Remote sensing and GIS in modeling visual landscape change: a case study of the northwestern arid coast of Egypt

Remote sensing and GIS in modeling visual landscape change: a case study of the northwestern arid coast of Egypt

Landscape and Urban Planning 73 (2005) 307–325 Remote sensing and GIS in modeling visual landscape change: a case study of the northwestern arid coas...

687KB Sizes 0 Downloads 151 Views

Landscape and Urban Planning 73 (2005) 307–325

Remote sensing and GIS in modeling visual landscape change: a case study of the northwestern arid coast of Egypt Yasser M. Ayad∗ Department of Anthropology, Geography and Earth Science, Clarion University, 335 Peirce Science Center, Clarion, PA 16214, USA Received 14 April 2003; received in revised form 2 August 2004; accepted 24 August 2004 Available online 10 November 2004

Abstract Land use planners in many countries have recognized the importance of the aesthetic values of landscape. Their desire to incorporate these values into decision-making processes has created a need to identify valid ways to quantify the scenic characteristics of landscapes. This has led to an increasing interest in the use of spatial data and geographic information systems (GIS) methodology in assessing visual attributes of the landscape. The objective of the present study is to assess the visual changes in a rapidly developing coastal area of Egypt using remotely sensed data (satellite images and aerial photographs) and raster GIS modeling. The analysis assesses changes between a period characterized by a vernacular, relatively natural landscape (1950s) and the beginning of the exploitation of the region for resorts (1990s). Using land use/land cover classes extracted from the satellite images and aerial photographs, four visual attributes of landscape are identified: land use/land cover diversity, activity (degree of naturalness), proximity to the shoreline, and topographic variety. A composite index is also developed. Although these attributes and the composite index rely mostly on the type of land use/land cover information on the landscape under consideration, the adopted techniques succeed in detecting several changes in the attributes, spatially locating them and mapping the magnitude of their changes. This study demonstrates what can be done to analyze and assess what is usually considered an incommensurable resource, the visual attributes of landscapes. It also reveals the extent of the impact of unplanned or ill-planned activities on one of the fragile resources of arid landscapes. © 2004 Elsevier B.V. All rights reserved. Keywords: GIS; Visual assessment; Landscape change; Developing countries; Arid landscape

1. Introduction Scenic landscapes are a major source of human enjoyment and in some cases have been the object of direct public action to preserve their quality (Bishop ∗

Tel.: +1 814 393 2990; fax: +1 814 393 2004. E-mail address: [email protected].

0169-2046/$20.00 © 2004 Elsevier B.V. All rights reserved. doi:10.1016/j.landurbplan.2004.08.002

and Hulse, 1994; Dearden and Sadler, 1989; Fabos et al., 1978; Itami, 1989; Moss and Nickling, 1989). However, the degree of appreciation and the specific perceptions of landscape values may widely differ from one society to another. These cultural characteristics play a key role in how people act on the landscape: different cultures see the physical environment in different ways (Opie, 1979). As Steinitz (1990) states:

308

Y.M. Ayad / Landscape and Urban Planning 73 (2005) 307–325

we can simultaneously judge landscapes in the coastal zone, in the desert, in the mountains, and in a great city as being “most beautiful” even though these landscapes exhibit great diversity. No single descriptive or predictive model can account for these equivalent judgments. Naveh (1995) traced this anthropological concept of culture based on socially transmitted rather than biologically determined behavior. In this sense, all landscapes inhabited, influenced, or modified by humans are the tangible products of interactions between nature and cultures (Altman and Chemers, 1980; Smardon, 1983; Naveh, 1995). Both natural and cultural processes—including planning and management decisions—lead to physical changes that will eventually be seen. As a result, the physical characteristics of the landscape can be identified by their visual attributes. With the development of land use planning, and its requirement for environmental data on which to base land use decisions, came an increased desire to elaborate valid means to quantify the scenic characteristics of landscapes (Litton et al., 1974; Zube et al., 1975; Smardon, 1983; Jakle, 1987; Zhang et al., 2000). Advancements in this area, particularly in methodology, are similar to developments in other research areas, which collectively contribute to the field of environmental management. The varied technological and methodological approaches presented at the “Our Visual Landscape” conference held in Ascona Switzerland in August 1999, clearly indicate the significant advances in computer simulation (Bishop et al., 2001; Danahy, 2001; Hehl-Lange, 2001; Miller, 2001), modeling (Ervin, 2001; Gimblett et al., 2001; Perrin et al., 2001), analysis and visualization (Muhar, 2001; Oh, 2001; Schmid, 2001) of both natural and man-made landscapes. However, while the opportunity exists for the incorporation of landscape or scenic assessment data in a range of environmental planning scenarios, the demand for this information is poor and ill-defined (Moss and Nickling, 1989), especially in most of the developing countries. This raises the questions about whether existing visual assessment procedures have the capacity to capture the relevant visual (aesthetic) qualities of landscapes that may be the outcome of natural processes, conservation initiatives or management strategies (Crawford, 1994; Moss and Nickling, 1989; Thorne and Huang, 1991; Smardon and Fabos, 1983).

Methodologically, geographic information systems (GIS) have widely contributed to the advancement of studies that evaluate the evolution of ecological and social fabric of landscapes. However, there is an increasing interest in the use of spatial data and GIS in assessing visual attributes of the landscape. As Bishop and Hulse (1994) stressed, if visual values can be identified using mapped data and the computational capabilities of GIS, then there exists the potential for the development of more objective and cost-effective procedures for the assessment of visual attributes and the impact of development upon their qualities. Early attempts to adopt GIS techniques such as Steinitz (1990) not only provided an assessment base for visual resource management but also incorporated ecological values in planning for sustainable landscapes. GIS were also used to identify areas at risk of visual impact from harvest activities proposed in an actual management plan in Wisconsin by Bergen et al. (1993). Buckley and Berry (1997) reviewed the issues involved and the potential benefits of integrating scientific visualization techniques with a GIS package (Arc/INFO), including model sensitivity analysis, interpretation of model output as well as traditional visual impact assessment. Crawford (1994) compared GIS results with those of a manually conducted study, and Panagopoulos (2001) used three-dimensional GIS to visualize landscape management in Monchique, Portugal. The later proved that the use of three-dimensional imaging for characterization of environmental sites helped planners perceive the whole picture and make better and quicker decisions. Furthermore, Walsh et al. (1999) used GIS, remote sensing, and population surveys to examine behavioral, geographical and environmental hypotheses about population–environment interactions. Gimblett et al. (2001) simulated human– environment interaction by integrating GIS data with statistical analysis, visualization and computer modeling. In addition to GIS analysis, remotely sensed data, especially satellite images and vertical aerial photographs, have been widely used in mapping spatial phenomena. Those tools give a bird’s eye view of the landscape that facilitates the assessment and management of large areas in a timely manner. Planning decisions, in many cases, have to be applied on a regional scale, and, due to time and effort constraints, might not consider specific locations or structures. Both remote

