PLANNING
André Botequilha Leitão
Centro de Valorização de Recursos Minerais (CVRM)
Instituto Superior Técnico - Universidade Técnica de Lisboa
Avenida Rovisco Pais - 1096 LISBOA CODEX - PORTUGAL
Tel: +351.1.841.7247; Fax: +351.1.841.7442; Email: p1118@alfa.ist.utl.pt
Abstract:
Landscape is being transformed at an accelerated rate in developed countries, due both to urban growth and an intense exploitation of natural resources. Urban fringe areas tend to be more affected as land use change processes are more dynamic as a consequence of urban expansion. Landscape description and evaluation, visual impact assessment and the definition of reclamation and conservation priorities are important to assist decision-making processes. Geographical Information Systems (GIS) can play an important role as a planning and management tool, supporting such decision-making processes.
The landscape evaluation model presented was developed in GIS Arc-Info vers. 7.01. It relies on two indicators: LVQ - landscape visual quality and LVF - landscape visual fragility. Landscape Capacity (LC) integrates both indicators in order to establish land use planning priorities. The model allows the simulation of land use changes scenarios and to assess the effects of visual barriers in minimising visual impacts.
The results of a case study in the south area of Loures Municipality are presented. The model allowed the identification of visual intrusions and landscape values, which presented high values of LVF. The former consists in priority areas for reclamation, the later for conservation.
The potential role of this model in landscape analysis, planning and management and for environmental impact assessments is discussed.
Keywords: geographical information systems, landscape planning, landscape visual fragility, landscape visual quality, visual impact assessment.
Due both to an intense exploitation of natural resources and urban growth, landscape is being transformed at an accelerated rate in developed countries (Bólos et al., 1992).
Being adjacent to the city of Lisbon, Loures Municipality is strongly influenced by its urban expansion. The concentration of activities belonging to the third sector (services) in Lisbon induced a residential function transfer from the core of Lisbon to the surroundings areas such as Costa do Sol, Amadora, Loures, etc. Between 1930 and 1980 Loures had seen its population density growing 10 fold, from 156 Hab/km2 to 1.482 Hab/km2. Similar to what it happens frequently in large metropolitan areas, and the Lisbon Metropolitan Area is no exception, Loures natural resources has been systematically and unnecessarily degraded.
Loures territory is divided in two areas: the north area, located north of the city of Loures, is predominately rural as the south area - the study area, with circa 100 sq., is largely occupied by urban / industrial areas, representing more than a third of its occupation (36%). The expectant areas constitute areas which were former agricultural areas located between existent urban areas, abandoned due to low agricultural potential who's owners expect that these could be converted to an urban land augmenting by several figures their value. These areas occupy a great percentage of this territory - 16%. Together actual and expectant urban areas sum up more than half of the South Loures territory. Another third (33%) is occupied by agriculture, 6% by forested areas, namely pine woods, deciduous forest including Eucalyptus sp. and mixed forest, and 9% by "natural" formations such as maquis, garrigue, etc. (Botequilha Leitão 1996).
Both urban and architecture quality that are characteristic of the surrounding urban areas of Lisbon are very low. According to the Loures Municipal Master Plan (PDM) studies (HP 1987) the major causes for the Loures environmental degradation are related to deficient allocation of urban / industrial areas, lack of a coherent urban structure and construction typologies adopted, inadequate integration of existing buildings in the landscape, degradation of areas located in the vicinities of urban / industrial areas and infrastructures (highways, railroads, airport, etc.), deficient exploitation of mineral resources (quarries), causing important landscape impacts, etc.
It is therefore necessary to define strategies to minimise landscape impacts of existing and future situations occurring as a consequence of incorrect land use planning and management.
