ITU A|Z • Vol 12 No 1 • March 2015 • 181-203 Scenario-based land use estimation: The case of Sakarya

Fatih TERZİ [email protected] • Department of Urban and Regional Planning, Faculty of Ar- chitecture, Technical University, Istanbul,

Received: July 2014 Final Acceptance: January 2015

Abstract This paper presents scenario-based modelling of urban land use which stem from interactions of the urban functions of Sakarya, one of the most important city for agricultural production and a vulnerable region in terms of natural re- sources and seismicity in Turkey. The purpose of the paper is to estimate the fu- ture land use pattern of Sakarya and discuss the environmental effects of alterna- tive spatial policies. ‘Whatif?’ approach which allows the users to develop various scenarios, provide a basis to measure the impact of new development areas on natural environment and revise the policies easily in relation to the land use esti- mation model results and future urban development pattern.

Keywords Land use, Land use policy, Sustainable development, Urbanization. 182

1. Introduction The paper has two objectives: The Estimation of future land use pat- first objective is to provide an input for tern are critically important for un- the land use planning, and the second derstanding the effects of alternative one is to develop the alternative local planning decision and assessing their land development strategies in the fu- impact on the environment in urban ture. The organization of the paper is planning process. Generally, the tech- as follows: The literature review related niques of future land use estimation to urban spatial modelling have been cover spatial and temporal processes. addressed in the second part. This These are the techniques which in- part is followed by the background clude uncertainty factor and developed information of case study area. In the on the basis of the alternative scenar- fourth part consists of methodology ios. Nowadays, urban growth model- and modelling procedure followed by ing studies witness a change towards the discussion section. The paper is -fi building ‘what if’ scenarios and dom- nalized with concluding remarks and ination of these scenarios over many the implications of the research for the modelling efforts (Batty and Torrens, future. 2005). ‘Planning Support Systems’ (PSS) constitute an informatics roof 2. Literature review: Modelling of ur- which unites all current and future in- ban spatial development formation technologies used for plan- As the human communities have be- ning with three concepts: information, come more dependent on the natural model and visualization (Klosterman resources, it is progressively becoming 1999; Geertman and Stillwell, 2004). an obligation to protect these natural Considering restraints of the planners resources, sustain the biological di- about the use of sources and informa- versity and ensure continuation of the tion, an approach of improving alter- healthy urban mechanism. The urban native ‘scenario-based estimations’ by planning which aims to organize the using the available database has been human actions by protecting the nat- introduced, instead of giving a precise ural resources and ensuring their sus- forecast about the future (Klosterman, tainability faces an increasing uncer- 1998). tainty about the future. The interaction This paper presents scenario-based between the human communities and modelling of urban land use which natural systems is of vital importance stem from interactions of the urban for the successful functioning of a city functions of Sakarya, one of the most (Barredo et.al, 2003; White et.al, 1997), important city for agricultural produc- and the land use pattern is emerged as tion and a vulnerable region in terms a result of these interactions. Howev- of natural resources and seismicity in er, it is very difficult to understand the Turkey. Modelling the prospective ur- nature of these interactions that are ban growth of Sakarya based on differ- tried to be explained by urban systems ent development scenarios is expected modelling (Openshaw, 1995). The ur- to contribute to the assessment of the ban modelling is building and execut- future urban development of Sakarya ing the complex mathematical models and determination of the strategies in order to estimate the urban growth regarding the future. Therefore, this form in the future with the aim of sup- paper aims to estimate the future land porting the planners, politicians and use pattern of Sakarya and to discuss other decision-making mechanisms. the environmental effects of alternative Role of the models in the planning spatial policies. In this paper, ‘what- process is to understand the behaviors if?’ model was used, since it allows the of the urban systems and help perform users to formulate various scenarios the urban development in compatible and select the most appropriate one with the public policies (Batty, 1976). among these and assess its impacts. In It is essential to analyze the land addition, this model provides the users use dynamics in the urban modelling with the opportunity to revise their as- approaches in order to understand sumptions and policies easily in rela- the urban functioning. The fact that tion to the model results. densely populated cities exhibit urban

ITU A|Z • Vol 12 No 1 • March 2015 • F. Terzi 183 expansion towards forests, wetlands classical urban transportation model and agricultural lands results in rap- known as four-step model was widely id depletion of the natural resources used, but this modelling techniques and degradation of the ecosystems. were found quite static. Growing un- Therefore, information concerning ur- derstanding of the impact of the in- ban growth rate, spatial development teraction between cities and transpor- form and concept is necessary for the tation systems on urban growth have planning strategies to designate the re- led to develop Integrated Land Use serves and capacities of the future nat- Transport (LU-T) Model (Sivakumar, ural resources and understand the cur- 2007). Earlier version of LU-T mod- rent and future impacts of the land use el was tried to be explained by gravi- changes. Estimating the future envi- ty type models, first one of which was ronmental results of the urban growth developed by Lowry in 1964. Later on, requires to forecast such changes in such integrated LU-T models as ME- land use and their effects on the natu- PLAN by Echenique, (1985), MUSSA ral resources. by Martínez (1992), and UrbanSim by One of the most important challeng- Waddell (2002) were developed to bet- es to be encountered by the natural and ter explain the interactions between social sciences in the years ahead will land use, transportation, the economy, be to understand the changes under- and the environment. Even though gone by the cities due to interactions of such LU-T models were developed and the humans and natural systems. There evolved, the 4-step model continues are so many complex processes in the to represent the transport modelling interaction of the cities with the nat- component (Sivakumar, 2007). ural systems and the change periods. Sivakumar (2007) groups the LU-T During this periods, urban systems are model in three category. The first subject to temporal and spatial chang- group is Travel Demand Models. This es as a result of the choices and actions type of models divide into five sub of the individuals. These choices and categories: Aggregate models, disag- actions are guided by the multiple fac- gregate trip-based models, tour-based tors. Households, business circles, real models, activity-based models and estate developers, managers, etc. are modelling freight demand. The second among the important factors of this group is operational integrated LU-T change. Households and business cir- models in which models grouped into cles decide about the production, con- two sub categories: Static models and sumption and selection of the location dynamic models. Static models are the within the city. Real estate developers earliest LU-T models which aim to are in search of the new investments. explain the relationships between the Managers decide the infrastructure land use and transportation systems. and service investments in consider- These type of models based on gravi- ation of the policies and laws. All these ty equations or input-output formula- decisions change structures of the eco- tions. Since early models approach the systems due to such actions as use of matter only from a mathematical per- the resources, conversion of the lands spective, land use spatial strategies and into the urban land, generation of policy analysis remains weak (Sivaku- emission and wastes. The recent mod- mar, 2007). Whereas, dynamic models elling studies conducted to understand defined as “…the development of im- the dynamic natures of the urban and proved modelling methodologies such natural systems concentrate on the re- as entropy-based interaction, random lations of the humans and natural sys- utility theory, bifurcation theory and tem (Alberti and Waddell, 2000). non-linear optimization, together with Another important challenge in ur- significant computational advances has ban land use modelling studies is to un- paved the way to the development of derstand the interactions between land dynamic land use-transport model sys- use and transportation system. Earlier tems” which divided into two catego- version of travel demand models were ries: General Spatial Equilibrium mod- based on estimating spatial movements els and Agent-based Micro-simulation and flows. In mid-twentieth century, models such as UrbanSim by Waddell