Y.M. Ayad / Landscape and Urban Planning 73 (2005) 307–325

309

Fig. 1. The location of the main study area and the major physiographic features.

sensing data and GIS techniques applied to a regional level may provide a handy and timely decision support. This often implies the use of coarse resolution datasets such as Landsat TM (30 m) and SPOT (20 m). This brings up the level of generalization issue, which falls under what Ervin (2001) calls “abstraction level”, or the scale at which some information will be obscured or generalized.

2. Visual attributes of arid coasts The present study will deal with the broader image of an arid coastal landscape in Egypt (Fig. 1); at this scale, details of specific structures and coastal settings are obscured. Therefore, it is necessary to review previous research that studied visual attributes of similar landscapes. Earlier studies that dealt with the visual

310

Y.M. Ayad / Landscape and Urban Planning 73 (2005) 307–325

perception of coastal settings, and those with coastal structures, assessed visual and aesthetic impacts for water resource type projects (Smardon et al., 1988), and classified coastal structures as either positive or negative (Knuston et al., 1993; Smardon, 1987). On the other hand, the case study does not focus on the visibility of a specific project form a certain viewpoint, but rather evaluates the extent of visual change in a rapidly developing region in Egypt. The selected time frame focuses on changes that occurred between a period of time that was characterized by a vernacular, relatively natural landscape (1950s), and the beginning of the exploitation of the region for different economic purposes (1990s). The identification of relevant visual characteristics is somewhat problematic, especially in an arid coastal landscape where visual studies have never been carried out or considered for further management strategies. To overcome this problem, inclusion and/or exclusion of visual attributes relied on three main considerations: (1) their appropriateness to the geographic scale and extent of the study area, (2) the culture of its users (inhabitants and visitors), and (3) their importance as revealed by previous studies. Because these studies identified important visual attributes in landscapes where forests, grasslands, water bodies, and more rugged topography existed, the attributes were either excluded or modified to better fit the more arid landscape under consideration. However, it is advisable to use caution when drawing conclusions or future planning recommendations from this study without conducting a more thorough analysis of the relevant characteristics of the study area including its spatial attributes and the ways it is perceived by local inhabitants and visitors. Visual preference is dependent on the cultural background of users and their reason for visiting the area. In a study by Steinitz (1990) for ACADIA national park, visitors considered any developed or urbanized landscape, even tourist-oriented commercial developments, as negative factors; on the other hand, Cherem and Traweek (1977) stated that among the positive factors were developed recreation areas. Furthermore, coastal commercial developments were acceptable by people of certain ages in another study by Banerjee and Gollub (1976). Thus, in the present case study, the built environment should also be taken into consideration as a potentially favorable factor especially when com-

bined with other visual resources such as the natural landscape, agricultural fields and/or other factors that directly touch the benefits of the area visitors. The dominant land use of the area under investigation is vacation and holiday making. The visual attributes are primarily based on the preferences of users with dense urban-life experience (e.g. Cairo and Alexandria). The main attraction for these visitors is the beach. Therefore, proximity to the shoreline is a significant factor. Also the type of accommodation and utilities that ensure good services are important. Many of the newly-built summer resorts in the area have mixed green areas integrated with other structures. This creates balanced open spaces that might be visually appealing and entertaining, but the created environment is ecologically incompatible with the original landscape. While this may not directly influence the visual experience of certain visitors, sustainable regional planning ought to be concerned about the environmental and ecological integrity of the landscape. In this sense, other environmental and ecological dictates should affect the selection of the visual attributes as well. This would take into consideration the interaction between humans and the environment in order to ensure the well being of a landscape that is sensitive to human use, as any damage or change to its environment is irreversible. 2.1. Visual attribute selection A review of specific landscape attributes identified in previous research was conducted. There is considerable agreement in this literature about scenic attributes (Fabos et al., 1978; Williamson, 1979; Mooney, 1983; Smardon and Fabos, 1983; Musick and Grover, 1991; Brown, 1994; Crawford, 1994). Williamson (1979) and Crawford (1994), for example, both point out that there is strong evidence to support the use of particular landscape attributes in the assessment of visual changes. They suggest that scenic quality increases as: (1) topographic ruggedness and relative relief increase, (2) the presence of water forms, water edge and water areas increase, (3) patterns of grasslands and forests become more diverse, (4) natural landscapes increase and man-made landscapes decrease, and (5) land use compatibility increases and land use edge diversity decreases. Accordingly, in order to identify scenic elements that might affect the scenic quality