To enable the municipalities such as Loures to fight these situations, and to plan and manage its territory it is necessary first to be able to identify and characterise situations where landscape impacts do occur and were impacts potentially could occur. The identification of landscape values is also important for conservation purposes and for valorisation of such situations as they represent a natural and cultural heritage of a territory. Human and financial resources are always scarce to overcome all the problems of the municipalities at the same time. Therefore is necessary to define priorities in site reclamation and conservation.
It is also necessary to enable the municipality with the appropriate tools regarding site selection for future activities. What it concerns to landscape impacts, it is important to evaluate the territory suitability to locate activities that constitute visual intrusions, such as quarries, landfills, etc. in order to minimise visual impacts.
Geographical Information Systems can play an important role in landscape studies (Steinitz 1990, Howes & Gattrell 1993, Miller et al. 1994, Haynes-Young & Green 1994, Falque et al. 1995, Botequilha Leitão 1996), assisting these tasks in different phases of the visual landscape planning and decision-making processes:
- data input, storage and manipulation - creation and update of data layers e.g. topography, land use, etc;
- outputs - cartography production, reports, public presentations, etc.;
- analysis - e.g. terrain analysis, land use analysis, identification of landscape values and visual intrusions;
- modelling / simulation - e.g. visibility studies, simulation of land use changes, visual impacts or effects of visual barriers.
Data update and map productions are extremely rapid and simplified tasks in a GIS environment compared with the traditional methods. These provide update information for modelling / simulation and to produce maps to assist several analysis, field works, to include on reports, to perform public presentations, e.g. presenting to the public site selection processes, etc. The almost immediate update of geographic data, and consequently of the plans based on this data, allows also a higher dynamism and flexibility to the planning process.
Terrain analysis such as slope, aspect, elevation range, determination of topographic features such as valleys, ridges, watersheds, or land use analysis are some of the tasks which frequently support visual landscape studies. These tasks are relatively fast and easy to do in a GIS. It is possible also to study visibility in vast areas, identifying points in a territory which are particularly exposed to human sight or hidden by landscape features (Kent 1986, Howes and Gattrell 1993). All these tasks are normally time consuming by traditional methods. Some of them, such as visibility studies are almost inoperative considering the tasks involved. Additionally land use heights can be incorporated into the DTM to provide a more accurate basis for visual analysis (Kent 1986, Miller et al. 1994).
Modelling landscape characteristics to perform impact simulation of different land use scenarios is an important characteristic. GIS "can serve as test beds for studying environmental processes or for analysing the results of trends, or of anticipating the possible results of planning decisions. By using the GIS in a similar way that a trainee pilot uses a flight simulator, it is, in principle, possible for planners and decision-makers explore a range of possible scenarios and to obtain an idea of the consequences of a course of action before the mistakes have been irrevocably made in the landscape itself" (Burrough 1986: p.7).
A GIS application for municipal planning was implemented in Arc View 3.0 within the framework of the SIGLA project (Muge et al. 1997) for which the presented research work was developed. Landscape quality indicators was integrated with several other indicators concerning other environmental components considered such as air, surface and groundwater, soils, noise, biodiversity, etc to enable the construction of an integrated environmental index. Simulating different land use scenarios, the GIS application allows to evaluate the environmental performance of a territory which resulted from these scenarios. This application was developed to constitute a Decision Support System (DSS) and it is being implemented in the National Environmental Agency (Direcção Geral do Ambiente) in Portugal, which was the contractor of this research project.
The model for Landscape Scenic Capacity evaluation and simulation presented was developed in the GIS platform ArcInfo, vers. 7.01 (Environmental Systems Research Institute, California), in a workstation UNIX Hewlett-Packard 9000 series, model 720, with 32 MB RAM. The GRID module of ArcInfo (Esri) was used to construct the Digital Terrain Models (and TIN module), to perform the visibility studies, to obtain the landscape indicators, etc. Working in a raster format was considered to be the best solution for the calculations and studies that was developed. The land use map was originally constructed as a coverage and converted to a "grid" to facilitate the overlay with the other grid layers. This model was totally implemented in Arc Info programming language - AML or Arc Macro Language (Esri). Layouts were produced with Arc View vers. 2.0 (Arc View 3.0 was not available at the time) since it offered major advantages to produce simple layouts compared with Arc/Plot, ArcInfo module for producing layouts and maps.