Scenario-based land use estimation: The case of Sakarya 184

(2002) (Sivakumar, 2007). (Meaille and Ward, 1990; Batty and Current operational LU-T models Longley, 1994; Veldkamp and Fresco, contain several shortcomings which 1996; White and Engelen, 1997; Tor- are highlighted by Sivakumar (2007). rens, 2000; Yang and Lo, 2003; Torrens, For example, the models highly rely 2006). on a) excessive spatial aggregation, b) Quite a few of the models which static equilibrium assumptions and have been developed in the last twenty c) four-stage travel demand model- years about the city and ecology mod- ling methods. In addition, the most elling (Meaille and Ward, 1990; Batty of LU-T models is operated under the and Xie, 1994a, 1994b; Landis, 1995; lack of endogenous demographic pro- Veldkamp and Fresco, 1996; Pijanows- cesses and intricate links between the ki et al., 1997; White and Engelen, land-use and travel demand compo- 1997; Clarke and Gaydos, 1998; Li and nents (Sivakumar, 2007). Yeh, 2000, Yang and Lo, 2003; Torrens, Latest LU-T modelling techniques 2006) are cell-based and have been de- called “Next-Generation Integrated veloped as separate software and/or a LU-T Models” have been developed component of the software of geomet- in response to the shortcomings of the ric information system. These models current LU-T models and they are cur- have the quality of vector or raster data rently in evolution. According to the and generally operated with the aid of Sivakumar (2007), “…these next-gen- software and visualized by integrated eration models build on the strengths with geographic information systems. and experience of currently opera- Such dynamic urban models are qual- tional models, which include generally ified as stochastic. BASS II (Bay Area strong microeconomic formulations of Simulation System) which was de- land and housing/floor-space market veloped by Landis (1992) for the San processes and coherent frameworks Francisco Bay Area of the USA and for dealing with land use-transport used by him for modelling of the ur- interactions… These ‘next-generation’ ban growth in the related region can be LU-T models are disaggregate, activi- given as an example for the dynamic ty-based and strive for greater integra- urban modelling studies. This model tion between the various components aims to perform a realistic simulation of the land use-transport system”. of the growth in the related region and Several analytical techniques have explain how the urban growth pattern been created by using various theo- and density will change in case the ur- ries about the urban modelling. Urban ban growth policies are applied on lo- form, urban growth and economy of cal or regional level (Landis, 1992). In the cities, inter-city relations, social the subsequent studies, Landis (1995) and economic functioning are some of developed CUF (California Urban Fu- the approaches on which these theories tures) model which he applied to Sac- are based. The first examples concern- ramento and San Francisco Bay Area. ing modelling of the urban systems Advantage of this model in relation to can be seen in Von Thünen (1966), the previous one is that it takes into Christaller (1966), Lösch (1954) and account the public policies and be- Alonso (1964) (Dökmeci, 2005). In the haviors of the private entrepreneurs subsequent years, it became possible to and can evaluate the alternative land make more complex calculations fol- use decisions on local-regional scale. lowing the progresses in computer and Landis improved the CUF model even software technologies and it was en- more and developed the CUF II mod- deavored to model the cities consider- el (Landis, 1997). In CUF and CUF II ing their complex characteristics. New models, a modelling technique which modelling searches have introduced was obtained from joint use of the sta- the dynamic modelling concept. This tistical estimation methods and geo- modelling approach is based on the graphic information systems was used. complex systems theory (Batty, 2005) In the dynamic urban modelling and has lately become a primary issue studies, there are also modelling tech- in the city and ecology models in terms niques which include uncertainty fac- of the geographic information systems tor (stochastic) and developed on the

ITU A|Z • Vol 12 No 1 • March 2015 • F. Terzi 185 basis of the complex systems theory. be a fully integrated, flexible and user Recently, modelling efforts which stip- friendly system and provide the follow- ulate creation of different scenarios ing opportunities to the users: with alternative policies have started “(1) It should enable the user to se- to prevail in the urban modelling stud- lect and use the most suitable analysis ies (Batty and Torrens, 2005). Plan- and estimation tools for a specific task/ ning Support Systems (PSS) which study. (2) To access local, regional or have come out in recent periods and national database with the aid of PSS lay a basis for such modelling studies and make association with the cor- include various tools supporting geo- responding analytical or prediction graphic information technology and model (3) To run the suitable models miscellaneous aspects of the planning to determine the alternative policy process (Geertman and Stillwell, 2004). options about the present situation Conceptual structure of the plan- and future and effects of different as- ning support systems is composed of sumptions and (4) To view such graph three components which are informa- results as graphics, maps, interactive tion, model and visualization (Klos- video/sound images.” terman, 1999a). Generally PSSs are Consequently, PSS is an integrated broader definitions of the geographic framework which is based on planning information systems which support theories and composed of the data, in- many aspects of the planning process formation, method and model as well (e.g. problem defining, data collection, as the tools working on a graphical spatial and process analysis, model- user interface (Geertman and Stillwell, ling, visualization, scenario building, 2003). prediction, simulation, reporting and Based on the PSS concept, vari- decision-making) (Harris and Bat- ous methodological approaches and ty, 1993; Klosterman, 1997; Brail and applications have been developed to Klosterman, 2001; Geertman and Still- support the planning process (Brail well, 2003; Geertman and Stillwell, and Klosterman, 2001; Geertman and 2004; Geertman and Stillwell, 2009). Stillwell, 2003; Geertman and Stillwell, Considering limited opportunities of 2009). Some of the popular applica- the planners about information and tions frequently seen in the literature sources, it becomes possible, with the are as follow: California Urban Fu- aid of PSS, to develop alternative sce- tures (CUF) (Landis, 1994), DUEM, narios and deliver estimations about (Xie, 1996), SLEUTH (Clarke et al, the future by using the existing data 1997), TRANUS (De la Barra, 2001), sets (Klosterman, 1998). UrbanSim (Waddell, 2000), What if? PSSs present special support for the (Klosterman, 1999a), CommunityViz specific phases of the planning pro- (Kwartler and Bernard, 2001), INDEX cess by integrating the GIS functions (Allen, 2001), Place3S (Snyder, 2003), and the computer-aided modelling Environment Explorer (Engelen et and visualization tools (Harris and al, 2003), the SPARTACUS (Lautso, Batty, 1993; Klosterman, 1997; Brail 2003), Land Use and Impact Assess- and Klosterman, 2001; Geertman and ment (LEAM) (Deal et al, 2005). In Stillwell, 2003; Geertman and Stillwell, addition to these, there are also mi- 2004; Geertman and Stillwell, 2009). cro-simulation models which are de- Harris and Batty (1993) associate the veloped for planning, design and green PSS concept to a system integrated space care and help planning processes with a set of computer-based mod- (Geertman and Stillwell, 2009). els and define three phases of PSS; i.e. Out of the models stated above, definition of the planning phases and “What if?” software developed by problems, designation of the system Klosterman (1999b) has been used in model and method and conversion of this paper. What if? is a strategy-cen- the basic data respectively in a man- tered planning tool and puts forth pos- ner to provide input for the model. sible alternatives regarding prospec- Brail and Klosterman (2001) mention tive spatial development form of a city about four characteristics for a perfect within the framework of the adopted PSS. According to them, a PSS should assumptions and formulated scenarios.