Y.M. Ayad / Landscape and Urban Planning 73 (2005) 307–325

of the northwestern coastal landscape of Egypt, specific visual attributes were selected. Those attributes were identified according to their significance to the objectives of the study and their ability to describe and to reveal certain visual qualities specific to the landscape under investigation. Specifically, four visual attributes were selected for the assessment: diversity, which is a measure of land use/land cover variety; activity, which describes the “degree of naturalness” of the land cover; water proximity, which describes how close an area is to the shoreline, and topographic variety, which is a measure of relative variations in terrain relief. The first two attributes, namely diversity and activity, were based mainly on the land use/land cover information, which were extracted from classified remotely sensed data (satellite imageries and aerial photographs). The other two attributes, on the other hand, were derived from contour lines and digital elevation models (DEM) of the area under investigation. 2.2. Study area specifics Previous studies suggest that scenic quality increases as the patterns of natural landscape (grasslands, tree cover, water bodies, etc.) become more diverse, but in the present study all land uses were included in the diversity index calculation. This stemmed from the fact that the studied area is highly dependent on summer resort/vacation-oriented activities, and therefore, one of the main interests of the holidaymakers and visitors is to find comfortable and proper accommodations, without which their visiting experience would be greatly affected. Therefore, all structures were included in the calculations. Making an assumption that, the diversity index means that a more diverse landscape, including some artificial areas, would be more visually favorable for holiday makers than less diverse and all natural areas. On the other hand, the adopted activity index takes into consideration the compatibility of the land use in terms of natural, semi-natural, artificial, or any combination between them. This classification respects the level of variation between the introduced land use and the original natural landscape. A “landscape compatibility” index was not considered in this study because of the difficulty in differentiating between the types of structures in a built up area (i.e. residential, industrial,

311

high-rise, low-rise, etc.) with the limited spatial resolution of the satellite imagery used. The present study classifies the land use in terms of aggregated activities (natural/artificial), and therefore measures the degree of naturalness. Furthermore, although water proximity and the topography of the study area did not evidently change through the period of study, both were considered as possible visual attributes that would give an idea about the collective visual condition at each given date. Finally, a composite visual index was developed to take into consideration different combinations of all four attributes.

3. Case study In Egypt, the Nile valley has concentrated human settlement, since for centuries it has represented an attraction for agricultural activities and formed a favorable environment for human occupation. Today, the populated regions represent only 4% of the country’s total area. Population density is increasing dramatically in the Nile Delta and valley and now exceeds 1200 person/km2 . For this reason, the nation is paying considerable attention to the development of Egyptian deserts in order to redistribute the population and to release the intense pressure on the cultivable land of the Nile valley. The demand for alternative productive land is always urgent because of this high rate of population growth. In order to cope with such demands, planning strategies implemented short-term solutions aiming at producing higher economic profit for the nation, but these strategies were based upon a poor understanding of environmental and social realities. This resulted in devastating effects on the natural resources and the social fabric in many places. The northwestern coast and desert were no exception; decision-makers saw this region as the future habitation of more than four million, with economic development varying from agriculture (rain-fed, or irrigated), pasturing, fishing, tourism, oil exploration, and petrochemical industries; they aimed to increase the total national and regional income by increasing the economic base of the nation, and to redistribute the large population concentrations existing in the metropolitan areas by creating new social and economic attractions in the region.

312

Y.M. Ayad / Landscape and Urban Planning 73 (2005) 307–325

One of the major and obvious changes that occurred along the northwestern coast of Egypt was the construction of summer resorts with their consequent and continued quarrying and leveling of the landscape. Quarrying practices involve the cutting of the ridges to make brick for use in resort construction. The natural landscape is therefore obliterated beyond recognition. The process destroys natural vegetation and creates an environment that contrasts with the less disturbed areas around it. Multistory buildings are now obscuring the view to the Mediterranean and therefore destroying aesthetic values and the visual experience of travelers along the coast. Furthermore, it is notable that studies concerned with the aesthetic importance of the landscape are completely missing from the planning process in Egypt in general and in the western coastal region specifically. The aesthetic elements of the landscape were always thought to be of secondary importance to the management and planning schemes, or were never considered as national resources.

4. The study area The northwestern coastal zone of Egypt, where the study area is located, may be divided into two main physiographic provinces: an eastern province between Alexandria and Ras El-Hikma (about 230 km west of Alexandria) and a western province between Ras ElHikma and Salloum at the Libyan border (Fig. 1). The landscape is divided into a northern coastal plain and a southern tableland (Salem, 1989). The coastal plain is wide in the eastern province (where the study area is located), and is characterized by alternating ridges and depressions running in a nearly east–west direction. The ridges vary in altitude and are dissected by many shallow erosion valleys. Some of these valleys discharge water into the Mediterranean, the others into depressions (Ayyad, 1978). The ridges are formed of limestone with a hard crystallized crust, and vary in altitude and lithological features according to age. The most prominent are the coastal, Abu-Sir, and GebelMariut ridges. The Abu-Sir ridge is separated from the coastal ridge by a depression with a mean surface elevation of five meters above sea level and a width that varies between 300 m and 1 km. It is filled with calcareous formations, highly saline in places and formed

almost totally of oolitic grains in certain localities. Between Abu-Sir and Gebel-Mariut ridges is the Mallahat Mariut depression. It has a width varying between 2 and 5 km with the surface mostly below sea level, and is filled mainly with brackish water and saline calcareous deposits of weathered and downwash material (Salem, 1989). The study area is a part of the rapidly developing northwestern coastal zone of Egypt, a strip of land that averages about 15 km in depth from the Mediterranean shoreline, and in some places exceeds 30 km. It is located between longitudes 29◦ 23 E and 29◦ 33 E and between latitudes 30◦ 49 N and 30◦ 58 N (Fig. 1). The pilot study area is located about 45 km west of Alexandria and extends about 19 km westward along the shoreline with a width that varies between 0.9 and 1.5 km inland (Fig. 2). The width of the pilot study area was selected based on an area bounded by the shoreline from the north and the contour line 20 m above sea level from the south, which represents the highest point on the first rocky ridge. This area accommodates the majority of holiday making and beach tourism. The physiography of the region physically, and therefore visually, separates the coastal strip from the remaining parts further inland by the series of ridges and depressions. Areas lying further inland are rarely used by visitors for beach and summer resorts. But it is important to note that the study area is part of a larger region that is exposed to heavy exploitation and other economic activities such as irrigated agriculture, industry, quarrying, and expanding urbanization.

5. Methods 5.1. Stage I: data preparation The selection of the datasets was based on detecting the changes that occurred between a period of time that was characterized by a vernacular, relatively natural landscape, and the beginning of the exploitation of the region for different economic purposes. The available data used in this study consists of SPOT XS satellite images acquired on 10 April 1987 and 2 September 1992 (three spectral bands of 20 m × 20 m spatial resolution), aerial photographs of scale 1:25,000 for September 1955 and November 1977, and a 1:50,000 topographic map of 1977.