This model, proposed formerly in 1995 and developed further in 1996 by Botequilha Leitão (1995, 1996), was based on a theoretical model presented in Bólos (1992) and adapted by the author. The proposed theoretical model evaluates Landscape Scenic Capacity (LSC) through three aspects that were considered descriptive of the present value of LSC: LVQ - landscape visual quality; LVP - landscape visual potential and LVF - landscape visual fragility. It is proposed that the integration of the three aspects be implemented through a rule-based system to support decision-making processes.
In the actual GIS model presented here, only LVQ and LVF were implemented and LSC resulted from the product of both referred components (Figure 1).
The methodology its based upon the following assumptions: (1) landscape quality is approached uniquely by its visual or scenic component, (2) the methodology was simplified to enable its implementation on a GIS environment, (3) consequently it was only considered some specific characteristics of the terrain such as slope and aspect combined in a single indicator - Slope / Aspect Indicator, aesthetic value of land use (or as visual intrusions) and (4) the presented model is considered as a first implementation of the proposed theoretical model to test the feasibility of its development within a GIS environment.
Figure 1. - Landscape Scenic Capacity evaluation methodology implemented in the presented model
Landscape visual quality (LVQ) evaluation is based upon two main features: (1) land use aesthetic or scenic value and (2) relief.
Aesthetic value, which range from -5 to +5, is assigned to each land use category by the author according to his personal judgement. Land uses considered as visual intrusions such as quarries, industrial facilities, airport, degraded areas, urban areas, military equipment, roads, etc. was assigned with negative values. Homogenous agro-systems (cereals, nurseries, and greenhouses) were assigned low positive values as heterogeneous agro-systems were assigned higher values. Forests, natural areas, archaeological and historical sites were assigned with the highest positive values. The basis for aesthetic value assignment was a land use map of South Loures developed by the National Centre for Geographic Information (CNIG) to the SIGLA project (Fig. 2).
Figure 2. - Land use map for South Loures Municipality (source map scale 1:25.000).
Relief was taken into acount by a Slope / Aspect Indicator, which resulted from the combination of these two topographical components: aspect and slope. These were considered as indicators for luminosity, which affect visibility conditions. South steep slopes were considered to receive more luminosity than others do and therefore to enhance an object located in these situations, enhancing its original aesthetic value. North steep slopes were considered as in shadowy conditions where objects would be less exposed to observation, reducing some of its original aesthetic value.
To calculate the topographic components, such as slope and aspect, and to perform visibility analysis it was necessary to produce a digital terrain model (DTM), a three dimensional representation of the study area. The DTM used for calculation of slope and aspect (TIN model) was derived from topographic data layers (contour lines and points) of 1:25.000 scale maps from Instituto Geográfico do Exército (IGEoE). To compute a new DTM to perform the visibility studies which integrates land use estimated heights (DTMlu) a LATTICE (Fig. 3) was derived from the original TIN. The calculation is carried out without using terrain data for the land surrounding the municipality.
LVQ resulted from the product of both components.
Figure 3. - Digital Terrain Model for South Loures Municipality (source map scale 1:25.000).
Landscape visual fragility (LVF) is defined as the degree of exposure of a point to human observation. The more exposed a object is to observation, the more fragile it is. Land use changes in those places are prone to induce high visual impacts. On the other hand, places hidden from observation, with low values of LVF are suitable to allocate land uses that constitute visual intrusions. The calculation of the LVF indicator is based in two components: (1) intervisibility indicator (II) and (2) relief (equal as in LVQ).