Scenario-based land use estimation: The case of Sakarya 186

Geertman and Stillwell (2003) empha- arising from natural structure of a city size that what if? is flexible and allows in order to control the future urban de- the users to specify importance of the velopment. In this research, each com- factors affecting urban development on ponent which constitutes geographical various levels. What if? model renders structure of Sakarya has not been ana- it possible to select a scenario among lyzed in detail, since such a study was the alternative scenarios and determine conducted by Provincial Directorate of impacts of each scenario. This model Environment and Forestry of Sakarya also enables the users to easily change in 2008. Instead, this paper focuses the assumptions and policies and rean- on geographical components, which alyze their results (Klosterman, 2001). would limit future urban development What-if?’ allows the users to formulate of Sakarya, have been examined as various scenarios based on different main groups and their characteristics spatial strategies, evaluate the each re- which directly provide input for the sult and assess its impacts. In addition, land use model have been defined. this model provides the users with the Sakarya exhibits wide-ranging and opportunity to measure the effects of very rich location features with its to- the each policy and assumption on the pography, geological conditions, soil future urban land use. Modelling pro- structure, water resources and land cedure has been included in the 4rd cover. This richness in natural struc- chapter of the study. ture is critically important for bio- logical diversity and sustainability of 3. Case study: Sakarya ecosystems. 6 main groups above have is located in the been dealt with, in this study, as con- northeast of the . It is trol factors of natural structure used in surrounded with in the east, Ko- future urban growth modelling of Sa- caeli and in the west, in karya. the south and in the north. Land use information is one of the While it was a village of Kocaeli in basic data used in modelling of the 1658, Sakarya became a sub-district in spatial development of the study area 1742, a district in 1852 and a province (Figure 1). According to this, land use in 1954. Today, following the earth- types and sizes of the study area where quake of 1999, Sakarya gained the sta- 13 land use types are available have tus of Metropolitan in 2000. Sakarya been given in Table 1. has 16 districts in total out of which Adapazarı, and Söğütlü are cen- 4. Data and methodology tral districts (Governorate of Sakarya, What-if? is a kind of bottom-up Provincial Directorate of Environment modelling type and starts the mod- and Forestry, 2008). Population of the elling procedure with homogeneous city is 888.556 according to 2011 popu- land units or uniform analysis zones lation census (TSI, 2012). (UAZs).Homogeneous land units Such macro issues as population (UAZ) describe the homogeneous increase, economic development and polygons (can be defined as geometri- upper scale investment decisions are cally closed areas or grids) generated by effective in spatial development of cit- the geographical information systems ies. Whereas, spatial development is in (Klosterman, 1999; 2001). These units close interaction with the natural struc- bear information of various parame- ture. If spatial growth of the city cannot ters regarding the land they represent be kept under control and guided well, such as land use type, soil capability then, the natural structure is adversely and distance to the city center. affected. Non-sustainable development The model performs land alloca- results in weakening of the life-support tion based on the related units (UAZs) systems of a city, increasing of the vul- concerning the estimated land use de- nerability against natural disasters and mands which differentiate as per the decreasing of the resistance as well as alternative public policy preferences deterioration of the overall life qual- and puts forth the spatial reflections ity. Therefore, it gains importance to of the probable regional trends in the describe the basic limiters and guiders future (e.g. growth trends for popula-

ITU A|Z • Vol 12 No 1 • March 2015 • F. Terzi 187

nated on the basis of the information obtained from three main phases: (1) land suitability analysis), (2) growth and (3) land allocation (Klosterman, 2001). Modelling procedure of the model can be seen in Figure 2.

4.1. Land suitability stage Land suitability analysis is com- posed of four steps: (1) determination of land suitability factors which help to determine the natural threshold(s), (2) designation of the factor weights, (3) grading of the factor variables and (4) determination of the land use types allowed for land conversion (Kloster- man, 2001; Asgary, Klosterman and Razani, 2007). a. Determination of the land suit- ability factors: Factors which are suitable or not for development of a certain land use. These fac- tors define the natural thresholds which are limiters and guiders for the future development. These are seismicity, land classification, soil groups, erosion status, hydrography and conservation sites. In addition to these informations of natural Figure 1. Land cover of Sakarya (SGM, 2011). environment, the distance variable are also considered as an influential factors on accessibility, urban form, Table 1. Land cover classification in Sakarya (Provincial land value and finally urban spatial Directorate of Environment and Forestry, 2008). development. These factors are ‘the Land cover class Size (Ha) Percent distance to’: railway stations, major road, road junction and city centers Prime farmland 87552,37 17,99 (Figure 3). Specific crop land 50958,14 10,47 b. Designation of the factor weights: It Planted agricultural land 20990,31 4,31 aims to specify relative importance of the factors in order to find out Marginal agricultural land 81466,25 16,74 relative conformity of each land use Degraded forest areas 36030,23 7,40 type in different locations (Table 2). Flooded forests areas 1447,65 0,30 c. Grading of the factor attributes/ features: During this step, different Forest areas 177198,68 36,41 areas/attributes/features of a cer- Grassland/Pasture 9726,13 2,00 tain factor are graded (e.g. There Swamp 363,07 0,07 are such qualities as forest, pasture, meadow, heathland in the vegeta- Ponds 6824,47 1,40 tion factor) (Klosterman, 2001). Rock fields 170,49 0,04 d. Determination of the land use types Sand 1726,71 0,35 allowed for land conversion: The fourth step includes determination Built-up areas 12182,73 2,50 of the current land uses which may Total 486637,22 100,00% be allowed to be converted into an- other land use (Table 3), if required tion and employment). UAZs are the during the land allocation process main executive units of the model and (Klosterman, 2001; Asgary, Klos- the estimated land use model is desig- terman and Razani, 2007).