Y.M. Ayad / Landscape and Urban Planning 73 (2005) 307–325

313

Fig. 2. The selected coastal strip.

The digital elevation model (DEM) was obtained by digitizing the 5 m interval contours lines from the topographic map. The digitized vector lines were then interpolated to a DEM with a 20 m raster cell size to match the SPOT dataset. This step was carried out using Arc/INFO 8.2 GRID module. To map the land use/land cover, the satellite images were classified using ERDAS IMAGINE 8.2. Both supervised and unsupervised techniques were applied for the 1987 and the 1992 images, respectively. The average overall accuracy was 87.14% for the first, and 90.25% for the second. Furthermore, the aerial photographs were scanned, and interpreted. Land use/land cover vector polygons were created, and then converted to raster to match the SPOT dataset. The final land use/land cover classification is outlined in Table 1.

Table 1 The identified land use classes at all dates (1955, 1977, 1987 and 1992) and their corresponding ID Class ID

Class name

Description

1 2 3 4

Dunes Crops Orchards Ploughed

5

Background

6

Urban

7 8

Flooded Salt marsh vegetation

Coastal sand dunes Crop-cultivated land Orchard plantation Ploughed fields prepared for agriculture Open land, including pasture land and sparse natural vegetation Urbanized zones including built-up and the surrounding areas Inundated salt marsh areas All salt marsh vegetation classes

314

Y.M. Ayad / Landscape and Urban Planning 73 (2005) 307–325

5.2. Stage II: data modeling and attribute computation Visual attributes were calculated in Arc/INFO GRID module for each date, and Arc Macro Language (AML) routines were written to automate the different steps of each procedure. Two visual characteristics were calculated: land use/land cover activity, by which the degree of naturalness was identified, and the land use/land cover diversity, by which the proportional distribution of different classes was calculated. Figs. 3 and 4 summarize the method and the steps of calculating each of those attributes.

Three major classes of activities were recorded (Table 2), by which the degree of naturalness was identified according to the proportional distribution of its corresponding land use classes in a block of 100 m × 100 m (5 × 5 cells) (Fig. 3). The proportional distribution was calculated using the following equation: 100  Ni x where PPi is the percentage of the proportional distribution of activity i, INT is a function that transforms the result into integer values, x the total number of cells in each block (5 × 5 cells for the present case), Ni the activity of identification i.

PPi = INT

Fig. 3. Method to extract the land use/land cover activity (degree of naturalness) classes.

Fig. 4. Method to extract the land use/land cover diversity classes.

Y.M. Ayad / Landscape and Urban Planning 73 (2005) 307–325 Table 2 The selected activity group of classes ID

Activity

Class names

Class ID

3

Natural

Dunes Background classes Flooded areas Salt marsh vegetation

1 5 7 8

2

Semi-natural

Crops Orchards Ploughed fields

2 3 4

1

Artificial

Urban areas

6

The final output was produced to represent the existing land uses in the study area with their corresponding ID of “degree of naturalness” (Table 3). Each resulting class was thus given a specific score that describes its importance (Crawford, 1994). Subsequently, due to the nature of the arid landscape under investigation and the scarce distribution of land use/land cover classes, the Shannon diversity index (Shannon and Weaver, 1949) was adopted. Table 3 The proposed combinations of the activity classes ID

Meaning

Classification

3

Only natural classes

3 ≥ 68

2

Only semi-natural classes

2 ≥ 68

Combination between natural and semi-natural classes

1 < 16

23

16 ≤ 2 < 68 16 ≤ 3 < 68 123

Equal share between all classes (natural, semi-natural, and artificial)

16 ≤ 1 < 68 16 ≤ 2 < 68 16 ≤ 3 < 68

13

Combination between artificial and natural classes

2 < 16

315

It is a popular measure of diversity in community ecology, and is more sensitive to rare land use/land cover. The value of this index represents the amount of “information” per individual land use/land cover class. Germino et al. (2001) is an example of previous studies in which ecological diversity indices were applied to visual landscape analysis. Therefore, as presented in Fig. 4, the Shannon diversity index (SHDI) (Rosenzweig, 1995; Morain, 1999) was calculated for the same 100 m × 100 m block according to the following equation:  SHDI = − Pi ln Pi where SHDI is the Shannon diversity index, i the class, Pi the proportional distribution of a class i in a specific study area. The value of Pi always varies between 0 and 1; therefore the logarithm of Pi will always be a negative value (except for 0 which will give infinity, and for 1 which will give 0). The presence of the negative sign outside the summation function is needed in order to maintain a positive result for the index. A SHDI is equal to 0 means that there is no diversity within the 100 m × 100 m block (only one class exists), and the greater the value of the index the more diverse the site. The proportional distribution was calculated for the 100 m × 100 m block for each class separately, and the first part of the SHDI equation (i.e. Pi ln Pi ) was therefore calculated. The computed results were then summed to produce one grid that contains the values of the SHDI. The cell size of the resulting grid is 100 m × 100 m. These values were therefore classified in order to represent four IDs of diversity. The final classification for the diversity is shown in Table 4. Furthermore, the topographic variations where derived from the number of elevation values contained in a 100 m × 100 m block (5 × 5 cells). Data were re-

16 ≤ 1 < 68 16 ≤ 3 < 68 12

1

Combination between artificial and semi-natural classes

Only artificial classes

3 < 16

Table 4 Different diversity classes extracted for the study area

16 ≤ 1 < 68 16 ≤ 2 < 68

ID

Description

SHDI

0 1 2 3

No diversity Low diversity Intermediate diversity High diversity

0 0 < SHDI ≤ 0.5 0.5 < SHDI ≤ 1 SHDI > 1

1 ≥ 68

The classification is based on the percentage proportional distribution of each activity class within a 5 × 5 cells block.

