processes involved in policy making. Figure 2 presents a conceptual model with five basic stages of floodplain management policy making: data collection, analysis, synthesis, assessment and decision-making, which integrate the most relevant aspects of flood policy formulation. In the first phase, a digital data base for the catchment area is implemented, collecting and storing different types of data, such as biophysical, socio-economic and perception data. The second step is analysis aiming at the characterisation of flood problems. The main components of the data are analysed in order to select the key variables for assessing the general situation of the catchment and the floodplain in terms of biophysical and regulatory issues, hydrological and hydraulic regimes and variables, socio-economic assessment and characterisation of perception patterns and causal effects. The next phase allows for the generation of a comprehensive synthesis of the catchment characteristics and for the integration of specific components in the flood plain risk areas. Public perception of flood risks, namely those of residents, shop owners, local authority politicians and professionals must not be neglected, jointly

26 FRANCISCO NUNES CORREIA ET AL.

Figure 2. Conceptual model for flood analysis and floodplain management.

GIS-BASED FLOOD ANALYSIS AND URBAN GROWTH MODELLING

27

with the physical and human processes that contribute to the increase of flood risk and vulnerability in the catchment. The subsequent two steps lead directly to the decision-making process in floodplain management. They incorporate the development of scenario generation and the assessment of the impacts of those scenarios on flood effects. The scenario formulation must be based on urban development patterns and on different options for flood alleviation measures. Four options can be considered in general terms: • the ‘do-nothing’ option, which assumes that urban development will grow as in former years, with few constraints and that no structural or non-structural measures will be implemented; • the structural measures option, incorporating measures for flood control such as building a dam in the catchment headwaters area and retention basins in the flood plain inside the city; • the non-structural option based on the application of flood plain regulation, zoning and regulatory constraints within the catchment; • a fourth option, which is a mix of structural and non-structural measures. For each of these scenarios, an assessment process can be generated using GIS capabilities. This process will include a comprehensive approach, integrating the main components that have been considered throughout the study. A multi-attribute assessment of the effects of the different types of measures could be identified and tested. This process could be a useful tool to support decision-making at the local level and facilitate the assessment and monitoring of the process within a comprehensive context. The graphical display abilities of the GIS are an important element in the efficient diffusion of information to the public, especially when this information is sometimes highly technical.

3. GIS-Based Modelling in Flood Studies Information in a GIS can be presented not only in the form of maps but also in the form of tables in which alphanumeric data are stored (Aronoff, 1989) with a linkage to the graphic features. As shown in Figure 3, GIS may contain both geometry data (coordinates and topological data) and attribute data describing properties of the graphic objects (Fedra, 1993). Flood plains and flooding areas are typically a geographical feature. Therefore, most problems in this area can be represented and analysed in a geographical context and take advantage of a GIS. In fact, it is estimated that 85% of the information handled in planning activities is associated with geographic entities. Both on global and local scales, scientists and planners have recognised the merits of an integrated and interdisciplinary approach for providing a more complete understanding of the problems at hand and the alternative solutions to these problems (Clark et al., 1991). This is certainly the case of flood plain planning and management.

28

FRANCISCO NUNES CORREIA ET AL.

Figure 3. Graphic and alphanumeric data in a GIS.

The integration of GIS and environmental models in general, including the hydrologic and hydraulic models considered in this article, can be achieved by three basic approaches as described by Fedra (1993). According to the first approach, the environmental models are built into GIS. This high level of integration requires a sufficiently open GIS architecture that provides the interfaces and linkages necessary for tight coupling. The user has an interactive access to the coupled system. This is shown in Figure 4. In the second approach, there is a common interface of the models and GIS with the user. Information sharing and transfer between the respective components can take place. This is schematically represented in Figure 5. Finally, in the third approach there are two separate systems, the GIS and the model, that may interchange files. The model reads some of its input data from the GIS files and produces some of its output in a format that can be displayed by GIS. This approach requires less software modification and GIS stays basically untouched. The software of the model may be adapted or stay untouched as well. In this later case links are established through a programmed interface. This is displayed in Figure 6.

GIS-BASED FLOOD ANALYSIS AND URBAN GROWTH MODELLING

29

Figure 4. Tight coupling of environmental model and GIS with interactive interface.

Figure 5. Model and GIS with a common interface with the user.

It is desirable that commercial GIS packages go towards the first approach with the inclusion of tools that may merge models and GIS. Models should become one of the multiple analytical functions of GIS. Additional input to the model would be generated by GIS, and GIS would be directly used to display the output. This is especially important when the intermediate output needs to be used for further analysis with GIS. This approach still requires significant research efforts.

30

FRANCISCO NUNES CORREIA ET AL.

Figure 6. Model and GIS in separate systems.

In Part I of this article, as well as in previous related publications, the third approach is used with very good results. This is the case of Correia et al. (1994, 1995, 1998a, b). In this following sections this approach is also used with the purpose of predicting possible urban growth scenarios in the Livramento River Basin.