Scenario-based land use estimation: The case of Sakarya 188

Figure 2. The flowchart of the model (Derived from Klosterman, 2001).

Following all steps and after all nec- 4.2. Growth: Population and density essary inputs are provided for the suit- projections ability factors, land suitability analysis The second phase of the model in- is obtained for each land use category cludes calculation of the future (target of the model. year) land use demands. The future This phase includes determination land use demands are calculated under and weighting of the factors which the different growth assumptions made control, guide and limit the growth for different land use demands. Ac- in the spatial development modelling. cordingly, future land use demands are Weighting means determination of the calculated by means of the projection factors influence in the location choice calculations about population and em- of new settlement areas and designa- ployment within the framework of cer- tion of the relative importance of the tain acceptances (Klosterman, 2001). factors in the site selection (Kloster- Population of Sakarya is continuous- man, 2001). Influence of each factor ly increasing especially in Adapazarı is graded over 100 points. High points district. This is mainly resulted from show that the related factor has a high industrialization and inclusion of Ko- level of impact on the location choice caeli and then Sakarya into develop- of new settlement areas. During this ment hinterland of Istanbul. Pursuant step, different attributes of a certain to the examination of the population factor are graded (Klosterman, 2001). trends on district basis as of 1990, all Grading is performed over 100 points districts have undergone population and the highest score shows the most increase except for Taraklı district, and preferred situation while 0 (zero) ex- only Taraklı constantly has lost popula- presses unfavorable region/location tion (Figure 4 and Table 4). (Table 2). Population values of each district were calculated for 2015, 2020 and 2025 (target year) based on the popu-

ITU A|Z • Vol 12 No 1 • March 2015 • F. Terzi 189

Figure 3. Suitability factors. lation increase trends of the districts. about 45 degree points out that popu- Polynominal equation which presents lation will increase significantly in the by far the best predictions about the future. As a matter of fact, population population trends of each district as of of Adapazarı has been calculated to 1990 was used and population values be 605214 in 2025. Besides, , of the next three terms (2015, 2020, , Karapürçek and Pamuko- 2025) were predicted with the related va are other districts with significant equation. The predicted populations population increase. Taraklı is, on the constituted the population parameters other hand, the only district that loses of the target year and taken as basis for population. the built-up area to be needed in the Size and number of the households future (Figure 4 and Table 4). and population density are other pop- In Figure 4, graph of Adapazarı ulation-related parameters used in the shows an almost linear nature in terms model (Table 5). Average household of population increase, but gradient of size was taken as 3.8 in the study area

Scenario-based land use estimation: The case of Sakarya 190

Table 2. The weight of the suitability factors (the number in parenthesis) and theirs attributes on 100 point-scales. Distance to railway Land classification (90) Points Points station (60) non-devel- 1., 2., 3., 4. Class 500 m 35 opable 5.class 90 1000 m 70 6.class 95 2000 m 90 7.class 100 4000 m 70 8.class 100 8000 m 35 Seismicity (100) 8001 m over 10 1.degree precautions Distance to major 50 required road (85) 2.degree precautions 35 250 m 100 required 3.degree precautions 15 500 m 90 required Soil groups (70) 750 m 80 non-devel- Organic Soils 1000 m 70 opable non-devel- Alluvial Coastal Soils 1500 m 60 opable Insufficiently Drained non-devel- 1501 m over 50 Alluvial Coastal opable non-devel- Distance to road Reddish Brown Soils opable junction (95) Brown Soils 0 500 m 100 Colluvial Soils 0 1000 m 90 Brown Forest Soils 35 2000 m 80 Vertisols 35 4000 m 70 Badly Drained Sierozems 65 8000 m 60 Limeless Brown Forest Soils 75 8001 m over 50 Distance to city Saline Brown Soils 100 centers (100) Erosion (100) 500 m 100 1.grade 0 1000 m 90 2.grade 25 2000 m 80 3.grade 50 4000 m 70 4.grade 75 8000 m 60 No risk 100 8001 m over 50

Conservation Hydrography (50) site (50) Archaeological site 0 Dam, Swamp, Ponds, Lake, Natural site 0 non-devel- Basin, River, Irrigated lands, opable Natural conversation area 0 Flood zone Urban conservation area 0

ITU A|Z • Vol 12 No 1 • March 2015 • F. Terzi 191

Table 3. Current land uses to be converted to future residential use. Response for Response for Land cover new residential Land cover new residential areas areas Prime farmland 0 Pasture/Grassland 1 Specific crop land 0 Swamp 0 Planted agricultural land 1 Ponds 0 Marginal agricultural land 1 Rock fields 1 Degraded forest areas 0 Sand 0 Flooded forests areas 0 Built-up areas 1 Forest areas 0

Figure 4. Population estimations.

(TSI, 2012). gross density values can be seen be- Amounts of the built-up areas and low district by district for Sakarya. The

Scenario-based land use estimation: The case of Sakarya 192

Table 4. Estimated population of the districts by years. 1990 2000 2007 2009 2011 2015 2020 2025 Adapazarı 171225 283752 377683 402310 429331 475935 543185 605214 Ferizli 5058 12379 12733 12914 13058 15483 15392 17006 Söğütlü 4839 7858 8306 8233 8306 9391 9443 10112 Akyazı 19331 23192 37729 41179 41738 46313 56038 61539 13405 17318 19802 20318 21317 22727 24479 26345 Hendek 23397 28537 45090 44418 44680 50679 59108 63361 Karapürçek 3211 4186 7467 7452 7388 8611 10307 11151 14500 24672 25607 27914 29615 32192 33943 37732 Kaynarca 3257 5064 5278 5144 5244 5858 5831 6201 10131 13793 13089 12560 11841 13264 12126 12186 10088 13200 15181 16047 16566 17791 19334 20917 Sapanca 14124 21727 23202 31614 32732 34420 39685 45553 Taraklı 5193 4146 2947 3055 2997 2355 1802 1376 Total 297759 459824 594114 861570 888556 915835 1069378 1145164 built-up areas include houses, trade, the land suitability analysis results and industry, public facility areas and all projection calculations, and change other human-made facilities. Accord- in the natural resources are analyzed. ingly, the five most densely populated The future land use pattern is obtained districts are respectively Adapazarı, following this phase. It is possible to Pamukova, Akyazı, Geyve and Kara- differentiate this pattern according to pürçek. The least populated districts the alternative scenarios (Klosterman, are Kocaali, Söğütlü and Karasu, start- 2001; Asgary, Klosterman and Razani, ing from the least populated one. 2007). This phase is composed of two steps which are developing scenario 4.3. Land allocation: Future growth and determining the spatial develop- pattern ment strategies which would affect the In the last phase of the model, the future urban growth. At the end of the future land use types are allocated “in land suitability analysis, the following the most suitable areas” on the basis of criteria are taken into account in the the demands obtained depending on given order while allocating the future