Y.M. Ayad / Landscape and Urban Planning 73 (2005) 307–325

316

Table 6 Classes of the different proximity distances to the shoreline

Table 5 Topographic variation classes ID

Description

Number of classes in Block

ID

Distance (m)

1 2 3 4

No variety Low variety Medium variety High variety

<4 4 to <8 8 to <13 >13

1 2 3 4 5

>1000 750–1000 500–750 250–500 0–250

classified to represent four classes of topographic variations as shown in Table 5. The proximity of the cells to the shoreline was then calculated and classified into five categories according to the distances presented in Table 6. The categories

Class description Furthest from the shoreline Far from the shoreline Midway Close from the shoreline Closest from the shoreline

were coded from 1 to 5, where 5 being the closest to the shoreline and 1 being the furthest. Finally, the composite visual index was calculated according to the criteria listed in Table 7 . The final classification of the visual index was based on the degree

Table 7 Selection scheme of the proposed composite visual index Code

Degree of naturalness

Land cover diversity

Topographic variety

Shoreline proximity

1

Natural Semi Natural/semi

High Intermediate

High Intermediate

Closest Close

2

Natural Semi-natural Natural/semi-natural

High Intermediate

High Intermediate

Intermediate

3

Natural Semi-natural Natural/semi-natural

Low None

High Intermediate

Closest Close

4

Natural Semi-natural Natural/Semi-natural

High Intermediate

Low None

Closest Close

5

Natural Semi-natural Natural/semi-natural

Low None

Low None

Closest Close

6

Natural Semi-natural Natural/semi-natural

Low None

High Intermediate

Intermediate

7

Natural Semi Natural/semi

High Intermediate

Low None

Intermediate

8

Natural Semi Natural/semi

Low None

Low None

Intermediate

9

Natural Semi Natural/semi

High Intermediate

High Intermediate

Far Furthest

10

Natural Semi Natural/semi

Low None

High Intermediate

Far Furthest

Y.M. Ayad / Landscape and Urban Planning 73 (2005) 307–325

317

Table 7 (Continued ) Code

Degree of naturalness

Land cover diversity

Topographic variety

Shoreline proximity

11

Natural Semi Natural/semi

High Intermediate

Low None

Far Furthest

12

Natural Semi Natural/semi

Low None

Low None

Far Furthest

13

Equal share Artificial/natural Artificial/semi-natural

High Intermediate

High Intermediate

Closest Close

14

Equal share Artificial/natural Artificial/semi-natural

High Intermediate

High Intermediate

Intermediate

15

Equal share Artificial/natural Artificial/semi-natural

Low None

High Intermediate

Closest Close

16

Equal share Artificial/natural Artificial/semi-natural

High Intermediate

Low None

Closest Close

17

Equal share Artificial/natural Artificial/semi-natural

Low None

Low None

Closest Close

18

Equal share Artificial/natural Artificial/semi-natural

Low None

High Intermediate

Intermediate

19

Equal share Artificial/natural Artificial/semi-natural

High Intermediate

Low None

Intermediate

20

Equal share Artificial/natural Artificial/semi-natural

Low None

Low None

Intermediate

21

Equal share Artificial/natural Artificial/semi-natural

High Intermediate

High Intermediate

Far Furthest

22

Equal share Artificial/natural Artificial/semi-natural

Low None

High Intermediate

Far Furthest

23

Equal share Artificial/natural Artificial/semi-natural

High Intermediate

Low None

Far Furthest

24

Equal share Artificial/natural Artificial/semi-natural

Low None

Low None

Far Furthest

25

Artificial

High Intermediate

High Intermediate

Closest Close

26

Artificial

High Intermediate

High Intermediate

Intermediate

Y.M. Ayad / Landscape and Urban Planning 73 (2005) 307–325

318 Table 7 (Continued ) Code

Degree of naturalness

Land cover diversity

Topographic variety

Shoreline proximity

27

Artificial

Low None

High Intermediate

Closest Close

28

Artificial

High Intermediate

Low None

Closest Close

29

Artificial

Low None

Low None

Closest Close

30

Artificial

Low None

High Intermediate

Intermediate

31

Artificial

High Intermediate

Low None

Intermediate

32

Artificial

Low None

Low None

Intermediate

33

Artificial

High Intermediate

High Intermediate

Far Furthest

34

Artificial

Low None

High Intermediate

Far Furthest

35

Artificial

High Intermediate

Low None

Far Furthest

36

Artificial

Low None

Low None

Far Furthest

of naturalness in the first place, and then combinations of the land use/land cover diversity, proximity to the shoreline, and the topographic variety.

6. Results and discussion The change in the degree of naturalness from 1955 to 1992 was evident especially on the coastal sand dunes and further inland on the slopes and at the top of the first rocky ridge (Fig. 5). The artificial and artificial/natural classes increased by 20 and 11%, respectively. On the other hand, semi-natural and semi-natural/natural classes started with a slow increase between 1955 and 1977 (from 6 to 10 and 5 to 6%, respectively) then almost disappeared in 1992. The Natural classes decreased gradually from 86% in 1955 to 68% in 1992 a total of 18% decrease in the period under investigation. The decreases in natural, semi-natural and combination classes of semi-natural/natural was in favor of both artificial and artificial/natural combinations which indicates an abrupt change of

the landscape substituting its intermediate activity compositions with more artificial activities. The vernacular landscape that existed in the area in the form of small-scale rain-fed agriculture and fig plantations in the period between 1955 and 1987 (as shown in Fig. 6a in semi-natural classes) totally disappeared in favor to the artificial activities such as summer resorts on the coastal sand dunes and other developments further inland. This describes a period of time following two decades of war where people tended to migrate to the region but still the influence of the national government plans for future development were not yet obvious. In 1977 it is clear that local inhabitants began to abandon their traditional way of life for a more industrialized one. Furthermore, land use/land cover diversity showed a general trend toward the transformation of the landscape into more homogenous compositions (Fig. 7). In 1987 the highly diversified areas reached their relatively highest values (about 6%) then started to decline in 1992 (Fig. 6b). This can be attributed to the beginning of the heavy construction period in 1987, which

Y.M. Ayad / Landscape and Urban Planning 73 (2005) 307–325

319

Fig. 5. The extracted land use/land cover activity (degree of naturalness) classes for the given dates.