The intervisibility indicator is based on the analysis of the terrain, calculating the total number of pixels that are visible from each of 119 selected observation points (the visual fields). The 119 visual fields grid layers are then added to form one single II layer. This indicator was not computed from all points in the territory due to simulation time constraints. Computation of II is highly time consuming. Additionally it was considered redundant to study intervisibility from all points since the territory is not frequented homogeneously by the population (Neuray 1982). Urban areas and roads are potentially more frequented than other places, e.g. places located in the middle of agricultural fields or forests, or sites that are inaccessible due both to topographic features e.g. steep slopes, peaks or due to the inexistence of trails or roads. According to this criteria - potential accessibility to the territory, it was selected 119 observations points, one single point per polygon classified as an urban area (located at the polygon centroid) and several points along roads and railways (Fig. 2).
Similarly to LVQ, LVF results from the product of both components.
Landscape Scenic Capacity (LSC) is defined as the quality of the territory to satisfy the requirements of certain uses or activities in what concerns scenic characteristics (Bólos 1992, Botequilha Leitão 1995, 1996) , e.g. recreation activities, tourism, nature observation, contemplation, etc.. In such cases where land uses constitute visual intrusions (quarries, landfills, industrial areas, degraded areas, etc.) the territory capacity for such scenic quality demanding activities is negative, i.e., they contribute negatively to the territory overall capacity for those activities. Besides aesthetic or scenic quality, LSC consider other characteristics of the landscape relevant to its planning and management, such as visual fragility. LSC is the product of LVQ and LVF.
All raster layers were produced with a 40m-grid resolution, and multiplied to obtain the final LSC data layer. The 40m-grid resolution was selected based on a sensibility test performed to assess the higher grid resolution that allowed the representation of the smaller land use polygon.
Visual impacts (VI) were then calculated by simulating land use changes in selected polygons, calculating new LSC values for each situation, and subtracting these to the original LSC values for situation 0, which was calculated before the land use changes (VI = LSCn - LSC0). For each simulation, it was necessary to compute new DTMlu´s as this model comes out modified each time a land-use change occurs. The intervisibility indices and new LSC layers are then recalculated for each simulation.
Some considerations concerning computation of certain indicators
The New Digital Terrain Model
To compute the visual sheds or visual fields required to the intervisibility indicator, it was constructed a new DTM - DTMlu (LATTICE model) that incorporates land use height data. The land use map of South Loures was used in these intervisibility calculations. Data on z-dimensions values was estimated by the author for each land use category e.g. water surfaces and streams, estuary, airport - 0m; roads, agricultural areas, salt-marshes - 1m; orchards and olive groves - 5m; woodland - 10m; industrial areas, quarries - 15m; urban areas - 20m. Visibility studies were carried out with the VISIBILITY command of GRID.
These estimations are somehow questionable but had the objective of testing the methodology rather than attaining a high accuracy on this or that particular height assumption. Additionally it was neither possible nor perhaps desirable to estimate a single height value for each urban area or building. In fact if it that was possible, the model would probably be extremely "heavy" and inoperative, at least with the hardware involved in the SIGLA project.
Parameters computation in a neighbourhood
The parameters that required the computation of certain characteristics within a (visible) neighbourhood were avoided, exception made to the intervisibility indicator (II). The excessive time to compute the visual fields was the key issue.
As referred Landscape Visual Potential (LVP) was the component of the three original components considered in the theoretical model that was not included in the presented model. This was due to excessive time to compute it. LVP has two aspects: quantitative and qualitative. Considering a certain observation point, it is possible to have either a wide view or a narrow view (with all the middle terms). Also, it can be a forested or a mountain landscape or other landscape that is considered to have a high scenic value, or over a industrial area or degraded area and therefore considered to have a low scenic value. Besides the areal characteristics of a certain view, it is therefore necessary to assess also its quality.