4. GIS-Based Urban Growth Analysis The case study of the Livramento river was already introduced in Part I. In this part, the main objective is to evaluate the consequences of different land-use policies on the flood regime. In fact, changes in the land-use of the catchment area induce different flood conditions that can be predicted by hydrological and hydraulic models (Correia et al., 1998b). There are two basic types of information needed for the analysis of urban growth dynamics: land-use data and Census data. In order to make the manipulation of these data more useful, the following steps were followed: (a) Identification of the boundaries of the polygons for the various years. This was done pairwise, from the oldest to the most recent data: 69 and 84, 69 and 89, 84 and 89, 89 and 91. This is necessary to be able of relating a polygon in a given year to the corresponding polygon in another year. (b) Definition of urban areas based on land-use maps. The information on the urban areas contained in the land-use maps was presented in separate data layers, one layer for each year. Each urban polygon was given an identification number, presented in Table V, so that we may follow its evolution over time. The urban polygons sometimes vanish, because they are absorbed by another polygon.

GIS-BASED FLOOD ANALYSIS AND URBAN GROWTH MODELLING

31

Table VI. Evolution of urban area and population in the town of Set´ubal

S (ha) Population Density (inhab. ha−1 ) 1S yr−1 1Population yr−1

1970

1981

1991

422.1 59395 140.7 – –

1029.1 89696 87.2 60 3000

1176.4 90527 77.0 15 83

(c) Data validation. For using GIS, the graphic units need to be validated and depicted as features for future links to the alphanumeric database. This process is done for each year separately, corresponding to different layers of information. (d) Linkage of land-use data to a simple database. The graphic files related to landuse were linked to an alphanumeric database with a simple structure, including only the label (land-use), ID no. and area. (e) Linkage of Census data to a complex database. The graphic files related to the Census data had to be linked to a more complex alphanumeric database, including data on dwellings and population. This information is only available for 1981 and 1991. With all the graphical and alphanumeric data properly linked to the databases, some of the parameters relevant to the characterisation of urban growth dynamics, such as area, perimeter and density values, can be immediately and automatically computed. One important aspect that can also be immediately investigated is what kind of land-use classes have preferentially been used for urbanisation. In the GIS this is computed by an overlay operation of the urban area at time (t) with the land-use map at time (t–1). The analysis and quantitative characterisation of urban growth dynamics will allow the projection of scenarios for the future. It is possible then to estimate the area of open land that will be converted into a developed area in Setúbal for each one of the four growth scenarios described above. To compute these values, it is necessary to consider the typologies of growth, the rates of increase in population and developed land in the last decades and the guidelines or restrictions established in the municipal Master Plan. This analysis must be done in the entire town of Setúbal and not only in the floodplain because the growth dynamics is not specific to this area. Direct measurement of landuse and population data can be done by the GIS using information from Census surveys. This data is summarised in Table VI. From the analysis of this table, the following conclusions can be formulated:

32

FRANCISCO NUNES CORREIA ET AL.

Table VII. Evolution of urban area and population in the surrounding area of Set´ubal

S (ha) Population Density (inhab. ha−1 ) 1S yr−1 1Population yr−1

1970

1981

1991

36.2 – – – –

303.1 – – 26 –

396.4 3708 9.5 9 370

• The town’s rate of growth decreased significantly between 1981 and 1991. This is certainly related to the economic crisis of this region during this period. The decrease in the rate of growth can be observed in terms of population and in terms of conversion of open land into built up areas; • In spite of the decrease mentioned above, newly developed areas are expanding at a rate above the needs of the population increase. This fact may be associated with a significant decrease in population density which was reduced by almost 50% from 1970 to 1991. If we analyse not only the town area but also the surrounding region (Table VII), we conclude that most of the newly urbanised areas grew up separately from the urban perimeter of Setúbal, especially between 1970 and 1981, originating new urban perimeters of small dimensions and densities, and confirming a ‘leap-frog’ type of development. The yearly rate of increase in developed areas in the surrounding region of Setúbal was approximately one half of the one registered in the town of Setúbal, in spite of the fact that the largest rates of increase in population occurred precisely in the surrounding areas. The identification of the typologies of growth, namely linear, continuous and discontinuous as presented in Section 4.3 of Part I, can be also done by GIS. This identification can be performed stepwise as displayed in Figures 7 and 8. For the linear growth it is necessary to build a separate layer with the road system only and then use the buffer function included in the GIS to create a file with polygons that are overlaid with the result of the urban increased polygons (Figure 7). These areas are measured directly through database operations in relation to the total growth and compared to the values for the different time periods. For the continuous growth typology, the procedure is similar, as can be seen in Figure 8. The buffer zone of a polygon at time (t–1) is crossed with the urban area at time t. This type of occurrence is characterised in terms of frequency and area. The analysis and quantitative characterisation of urban growth dynamics will allow the projection of scenarios for the future, according to the options presented in Section 4.3 of Part I.

GIS-BASED FLOOD ANALYSIS AND URBAN GROWTH MODELLING

Figure 7. Stepwise procedure for identification of linear growth typology.

Figure 8. Procedure for identification of continuous growth typology.

33

34

FRANCISCO NUNES CORREIA ET AL.

As a result of this analysis, the final conclusions were that the urban growth scenarios in Setúbal for the next decade can be quantitatively characterised as follows: • Trend Scenario Growth of the urban perimeter jointly with the growth of new areas outside the urban perimeter, corresponding to the continuation of a ‘leap-frog’ pattern of conversion into an urban land-use of open land. A growth of 150 ha in the urban perimeter can be expected in the next ten years. A growth of 350 ha can be expected outside the urban perimeter during the same period of time. This growth will occur mainly in the old farms located in the flood plain and in the Eastern plateau. This is the worst scenario from a flood management point of view and deserves attention when formulating flood policies. • Official Scenario Growth of 15 ha in the urban perimeter and 90 ha in the surrounding region. However, growth in the surrounding are...