replaced the previous landscape with more of the same structured summer resorts. This was also reflected in the continuous decline of the areas of no diversity (from 65% in 1955 to 42% in 1992). Finally, both intermediate and low diversity areas slightly increased during the period under investigation. Although the more homogenous areas indicate a transition to a one land cover/land use class, the resolution of the satellite imagery and the 5 × 5 cell blocks might obscure internal variations in summer resorts, other structures and finer land cover/land use classes. Many of the summer resorts include other natural and semi-natural landscape features such as grasslands, trees and artificial water bodies, which,

if considered, might increase their visual appreciation. Furthermore, the proposed composite visual index, which takes into consideration all four landscape attributes as shown in Table 7, creates three main groups of visual resources ranked primarily by the degree of naturalness. The first group, which values are from 1 to 12, represents the natural, semi-natural and natural/semi-natural classes with varied combinations of land cover/land use diversity, topographic variety and shoreline proximity. The second group, on the other hand, which values are from 13 to 24, represents the same combinations with equal share, artificial/natural, and artificial/semi-natural degree of natu-

320

Y.M. Ayad / Landscape and Urban Planning 73 (2005) 307–325

Fig. 6. (a) Changes of the land use/land cover activity (degree of naturalness) classes between 1955 and 1992. Areas classified as “natural” were 86% in 1955, 82% in 1977, 74% in 1987, and 68% in 1992. (b) Changes of the land use/land cover diversity classes between 1955 and 1992. (c) Changes in the calculated composite visual index between 1955 and 1992 for the three main groups.

Y.M. Ayad / Landscape and Urban Planning 73 (2005) 307–325

321

Fig. 7. The extracted land use/land cover diversity classes for the given dates.

ralness classes. Finally, the third group, which values are from 25 to 36, represents the same combinations with only the artificial activity class. Basing the classification primarily on the degree of naturalness showed a notable change in many categories such as 5, 8, 10 and 12, which decreased from 31.7 to 17.7, 14.9 to 10.7, 16.9 to 9.6 and 9.8 to 6.5%, respectively (Fig. 6c). On the other hand, the smaller portion of the landscape under investigation had mixed and artificial classes, among which an increase of about 8% in category 29 was clear, which reflects the opposite change of category 5. This reveals the transformation from all natural combinations to all artificial classes with both low diversity and topography but closer to the shoreline. The presented method did not detect any

variation in the artificial classes closer to the shoreline (Fig. 8). This may be attributed to either the absence of higher land cover diversity that combines other natural classes, or the coarse spatial resolution of the used imageries. Other changes that could not be directly related to the degree of naturalness were detected. Categories 21 and 33 have increased by 4 and 2%, respectively. This increase is accompanied with a decrease in other diversity, topography and proximity to the shoreline combinations. There may be difficulties interpreting such results using the current representation. Other statistical methods for interpretation might be useful in tracking the changes in each category and assessing the omitted and committed values between categories.

322

Y.M. Ayad / Landscape and Urban Planning 73 (2005) 307–325

Fig. 8. The distribution of the calculated composite visual index for the given dates.

7. Conclusions The present study is an attempt to deploy remotely sensed data and GIS technology in modeling changes in the visual attributes of an arid coastal landscape. After the extraction of land use/land cover classes from the satellite images and the aerial photographs, four visual attributes of landscape were extracted: land use/land cover diversity, activity (degree of naturalness), proximity to the shoreline, and the topographic variety. Although those attributes and the suggested composite index relied mostly on the type of land cover/land use information on the landscape under consideration, the suggested techniques were able to detect several

changes in each attribute, spatially locate them and map the magnitude of their changes. This suggests that much could be learned from further studies evaluating the importance of aesthetic resources to the planning and management processes in the landscape under investigation specifically and in arid coastal landscapes of Egypt in general. Visual assessments in particular need to be taken into consideration by the Egyptian governmental agencies in developing strategies that involve outstanding visual values. The western Mediterranean region, like other areas of scenic attraction (e.g. Sinai and the Red Sea coastal land), is economically dependent upon the tourist trade. In this case, the identification, description and analysis

Y.M. Ayad / Landscape and Urban Planning 73 (2005) 307–325

of visual characteristics are important steps in understanding how these development strategies may affect such characteristics. The present study incorporated some aesthetic aspects in the analysis of the landscape. It demonstrated what could be done to analyze and assess an important but elusive resource. The tourism resource base may be compromised by failure to integrate scenic values into resource management decision-making. Yet to the best of the author’s knowledge, the Egyptian government has undertaken no studies, employed no personnel nor adapted any specific methodologies for undertaking aesthetic evaluation. Although ecological arguments might provide the most satisfying and fundamental basis on which to plan new landscapes, it is evident that, in practice, programs should also be justified in terms of their aesthetic potential and be implemented by professionals whose principal training is in the discipline of landscape planning. The methodology employed here could easily be extended to include other visual attributes, such as color, water bodies, water body edges, visibility key points and visual accessibility, as well as non-visual attributes, such as smell, texture and even symbolic values. The future exploration of pertinent attributes for arid coast landscapes in general and those specific to the northwestern coast of Egypt are definitely encouraged. At the perceptual level, people who live in and use the landscape must be involved in the attribute selection. Many methods were introduced by several authors and ought to be considered as a basis for future work in the region under investigation (e.g. Hands and Brown, 2002; Kaltenborn and Bjerke, 2002; Lange, 2001; Langers and Goossen, 2000; Nohl, 2001; Smardon, 1987; Steinitz, 1990, 2001). Future visual perception studies should be conducted in order to extract those attributes that can refine the present study’s results. A variation in results is expected between visitors and inhabitants since residents might be more tolerant of coastal development if economic income is generated (Zube and McLaughlin, 1978) and visitors, depending on their aim (holiday making and summer resorts in the present study), might be tolerant of some coastal development that will ensure a level of service to accommodate their stay but not alter the beach landscape. This constitutes an important aspect that needs further exploration as aesthetic values and socio-economic benefits might contradict with the eco-