The LVP layer could be obtained by computing, for each point (or cell) in the study area (circa 64.500 cells), its visual field, overlay it with the LVQ layer to obtain the LVQ values for each visual field considered, compute a LVQ average value for that observation point and finally store it in the respective cell. If we consider 10 seconds per visual field computation it would take 7 days round the clock to compute one single LVP layer. If we consider additionally that the DTMlu is changed every time a new simulation is performed, and that in consequence it changes also some of the visual fields, it is obvious that a new, faster method is necessary to compute these "visual neighbourhood" dependent parameters.
One possible solution could be to calculate these parameters "outside" ArcInfo, using faster visibility algorithms applied in fast image analysis software e.g. PERICOLOR (Matra Systems, Guyancourt, France), and to import such results into the ArcInfo core application through IAC (Inter-Application Communication).
Visibility Studies
In a preliminary phase simple tests were conducted to assess the VISIBILITY command performance, namely (1) if with the supplied DTM the command was getting a good "reading" of the territory, and (2) comparing the VISIBILITY results in two situations (Botequilha Leitão 1996):
- visual field computation from a selected observation point where the surface was the original DTM, without the land use heights incorporated into the model (Fig. 4);
- visual field computation from a selected observation point where the surface was the DTMlu, with the land use heights incorporated into the model (Fig. 5).
Figure 4. Visibility Studies. Situation 1. Visual field with the
Digital Terrain Model without land use heights incorporated.
Figure 5 Visibility Studies. Situation 2. Visual field with the
Digital Terrain Model with land use heights incorporated.
The point selected for observation was virtually located at the Várzea de Loures which is the floodplain of the River Trancão, being a quite flat extensive terrain, with an altitude of 3 meters above sea level (Situation 1). Its surroundings have both extensive flat and hilly areas, with steep slopes. In the second situation, since the land use heights was incorporated into the DTM, and that the observation point is located in an urban area, it means that the virtual observer, instead of being at the ground he is standing at the roof of a 6 storey building. It is important to notice that the VISIBILITY command does not allow observation points located below the considered topographic surface (SPOT HEIGHT). Therefore 90% of the observation points (urban areas) are considering the maximum impact that could occur. However, urban areas are represented by a single observation point, which in some situations (large urban areas) is scarcely representative of the total area.
As an example, lets face at the hillside located East to the observer (Figures 4, 5: Vale de Figueira Alto de S. Lourenço). It is possible for the observer to see the lower part of the hillside but not the top of the hill. This is due to the fact that the hill as very steep slopes on its lower part (higher density of the contour lines) but it softens in its upper part ending in a sort of a plateau, which lies hidden from the observer.
Now, to compare both situations, namely between the visual fields with DTM and DTMlu, lets turn to South and observe the visual fields on the hillside formed by Cabeço da Aguieira, namely when the river Trancão makes a turn to Southeast. Comparing both figures 4 and 5 the visual fields are different: the visual field in figure 4 is bigger than in figure 5. This can be explained by its urban ocupation (Fig. 2). The buildings form a visual screen hiding everything beyond them, including the top of the hill.
It seems that the command VISIBILITY is doing a relatively good interpretation of the terrain and that it distinguishes between both situations. Obsviously that the simplifications introduced by estimates of land use heights and by generalisation of these heights within each land use class generates some discrepancies. However these were not accounted for in this study.
Landscape Visual Quality (LVQ)
As we could expect the LVQ map (Fig. 6) reflects the evaluation conducted by the author - lowest values registered for South Loures concentrates in industrial and urban areas; high values are assigned to agricultural areas and woodlands.
Figure 6. - Landscape Visual Quality indicator for South Loures
Municipality - situation 0 (source map scale 1:25.000)
However, the influence of the relief component on the LVQ indicator induces some changes in the original scenic values. Let's take as an example two shrub formations in different situations: (1) in a south exposed slope, not too steep, located in the top centre of the LVQ map, indicated as Shr1 (Fig. 2) and (2) a north steep slope, located roughly in the centre Northeast of the map (Shr2). The relief value was determinant to differentiate between these two situations: the first (Sh1) is classified as Very High due to its favourable topographic situation, enhancing its original value. The shrub formation is more illuminated, consequently more exposed. The second, less rated, is classified as High due to its shadowy situation.