Table 5. Number of households by the districts (Provincial Directorate of Environment and Forestry, 2008). Districts Population Household Number of Built-up areas Gross density size household (ha) (p/ha) Adapazarı 429331 3.8 112982 5688.24 75.48 Ferizli 13058 3.8 3436 347.4 37.59 Söğütlü 8306 3.8 2186 457.06 18.17 Akyazı 41738 3.8 10984 807.2 51.71 Geyve 21317 3.8 5610 428.53 49.74 Hendek 44680 3.8 11758 1280.05 34.90 Karapürçek 7388 3.8 1944 151.54 48.75 Karasu 29615 3.8 7793 1074.18 27.57 Kaynarca 5244 3.8 1380 125.95 41.64 Kocaali 11841 3.8 3116 655.59 18.06 Pamukova 16566 3.8 4359 228.05 72.64 Sapanca 32732 3.8 8614 735.08 44.53 Taraklı 2997 3.8 789 77.08 38.88

ITU A|Z • Vol 12 No 1 • March 2015 • F. Terzi 193 land use types to the land units (UAZs) trol parameters are the factors which obtained for different land use types: determine future growth pattern of the land suitability, size of the land, spatial city. control parameters and randomness. After all phases are introduced to Spatial control parameters are among the system, the model is run and sim- the remarkable ones of these criteria. ulation results of spatial development Spatial control parameters describe are obtained for each scenario in the how future growth of a settlement will specified target years (e.g. once in ev- be guided and what kind of limitations ery 5 years, 15-year estimation in to- there will be. Master plan of a city can tal). Hence, urban growth form and be given as a proper example. Master its impact on the natural resources are plans define spatial location choice of found out. Consequently, each scenario future land use demands. Another spa- shows how the urban growth will take tial control parameter can be transport place in the following years and what axis. According to this, linear growth results it will bring, on condition that of the city can be induced along the the assumptions are accurate and the transport axis and it can be defined as a adopted strategies are implemented. spatial development control parameter. Another spatial development control 4.3.1. Scenario-1: Controlled devel- parameter can be infrastructure invest- opment scenario ments. It can be foreseen that future According to this scenario, continu- urban growth will continue primarily ation of the distinctive characteristics in the areas with completed infrastruc- of development (house type, densi- ture. To sum up, spatial development ty, mono-centric development trend, contl parameters are the factors which household size) in the future was an- determine future growth pattern of a alyzed and simulation of the future city. urban growth form was put forward. On the basis of the demands ob- Leading decisions of the current con- tained from the land suitability analy- struction plan were not taken into ac- sis results and projection calculations, count in this scenario. It was consid- the future land use types are allocated ered that location choice could also be “in the most suitable areas”, and change made outside the development areas in the natural resources are analyzed. projected by the current construction In the definition of “the most suitable plan for the size of the future built-up areas” emphasized above, the rules are area. It was accepted that built-up area determined by the user again. The fol- demand would be met from planted lowing criteria should be taken into ac- lands, marginal agricultural lands and count for the spatial allocation of the grasslands, and all other natural ar- future land use: land suitability, size eas would be maintained and exclud- of the land, spatial control parameters ed from settlement activities. In other and randomization. Spatial control words, ecologically sensitive regions parameters are among the remarkable would be protected completely and ones of these criteria. Spatial control new settlement demands would be di- parameters describe which fields are rected to the remaining areas. This sce- prioritized for the future growth of a nario assumes that new development settlement. As an example, consider- areas would be concentrated near and ing that the city will develop along the around the current district centers. transport axis, linear growth can be de- This scenario, in a sense, claims to con- fined as a spatial development control trol the future development which is parameter. Thus, the urban growth can why it is named as controlled develop- be thought to primarily take place on ment scenario. Acceptances and spatial the lands in parallel with the transport development strategies of this scenario axis. Another spatial development con- have been explained below. trol parameter can be infrastructure. • Built-up areas were considered to It can be foreseen that future urban develop based on the current den- growth will continue primarily in the sity degree of each district. areas with completed infrastructure. • It was considered that the future To sum up, spatial development con- built-up area demand would be met

Scenario-based land use estimation: The case of Sakarya 194

from planted lands, marginal agri- ditions would prevail, to a certain cultural lands and grasslands and extent, in the future urban forma- that forest lands, fertile agricultural tion. lands and ecologically sensitive re- • Built-up areas were considered to gions (vineyards, orchards, reeds, be 20% less dense than the current stream beds, basins, etc.) would not situation in each district. be opened to settlement. • It was considered that the future • While determining the future ur- built-up area demand would be met ban morphology, effect of the city from planted lands, marginal agri- center is decisive. Since the city has cultural lands and grasslands and a flat topography, development was that forest lands, fertile agricultural accepted to take place around dis- lands and ecologically sensitive re- trict centers in each district. gions (vineyards, orchards, reeds, • Order or precedence is as follows stream beds, basins, etc.) would not for the location choice of the future be opened to settlement. built-up areas: 1- Estimated devel- • It was accepted that main axis would opment pattern (around the cen- be decisive in determination of the ters), 2- existence of great lands, 3- future urban morphology and that land suitability and 4- randomness. urban growth would take place Within the framework of these as- physically along the main transport sumptions, results of the modelling axis in each district. and simulation works conducted for 3 • Order or precedence is as follows terms until 2025 have been stated be- for the location choice of the future low. built-up areas: 1- Estimated devel- opment pattern (along the road net- 4.3.2. Scenario-2: Sprawling develop- works), 2- existence of great lands, ment scenario 3- land suitability and 4- random- The main difference between this ness scenario and the previous one is that Within the framework of these as- this scenario analyzes the situation sumptions, results of the modelling where the future urban development and simulation works conducted for is 20% less dense than the current sta- 3 terms until 2025 as well as their im- tus and new settlement demands are pacts on the natural structure have met along the transport axis and it been stated below. puts forth a simulation for the future urban growth form. In some ways, 5. Simulation results and discussion this development form aims to reveal According to results of both scenar- low-density sprawl formation along ios, increase in the amount of built-up the transport axis. Leading decisions of areas and decrease in the natural areas the current construction plan was not (planted land, marginal agriculture taken into account in this scenario as and grassland) have been given in Ta- well. It was accepted that built-up area ble 6 and Table 7 respectively for the demand would be met from planted controlled scenario and sprawling de- lands, marginal agricultural lands and velopment scenario. According to both grasslands, and all other natural areas development scenario, the first three would be maintained and excluded districts with the highest change in the from settlement activities. The fact that built-up areas are respectively Kara- city centers will not have any influence pürçek, Akyazı and Adapazarı (Table on the new development areas means 6, 7). Two scenarios gave different re- that there is not any control parameter sults in terms of spatial development in location choice of the new settle- pattern because of the fact that density ment areas. In this scenario, transport value was taken 20% lower in the 2nd network stands out as the most deci- scenario and main transport axis were sive factor in the future urban forma- decisive in the urban growth form. tion. Acceptances and spatial develop- From the perspective of the impacts ment strategies of this scenario have undergone by the natural areas, the been explained below. highest loss (the most affected groups • It was considered that market con- of 20% in overall Sakarya) under