323

logical sustainability of the landscape (Parsons, 1995), especially when more green areas, tree plantations, and artificial water bodies (e.g. marinas, pools, ponds, etc.) are introduced. This calls for studies that interrelate visual, ecological, and socio-economic attributes to reveal the significance of changes that occur in each and to study the correlation between them and therefore achieve a higher level of landscape understanding and a good basis for planning future development. A basis for such theories was presented by Thorne and Huang (1991). It is therefore important to note that awareness of the value of the visual qualities of the arid coastal landscape needs to be promoted. The population’s perception of the landscape and its cultural interactions with spatial visual peculiarities and with changes in the physical land formations need to be evaluated. Furthermore, the continuously evolving technologies of satellite image processing and aerial photography interpretation can be adapted to the study of the evolution of the visual attributes of the arid coastal landscape, especially at a regional scale. On the other hand, at a finer scale, field surveys using GPS and handheld computers might be considered to improve data collection. Finally, lessons can be drawn from the present study. Future development planning strategies for similar areas should accommodate the visual resources of arid coastal landscape. Irreversible acts can damage not only the aesthetic component but may also have drastic effects on the ecological well-being of arid coasts. Consequently, this study showed that as the northwestern coast of Egypt has undergone extensive development, unplanned or ill-planned activities have altered landscape composition in ways that will be obvious for many decades and that have deeply impacted one of the most fragile resources of arid landscapes: their visual appeal.

Acknowledgment Important parts of this research were carried out as part of the author’s Ph.D. thesis at the University of Montreal. Special thanks goes to the author’s advisor, Dr. Michele Guenet, and co-advisor, Dr. Gerald Domon, whose contributions provided the author with the ability to develop the methodologies presented in this article. The Ph.D. thesis was supported

324

Y.M. Ayad / Landscape and Urban Planning 73 (2005) 307–325

by the Canadian International Development Agency’s Programme Canadien de Bourses de la Francophonie. Thanks also to the Department of Environmental Studies, Alexandria University, Egypt, for providing the satellite images, aerial photographs, and maps that were used in this paper.

References Altman, I., Chemers, M., 1980. Culture and Environment. Brooks/Cole Publishing Company. Ayyad, M.A., 1978. Regional environmental management of Mediterranean desert ecosystems of northern Egypt (REMDENE). A Project Proposal Submitted to the US Environmental Protection Agency for Support. Mimeo. Banerjee, T., Gollub, J., 1976. The public view of the coast: toward aesthetic indicators for coastal planning and management. In: Suefeld, P., Russel, J.A. (Eds.), Proceedings of the 7th International Conference of EDRA, Vancouver, BC. Bishop, I.D., Ye, W.-S., Karadaglis, C., 2001. Experiential approaches to perception response in virtual worlds. Landscape Urban Plan. 54, 115–123. Bishop, I.D., Hulse, D.W., 1994. Prediction of scenic beauty using mapped data and geographic information systems. Landscape Urban Plan. 30, 59–70. Bergen, S.D., Fridley, J.L., Ganter, M.A., Schiess, P., 1993. Predicting the visual effect of forest operations. J. For. 93 (2), 33–37. Brown, T., 1994. Conceptualizing smoothness and density as landscape elements in visual resource management. Landscape Urban Plan. 30, 49–58. Buckley, D.J., Berry, J.K., 1997. Integrating advanced visualization techniques with Arc/INFO for forest research and management. In: Proceedings of the 1997 ESRI International User Conference, San Diego, CA. Cherem, G.J., Traweek, D.E., 1977. Proceedings of the River Recreation Management and Research, Saint Paul, MN. Visitor employed photography: a tool for interpretive planning on river environments. Crawford, D., 1994. Using remotely sensed data in landscape visual quality assessment. Landscape Urban Plan. 30, 71–81. Danahy, J.W., 2001. Technology for dynamic viewing and peripheral vision in landscape visualization. Landscape Urban Plan. 54, 125–137. Dearden, P., Sadler, B., 1989. Landscape evaluation: approaches and applications. Westernal Geographical Series, vol. 25. University of Victoria. Ervin, S.M., 2001. Digital landscape modeling and visualization: a research agenda. Landscape Urban Plan. 54, 49–62. Fabos, J.Y., Greene, C.M., Joyener Jr., S.A., 1978. The METLAND landscape planning process: composite landscape assessment, alternative plan formulation and plan evaluation. Part 3 of the Metropolitan Landscape Planning Model. Massachusetts Agricultural Experiment Station and U.S. Department of Interior Office of Water Research and Technology.

Germino, M.J., Reiners, W.A., Blasko, B.J., McLeod, D., Bastian, C.T., 2001. Estimating visual properties of rocky mountain landscapes using GIS. Landscape Urban Plan. 53, 71–83. Gimblett, R., Daniel, T., Cherry, S., Meitner, M.J., 2001. The simulation and visualization of complex human–environment interactions. Landscape Urban Plan. 54, 63–78. Hands, D.E., Brown, R.D., 2002. Enhancing visual preference of ecological rehabilitation sites. Landscape Urban Plan. 58, 57–70. Hehl-Lange, S., 2001. Structural elements of the visual landscape and their ecological function. Landscape Urban Plan. 54, 105–113. Itami, R., 1989. Scenic perception: research and application in U.S. visual management systems. In: Dearden, P., Sadler, B. (Eds.), Landscape Evaluation: Approaches and Applications. University of Victoria. Jakle, J.A., 1987. The Visual Elements of Landscape. The University of Massachusetts Press. Kaltenborn, B.P., Bjerke, T., 2002. Associations between environmental value orientations and landscape preferences. Landscape Urban Plan. 59, 1–11. Knuston, M.G., Leopold, D.J., Smardon, R.C., 1993. Selecting islands and shoals for conservation based on biological and aesthetic criteria. Environ. Manage. 17, 199–210. Lange, E., 2001. The limits of realism: perceptions of virtual landscapes. Landscape Urban Plan. 54, 163–182. Langers, F., Goossen, M., 2000. Assessing quality of rural areas in The Netherlands: finding the most important indicators for recreation. Landscape Urban Plan. 46, 241–251. Litton, Burton, R., Tetlow, R.J., Sorensen, J., Beatty, R.A., 1974. Water and landscape: an aesthetic overview of the role of water in the landscape. Water Information Center Inc. Miller, D., 2001. A method for estimating changes in the visibility of land cover. Landscape Urban Plan. 54, 91–104. Mooney, M.B., 1983. Classifying visual attributes of wetlands in the St. Laurence-Eastern Ontario region. In: Smardon, R.C. (Ed.), The Future of Wetlands: Assessing Visual-cultural Values. Osmun Publishers, Allanheld, pp. 99–117. Morain, S., 1999. GIS Solutions in Natural Resources Management: Balancing the Technical-political Equation. OnWord Press. Moss, M.R., Nickling, W.G., 1989. Environmental and policy requirements: some Canadian examples and the need for environmental process assessment. In: Dearden, P., Sadler, B. (Eds.), Landscape Evaluation: Approaches and Applications. University of Victoria. Muhar, A., 2001. Three-dimensional modeling and visualization of vegetation for landscape simulation. Landscape Urban Plan. 54, 5–17. Musick, H.B., Grover, H.D., 1991. Image textural measures as indices of landscape pattern. In: Turner, M.G., Gardner, R.H. (Eds.), Quantitative Methods in Landscape Ecology: The Analysis and Interpretation of Landscape Heterogeneity. SpringerVerlag, New York, pp. 77–103. Naveh, Z., 1995. Interactions of landscapes and cultures. Landscape Urban Plan. 32, 43–54. Nohl, W., 2001. Sustainable landscape use and aesthetic perceptionpreliminary reflections on future landscape aesthetics. Landscape Urban Plan. 54, 223–237.