Landscape Visual Fragility (LVF)
The 0 values in the LVF map (Fig. 7) represent almost a third of the covered territory. It presents a more or less uniform distribution across the study area except along narrow valleys and small patches of expectant land in the middle of large urban areas, where observers' sight can't reach. Concerning its distribution among different land use categories, the great majority of the territory with 0 values is agricultural fields (60%), expectant land included, with reduced height (1m). Urban and industrial areas represent 20%. Frequently, these correspond also to situations in the middle of large areas, caused both by the fact that in each single area it was placed only one observer, revealing an insufficiency, and also by the topography itself, e.g. narrow valleys. Woodlands and natural areas occupy 10%. Some of them, namely the marshes (natural areas) along the river Tejo (estuary), result from a "shadow" effect that occurs near the border of urban and industrial areas. The observer can't see those border areas because he is located in the roof, frequently in the middle of the area.
Figure 7 - Landscape Visual Fragility indicator for South Loures
Municipality - situation 0 (source map scale 1:25.000).
High LVF values presents a strong correlation with the general topography of the study area: (1) areas with steep slopes, in asymmetric valleys e.g. Costeiras de Lisboa, high isolated hills e.g. Montemor (geodesic vertex (v.g.) South of Loures) or very exposed ridges e.g. Pinheiro de Loures (Northwest of Loures) constitute situations that are highly exposed to observation; (2) extensive flat cultivated open fields such as founded in the Loures flood plain (Várzea de Loures), located East - Northeast of Loures are also extremely vulnerable to human observation; (3) low height land uses (estuary, salt marshes) along Tejo river banks (East of the study area) form a relatively large strip (+ 700m) constituting a more exposed situation considering its surroundings and generating consequently higher LVF values.
Almost half of these areas (45%) are occupied by urban or industrial uses; added the degraded areas, it sums 51%. If we considered the fact above mentioned that the quality of these urban-industrial spaces are, in general, very low, representing therefore visual intrusions, the situation seems to be of concern. Additionally, and considering not only the visual impact perspective but also the land use planning point of view, the urban-industrial occupation is inadequate since some of these areas present very steep slopes. Examples of these situations are the urban areas located along the Costeiras de Lisboa, Alto de S. Lourenço, North of Odivelas and Loures (Pinheiro de Loures). Concerning degraded areas such as quarries, degraded urban areas, etc., half of the total surface area occupied by these in the study area is located also in high LVF situations. A third is occupied by cultivated land, such as irrigated fields in the Loures flood plain and expectant land scattered all over the territory. Woodlands and "natural" formations occupy just 8%, being the rest distributed by other land uses.
Landscape Scenic Capacity (LSC)
Landscape Scenic Capacity (LSC) scale is two-folded: the positive scale regards those situations where scenic quality is positive, the negative scale those where LVQ is negative (Fig. 8).
Figure 8. - Landscape Scenic Capacity indicator for South Loures
Municipality - situation 0 (source map scale 1:25.000).
The former represents a priority scale for landscape conservation, the later for reclamation. Those situations where both the landscape or scenic value is high and it is highly exposed to observation represent highly vulnerable areas where land use changes or any action leading to its present value degradation should be regarded with extreme caution. Examples are some areas located Southwest and South of Loures, others along the Costeiras de Lisboa, some cultivated fields in the Loures flood plain, along the river Trancão or along the Tejo estuary.
The opposite situations, represented by such areas above mentioned where urban areas, degraded areas, quarries are highly exposed, constitute situations where measures should be taken, namely through site reclamation or visual impacts minimisation e.g. placing visual barriers. Examples for those situations where already pointed out in the former chapter (Costeiras de Lisboa, North of Odivelas, etc.).