ITU A|Z • Vol 12 No 1 • March 2015 • F. Terzi 195

Table 6. Changes in built up areas and natural areas in controlled development scenario (% change). Adapazarı 2015 2020 2025 Akyazı 2015 2020 2025 Planted agricultural land -1.40 -3.02 -3.97 Planted agricultural land -0.16 -0.87 -1.35 Marginal agricultural land -2.75 -6.61 -10.36 Marginal agricultural land -3.78 -9.04 -11.65 Pasture/Grassland -10.66 -27.45 -41.45 Pasture/Grassland 0.00 -0.80 -1.76 Built-up areas 10.86 26.52 40.97 Built-up areas 10.99 34.38 47.54 Ferizli 2015 2020 2025 Geyve 2015 2020 2025 Planted agricultural land -0.94 -0.94 -1.30 Planted agricultural land 0.00 0.00 0.00 Marginal agricultural land -1.58 -1.58 -2.51 Marginal agricultural land -0.36 -0.60 -0.85 Pasture/Grassland 0.00 0.00 -1.94 Pasture/Grassland 0.00 0.00 0.00 Built-up areas 18.70 18.70 30.37 Built-up areas 6.76 14.91 23.68 Hendek 2015 2020 2025 Karapürçek 2015 2020 2025 Planted agricultural land 0.00 0.00 0.00 Planted agricultural land -5.13 -13.26 -17.73 Marginal agricultural land -0.69 -1.29 -1.61 Marginal agricultural land -1.64 -2.04 -2.04 Pasture/Grassland 0.00 0.00 0.00 Pasture/Grassland 0.00 0.00 0.00 Built-up areas 3.61 8.51 11.05 Built-up areas 16.99 39.53 51.41 Karasu 2015 2020 2025 Kaynarca 2015 2020 2025 Planted agricultural land -1.48 -3.71 -8.62 Planted agricultural land 0.00 0.00 -0.02 Marginal agricultural land -0.55 -0.88 -1.37 Marginal agricultural land -0.12 -0.12 -0.17 Pasture/Grassland -2.36 -2.36 -2.62 Pasture/Grassland 0.00 0.00 0.00 Built-up areas 8.71 14.67 27.43 Built-up areas 11.95 11.95 18.29 Kocaali 2015 2020 2025 Pamukova 2015 2020 2025 Planted agricultural land -2.82 -2.82 -2.82 Planted agricultural land 0.00 0.00 0.00 Marginal agricultural land -0.95 -0.95 -0.95 Marginal agricultural land -0.02 -0.02 -0.02 Pasture/Grassland 0.00 0.00 0.00 Pasture/Grassland -0.28 -0.68 -1.00 Built-up areas 12.02 12.02 12.02 Built-up areas 1.95 4.72 6.97 Sapanca 2015 2020 2025 Söğütlü 2015 2020 2025 Planted agricultural land -4.10 -12.86 -23.74 Planted agricultural land 0.00 0.00 0.00 Marginal agricultural land -0.32 -17.46 -31.93 Marginal agricultural land -1.50 -1.55 -2.53 Pasture/Grassland 0.00 -1.26 -1.26 Pasture/Grassland -14.56 -15.05 -16.08 Built-up areas 5.16 21.32 39.17 Built-up areas 13.12 13.70 21.89

the controlled development scenar- In the controlled development io is grasslands (decrease of 41%) in scenario, three important corridors Adapazarı, planted lands (decrease of emerged regarding future development 24%) and grasslands (decrease of 32%) pattern of Sakarya. The first one is in Sapanca, planted lands (decrease Söğütlü and Ferizli axis towards north of 18%) in Karapürçek and grasslands with Adapazarı in the center. The sec- (decrease of 16%) in Söğütlü. These ond axis is Karapürçek – Akyazı with rates were calculated as follow for Adapazarı in the center again. Even the sprawling development scenario: though development pattern of this Planted agricultural land (decrease of axis gives a scattered pattern for 2025, 29%) and marginal agriculture (de- it has the potential of being a strong fo- crease of 16%) in Adapazarı, planted cus. The third axis is Adapazarı-Sapan- lands (decrease of 32%) and marginal ca (Figure 5). agricultural land (decrease of 28%) in Natural limiting factors play a crit- Sapanca and planted lands (decrease of ical role in morphological formation 28) in Karapürçek, planted lands (de- of the built-up area within the study crease of 9) in Akyazı and finally plant- area. What is intriguing in the future ed lands (decrease of 7) in Kocaali. modelling of the spatial development Scenario-based land use estimation: The case of Sakarya 196