Y.M. Ayad / Landscape and Urban Planning 73 (2005) 307–325 Oh, K., 2001. Landscape information system: a GIS approach to managing urban development. Landscape Urban Plan. 54, 79–89. Opie, J., 1979. Seeing desert as wilderness and as landscape—an exercise in visual thinking. In: Proceedings of Our National Landscape: A Conference on Applied Techniques for Analysis and Management of the Visual Resource, Nevada, USA. Panagopoulos, T., 2001. Visual landscape management and visual impact assessment in Monchique, Portugal. In: Proceedings of the International Conference on Forest Research: A Challenge for an Integrated European Approach, Thessaloniki, Greece. Parsons, R., 1995. Conflicts between ecological sustainability and environmental aesthetics: conundrum, can¨ard or curiosity. Landscape Urban Plan. 32, 227–244. Perrin, L., Beauvais, N., Puppo, M., 2001. Procedural landscape modeling with geographic information: the IMAGIS approach. Landscape Urban Plan. 54, 33–47. Rosenzweig, M.L., 1995. Species Diversity in Space and Time. Cambridge University Press, New York, NY. Salem, B.B., 1989. Remote sensing of vegetation and land use in the northwestern desert of Egypt. Ph.D. Thesis. Faculty of Science, Alexandria University, Egypt. Schmid, W.A., 2001. The emerging role of visual resource assessment and visualization in landscape planning in Switzerland. Landscape Urban Plan. 54, 213–221. Shannon, C.E., Weaver, W., 1949. The Mathematical Theory of Communication. University of Illinois Press, Urbana, IL. Smardon, R.C. (Ed.), 1983. The Future of Wetlands: Assessing Visual-cultural Values. Allanheld, Osmun and Co. Publishers Inc.. Smardon, R.C., 1987. Visual access to 1000 lakes. Landscape Architect. 77 (3), 86–91. Smardon, R.C., Fabos, J.Y., 1983. A model for assessing visualcultural values of wetlands: a Massachusetts case study. In: Smardon, R.C. (Ed.), The Future of Wetlands: Assessing Visualcultural Values, pp. 149–170. Smardon, R.C., Palmer, J.F., Knopf, A., Grinde, K., 1988. Visual resources assessment procedure for the U.S. army Corps of Engineers. Instruction Report No. EL-88-1. Department of the Army, U.S. Army Corps of Engineers, Washington, DC.

325

Steinitz, C., 1990. Toward a sustainable landscape with high visual preference and high ecological integrity: the loop road in Acadia National Park, USA. Landscape Urban Plan. 19, 213–250. Steinitz, C., 2001. Visual evaluation models: some complicating questions regarding memorable scenes. Landscape Urban Plan. 54, 283–287. Thorne, J.F., Huang, C., 1991. Toward a landscape ecological aesthetic: methodologies for designers and planners. Landscape Urban Plan. 21, 61–79. Walsh, S.J., Entwisle, B., Rindfuss, R.R., 1999. Landscape characterization through remote sensing, GIS and population surveys. In: Morain, S. (Ed.), GIS Solutions in Natural Resource Management: Balancing the Technical-political Equation, 1st ed. OnWord Press, Santa Fe, pp. 251–265. Williamson, D., 1979. Scenic perceptions of Australian landscapes. Landscape Aust. 1 (2), 94–101. Zhang, Z., Tsou, J.Y., Lin, H., 2000. GIS for visual impact assessment. In: Proceedings of the 21st Asian Conference on Remote Sensing, Taipei, Taiwan. Zube, E.H., Brush, R.O., Fabos, J.Y. (Eds.), 1975. Landscape Assessment: Values, Perceptions, and Resources. Hutchinson and Ross Inc., Dowden. Zube, E.H., McLaughlin, M., 1978. Assessing perceived values of the coastal zone. In: Coastal Zone’78: Proceedings of the Symposium on Technical, Environmental, Socioeconomic and Regulatory Aspects of Coastal Zone Management, San Francisco, CA, March 14–16. American Society of Engineers, New York.

Yasser M. Ayad is currently Assistant Professor of GIS, Clarion University of Pennsylvania, Department of Anthropology, Geography and Earth Science, Clarion, PA, USA. He received his Ph.D. degree in environmental planning in 2000 from the University of Montreal, Canada, M.Sc. degree in environmental studies in 1993 from the University of Alexandria, Egypt and B.Sc. degree in architectural engineering in 1988 from the University of Alexandria, Egypt. His areas of interest are GIS, Landscape Ecology, Landscape Indices, Regional and Urban Planning.