Simulations of Visual Impacts
For visual impact simulation it was altered the existent land use at point P1 (Fig. 8) - agricultural land to a 20 meters high building, which we will call scenario 1; olive groves and low shrubs were replaced by forest stands in the surroundings of point P2 - scenario 2. At the first point considered both LVQ and LVF are low and consequently also LSC is low. It was intended with this scenario to test a land use change from a land use with a relatively low average height to one with a higher average height. It was expected that this change would increase visual impacts in the surrounding area. At the second point, land use change was tested not at the point itself but in the surroundings, in this case around an existent quarry simulating a visual barrier. The quarry is situated in a high LVF location. It was expected that this change would reduce the quarry negative visual impact.
Scenario 1 caused an average negative impact of +50%, i.e., the LSC value for the attained areas has diminished in average half their value. The spatial dispersion of the impact is quite significant as it can be observed in Fig. 9. It was registered positive impacts in some situations, negative in others, although the global result is negative. This means that among the land uses seen by the 119 observers in the situation 0, the land uses assigned with high LVQ were predominant. After the land use change these were less seen, diminishing the overall LSC value. In fact it represents a real situation since it is not rare that e.g. a building is built in front of others resulting in a loss of views. However, this can be true also for the inverse situation where the new land use acts as a visual barrier towards unpleasant views or visual intrusions. In this case, LSC value for the resulting scenario is higher than in the situation 0.
Fig. 9 - Landscape Visual Impacts - Scenario 1.
Although point 1 was classified with a low LVF value the land use change resulted in an impact across a large extension. This situation can be explained by its topographic characteristics and by the land use in the surrounding areas. Point 1 is located in a ridge and surrounded by forests. When a 20 meters building substitute a land use with only 2 meters, it rises up from the forest canopy (10 meters high) which surrounds it. The building is now 10 meters above the forest canopy and additionally it is in a ridge where exposure is normally higher than surrounding areas. Therefore it can be concluded that LVF indicator is not absolute since places classified with low LVF values in situation 0 can turn out to generate high visual impact. This situation can be partially resolved by simulating the land use proposed and calculate visual impacts induced. One thing is for sure: places with high LVF are always prone to induce high visual impacts unless visual barriers are used. Intervisibility relationships in a territory are therefore complex and LVF indicator map can only be used as a reference and not as a final map.
Scenario 2 caused an average positive impact of + 277%, i.e., the LSC value for the attained areas has augmented in average three times their value. The impact is localised in the surrounding areas as it can be observed in Fig.10.
Fig. 10 - Landscape Visual Impacts - Scenario 2.
Positive impacts are due to two situations: (1) the visual barrier is diminishing the quarry LVF value and (2) by the positive value of the visual barrier itself. Therefore visual barriers has a twofold effect of attenuation and valuation. The importance of the visual barriers in these situations are not restricted by its role as a "barrier" but are also contributing by its intrinsic value for the overall LSC value.
In this manner is possible therefore to model the effect of visual barriers in minimising negative visual impacts and in the contribution for increasing the landscape scenic value.
The Landscape Visual Fragility indicator can be used per se, constituting an objective indicator for visual impact. However land use changes must be simulated to assess the impact of such changes since locations assigned with low LVF values may generate high visual impacts.
The positive scale of the Landscape Scenic Capacity map can be of assistance to define priorities for landscape value conservation, the negative scale for landscape reclamation.
The Visual Impacts simulation allows studying the effects of land use changes and to compare various alternatives based on generated impacts. It allows also studying the effects of visual barriers allocation.
It is important to stress nevertheless that there are several issues that need to be address to ameliorate the model presented. As an example, among others, it would be necessary to test the DTM quality - errors associated to visual fields and by using more detailed land use height estimates. To increase the percentage of covered territory by the observers studying its influence versus increased number of observers, to consider the observer distance to weight impact significance and the introduction of population densities and traffic volume to weight accessibility. However it is also important to remind that all these improvements represent also that the model would be "heavier", increasing simulation time, and that not always the required data to put it all in practice is available. It is therefore essential that a trade-off between precision and feasibility is established assuring that the model application is not compromised by excessive refinements.