Table 7. Changes in built up areas and natural areas in sprawling development scenario (% change). Adapazarı 2015 2020 2025 Akyazı 2015 2020 2025 Planted agricultural land -5.08 -16.77 -29.87 Planted agricultural land -2.33 -6.80 -8.84 Marginal agricultural land -4.44 -10.57 -15.97 Marginal agricultural land -1.72 -4.02 -5.97 Pasture/Grassland -0.61 -0.95 -1.89 Pasture/Grassland 0.00 0.00 0.00 Built-up areas 13.58 33.18 51.25 Built-up areas 13.82 42.92 59.34 Ferizli 0.20 0.45 0.70 Geyve 2015 2020 2025 Planted agricultural land -1.27 -1.27 -1.79 Planted agricultural land -0.26 -0.50 -0.77 Marginal agricultural land -1.50 -1.50 -2.81 Marginal agricultural land 0.00 0.00 0.00 Pasture/Grassland -1.72 -1.72 -2.79 Pasture/Grassland 8.29 18.75 29.55 Built-up areas 23.24 23.24 37.82 Built-up areas 0.20 0.45 0.70 Hendek 0.20 0.45 0.70 Karapürçek 2015 2020 2025 Planted agricultural land 0.00 0.00 0.00 Planted agricultural land -5.73 -5.73 -5.73 Marginal agricultural land -0.80 -1.56 -1.95 Marginal agricultural land -9.53 -21.88 -27.74 Pasture/Grassland 0.00 0.00 0.00 Pasture/Grassland 0.00 0.00 0.00 Built-up areas 4.44 10.66 13.79 Built-up areas 20.79 49.84 63.65 Karasu 0.20 0.45 0.70 Kaynarca 2015 2020 2025 Planted agricultural land -2.17 -3.58 -7.06 Planted agricultural land -0.03 -0.03 -0.03 Marginal agricultural land -1.88 -3.18 -5.72 Marginal agricultural land -0.14 -0.14 -0.21 Pasture/Grassland 0.00 0.00 0.00 Pasture/Grassland 0.00 0.00 0.00 Built-up areas 10.97 18.34 34.32 Built-up areas 15.00 15.00 22.94 Kocaali 0.20 0.45 0.70 Pamukova 2015 2020 2025 Planted agricultural land -2.93 -2.93 -2.93 Planted agricultural land 0.00 0.00 0.00 Marginal agricultural land -7.29 -7.29 -7.29 Marginal agricultural land -0.15 -0.30 -0.46 Pasture/Grassland 0.00 0.00 0.00 Pasture/Grassland 0.00 0.00 0.00 Built-up areas 15.04 15.04 15.04 Built-up areas 2.63 5.70 8.77 Sapanca 0.20 0.45 0.70 Söğütlü 2015 2020 2025 Planted agricultural land -3.43 -16.81 -32.43 Planted agricultural land 0.00 0.00 -0.47 Marginal agricultural land -7.32 -18.48 -27.58 Marginal agricultural land -2.10 -2.21 -3.38 Pasture/Grassland -1.02 -1.40 -1.40 Pasture/Grassland -0.46 -0.46 -0.46 Built-up areas 6.46 26.63 48.99 Built-up areas 16.40 17.26 27.15 form of Sakarya is that the new built- tegrity and it can consequently create up areas can be added to the current a controlled and compact development structure in some districts while they form. The most effective control -pa are separated from one another with rameter in the development form of natural thresholds due to natural lim- this scenario is proximity to the center, iters in some others. For instance, i.e., meeting the new settlement de- in western Adapazarı, Söğütlü and mands in the great lands closest to the Hendek, the new built-up areas can district centers. be added to the existing ones while In the sprawling development sce- northern Adapazarı, Akyazı, Sapanca, nario, transport axis became the main Karapürçek, Geyve and Karasu pres- decisive for the future formation of ent a separated structure as the future Sakarya. Additionally, density value of urban form. However, this negative each district was taken 20% less than disintegration can be turned into an the current density values. As a matter advantage with an integrated planning of fact, both these acceptances of the approach if they are projected as urban scenario are two of the factors effec- areas separated from each other with tive in low-density and automobile-de- natural thresholds and having self-in- pendent settlement pattern which are ITU A|Z • Vol 12 No 1 • March 2015 • F. Terzi 197

Figure 5. Simulation results by the years for both scenarios (controlled development on the right, sprawling development on the left). defined as urban sprawl in cities of rent settlement unlike the former sce- the western countries, especially in nario and in contrary, they developed the USA. Regarding future settlement separately from the existing settlement. form of Sakarya, the simulation re- One other intriguing finding is that an- sults imply a settlement not aggregat- other city shows up in Akyazı, as an ed around the current city center, but alternative to the current settlement. mostly linear and separated from the Although the settlement form obtained existing settlement. One of the inter- from this scenario exhibits a pattern esting findings on district level is that irrelevant to the current settlements, new development areas in Adapazarı it presents an alternative development are located linearly along the highway pattern for the city which faces such which extends towards northwest of destructive natural threat as earth- the city, they are not added to the cur- quake. This scenario enables evalua-

Scenario-based land use estimation: The case of Sakarya 198 tion of the new settlement areas, which ban development. come forward in the settlement pattern Both scenarios accept that the future obtained in 2025, as new satellite cities. built-up area demand would be met What is important here is to plan the from planted lands, marginal agricul- new settlement demands expected to tural lands and grasslands and that for- come up in 2025 based on the satellite est lands, fertile agricultural lands and city concept within the plan integrity ecologically sensitive regions (vine- and coordination. yards, orchards, reeds, stream beds, 6. Conclusion basins, etc.) would not be converted to Nowadays, it is becoming a common residential areas. practice to develop urban development According to both scenarios, the scenarios with ‘what if?’ type model three districts with the highest change and perform simulation studies based of built-up area are respectively Kara- on these scenarios. ‘Planning Support pürçek, Akyazı and Adapazarı. Two Systems’ (PSS) constitute an informat- scenarios gave completely different ics roof which unites all current and urban spatial pattern because of the future information technologies used different spatial strategies. In addition, for planning. Considering restraints of natural limiters play a critical role in the planners about the use of sources morphological formation of the built- and information, PSS has been built to up area within the study area. improve the alternative scenario-based As can be seen, different develop- estimations by using the available data- ment patterns with different assump- base instead of giving a precise forecast tions and different scenarios can have about the future. the potential to be a suitable develop- This paper presents scenario-based ment pattern for the city with appro- modelling of the complex systems priate planning activities and control which stem from interactions of the mechanisms even though they have urban functions of Sakarya, which different dynamics, planning activi- is a vulnerable area in terms of nat- ties and application tools. This, as a ural resources and seismicity, both result, ensures that such modelling among themselves and with one an- approaches present the possibilities to other. Modelling the prospective urban see alternative futures for the cities and growth of Sakarya based on different introduce the possible challenges and development scenarios is expected opportunities. to contribute to the assessment of the What if? allows to work with the future urban development of Sakarya GIS data and it has been easy to apply and determination of the strategies re- to the any size of cities ranging from garding the future. neighborhood to the large regions. What if? Model used in this study What-if modelling procedure can be renders it possible to select a scenario developed by adding a transportation among the alternative scenarios and modelling component to estimate fu- determine impacts of each scenario. ture travel demand and pattern which This model also enables the users to is very much connected to future land easily change the assumptions and pol- use pattern. icies and reanalyze their results. Two different scenarios were devel- Acknowledgment oped in this study. The first scenario This paper is produced from the is more controlled, accepts the cur- project supported by Istanbul Techni- rent development dynamics and tries cal University-Scientific Research Proj- to predict the new settlement areas by ects Supported Program- Project No: using current density valuesThe sec- 36131: Mekânsal Büyümenin Planlama ond scenario is sprawling development Destek Sistemleri Yardımıyla Senaryo scenario. As can be inferred from its Tabanlı Modellenmesi: Sakarya Örneği name, this scenario projects develop- ment with lower density (density val- References ues 20% lower than the current situa- Alberti, M. (1999). Urban patterns tion were adopted) and takes the main and environmental performance: what axis as decisive and leading in the ur- do we know? Journal of Planning Edu-