Financial support from Junta Nacional de Investigação Científica e Tecnológica within the PRAXIS XXI program (M.Sc. and Ph.D. grants) is gratefully acknowledged. Thanks are also due to Câmara Municipal de Loures for the topographic data used in this research project.
Bolós, M. et al. 1992. Manual de Ciencia del Paisaje. Teoria, métodos y aplicaciones. Colección de Geografia. Masson, S.A.. Barcelona.
Botequilha Leitão, A. 1996. "Integração de Técnicas de Avaliação da Paisagem em Sistemas de Informação Geográfica". Dissertação para a obtenção do grau de Mestre. Instituto Superior Técnico. Lisboa.
Botequilha Leitão, A. 1995. "Um modelo de avaliação da qualidade paisagística em Sistemas de Informação Geográfica" in Proceedings do III Encontro sobre Sistemas de Informação Geográfica. USIG - Associação dos Utilizadores de Sistemas de Informação Geográfica. Lisboa.
Burrough, P.A. 1986. Principles of Geographical Information Systems for Land Resources Assessment. Monograph on Soil and Resources Survey. Claredon Press. Oxford.
Câmara Municipal da Amadora (CMA). 1990. Plano Director Municipal da Amadora. Dossier A - Estudos de Caracterização ao Nível Regional. Gabinete do PDM. Amadora.
Falque, M.-C., Bonnefont, E., Languepin, M., Butel, J. & Esteben, M., 1995. Évaluation des paysages de L´Hérault: une approche méthodologique et une application SIG adaptée. Mappe Monde, 2: 14-17.
Hadrian, D.R., Bishop, I.D. & Tree, M., 1988. "Automated mapping of visual impacts in utility corridors". Landscape and Urban Planning, 16. 261-282.
Haynes-Young, R., Green, D.R. & Cousins, S.H. (Eds.), 1993. Landscape Ecology and GIS. Taylor and Francis. London.
Hidrotécnica Portuguesa (HP). 1987. Plano Director Municipal de Loures. Estudos Prévios. Vol. 2 - Caracterização Biofísica. Tomo 1 - Texto, Quadros e Figuras. Lisboa.
Howes, D. and Gatrell, T. 1993. "Visibility analysis in GIS: issues in the environmental impact assessment of windfarm developments" in Hart, J.; Ottens, H.F.L. and Scholten, H.J., editors: Proceedings of 1993 European Conference on Geographical Informations Systems. Utrecht. 2.: p. 861-870.
Kent, M. 1986. "Visibility analysis of mining and waste tipping sites - a review". Landscape and Urban Planning, 13: p. 101-110.
Miller, D.; Morrice, J.; Horne, P. and Aspinall, R. 1994. "Characterisation of landscape views" in: Hart, J., Ottens, H.F.L. and Scholten, H.J., editors. Proceedings of 1994 European Conference on Geographical Information Systems / MARI. Utrecht. 1: p. 223-232.
Muge, F.; Botequilha Leitão, A.; Neves, N.; Grueau, C.; Ribeiro, L.; Figueira, R ; Nunes, C. and Cortêz, L. 1997. "Modelling Integrated Environmental Indicators in a Geographical Information System" in Reis Machado, J., editor: Proceedings of International Conference on Environmental Challenges in a Expanding Urban World and the Role of Emerging Information Technologies. Costa da Caparica. Lisboa.
Neuray, G., 1982. Des Paysages. Pour qui? Pourquoi? Comment? Les Presses Agronomiques de Gembloux. Belgique.
Steinitz, C., 1990. Toward a sustainable landscape with high visual preference and high ecological integrity: the Loop Road in Acadia National Park, U.S.A.. Landscape and Urban Planning, 19: 213-250.