ITU A|Z • Vol 12 No 1 • March 2015 • F. Terzi 199 cation and Research, 19(2), 151-163. Bay area. Environment and Planning B: Alberti, M., & Waddell, P. (2000). Planning and Design, 24, 247–261. An integrated urban development and Clarke, K. C., & Gaydos, L. J. (1998). ecological simulation model. Integrat- Loose-coupling a cellular automa- ed Assessment, 1(3), 215-227. ton model and GIS: long-term urban Allen, E., 2001. INDEX: software for growth prediction for San Francisco community indicators. In: R. K. Brail, and Washington/Baltimore. Interna- R. E. Klosterman ed. Planning Support tional Journal of Geographical Informa- Systems: Integrating Geographic In- tion Science, 12(7), 699-714. formation Systems, Models and Visu- De La Barra, T. (2001). Integrated alization Tools. ESRI Press, Redlands, land use and transport modeling: the California, 229-262. TRANUS experience. In: R. K. Brail, Alonso, W. (1970). Location and R. E. Klosterman ed. Planning Support Land Use, Harward University Press, Systems: Integrating Geographic In- Cambridge, Massachusetts, USA. formation Systems, Models and Visu- Asgary, A., Klosterman, R., & alization Tools. ESRI Press, Redlands, Razani, A. (2007). Sustainable Urban California, 129- 156. Growth Management Using What if? Deal, B., Pallathucheril, V., Sun, International Journal of Environmental Z., Terstriep, J., & Hartel, W. (2005). Research, 1(3), 218-230. LEAM technical document: overview Barredo, J. I., Kasanko, M., McCor- of the LEAM approach. Department of mick, N., & Lavalle, C. (2003). Model- Urban and Regional Planning, Univer- ling dynamic spatial processes: simula- sity of Illinois at Urbana-Champaign. tion of urban future scenarios through Dökmeci, V. (2005). Planlamada cellular automata. Landscape and Ur- sayısal yöntemler: İTÜ Yayınevi. ban Planning, 64(3), 145-160. Echenique, M. (1985). The Use of Batty, M. (1976). Urban modeling: Integrated Land Use and Transport algorithms, calibrations, predictions. Models: The Cases of Sao Paulo, Brazil Cambridge: Cambridge University and Bilbao, Spain, In: M. Florian (Ed.) Press. The Practice of Transportation Plan- Batty, M. (2005). Cities and com- ning, pp. 263-286. plexity: Understanding cities with cel- Engelen, G., White, R., & Nijs, T. lular automata, agent-based models, (2003). Environment Explorer: spatial and fractals. Cambridge, MA: The MIT support system for the integrated as- Press. sessment of socio-economic and envi- Batty, M., & Longley, P. A. (1994). ronmental policies in the Netherlands. Fractal cities: a geometry of form and Integrated Assessment, 4(2), 97-105. function: Academic Press, London Geertman, S. and Stillwell, J. (edi- Batty, M., & Torrens, P. M. (2005). tors). (2003). Planning Support Sys- Modelling and prediction in a complex tems in Practice, Springer-Verlag: Ber- world. Futures, 37(7), 745-766. lin, Heidelberg, Newyork. Batty, M., & Xie, Y. (1994a). From Geertman, S., & Stillwell, J. (2004). cells to cities. Environment and Plan- Planning support systems: an invento- ning B: Planning and Design, 21, 31–48. ry of current practice. Computers, En- Batty, M., & Xie, Y. (1994b). Mod- vironment and Urban Systems, 28(4), elling inside GIS: Part 2. Selecting 291-310. and calibrating urban models using Geertman, S., & Stillwell, J. C. H. ARC-INFO. International Journal of (2009). Planning support systems best Geographical Information Systems, practice and new methods (Vol. 95): 8(5), 451-470. Springer. Brail, R. K., & Klosterman, R. E. Harris, B., & Batty, M. (1993). Loca- (2001). Planning support systems: in- tional models, geographic information tegrating geographic information sys- and planning support systems. Journal tems, models, and visualization tools: of Planning Education and Research, Esri Press. 12(3), 184-198. Clarke, K. (1997). A self-modifying Klosterman, R. E. (1997). Planning cellular automaton model of histori- support systems: a new perspective cal urbanization in the San Francisco on computer-aided planning. Journal

Scenario-based land use estimation: The case of Sakarya 200 of Planning Education and Research, by the integration of constrained cel- 17(1), 45- 54. lular automata and GIS. International Klosterman, R. E. (2001).The What Journal of Geographical Information if? Planning support system. . In: R. K. Science, 14(2), 131-152. Brail, R. E. Klosterman ed. Planning Martínez, F.J. (1992). Toward the Support Systems: Integrating Geo- 5-Stage Land Use-Transport Model, In: graphic Information Systems, Models Selected Papers of the 6th World Con- and Visualization Tools. ESRI Press, ference on Transportation Research; Redlands, California, 263-284. Land Use Development and Globaliza- Klosterman, R. (1998). Computer tion, Lyon, France, July, pp. 79-90. applications in planning. Environment Méaille, R., & Wald, L. (1990). Using and Planning B: Planning and Design, geographical information system and 25, 32-36. satellite imagery within a numerical Klosterman, R. (1999a). New per- simulation of regional urban growth. spectives on planning support systems. International Journal of Geographical Environment and Planning B: Planning Information System, 4(4), 445-456. and Design, 26(3), 317-320. Openshaw, S. (1995). Human sys- Klosterman, R. E. (1997). Planning tems modelling as a new grand chal- support systems: a new perspective lenge area in science. Environment and on computer-aided planning. Journal Planning A, 27, 159-159. of Planning Education and Research, Pijanowski, B. C., Long, D. T., Gage, 17(1), 45-54. S. H., & Cooper, W. E. (1997). A land Klosterman, R. E. (1999b). The what transformation model: conceptual ele- if? Collaborative planning support ments, spatial object class hierarchies, system, Environment and Planning B: GIS command syntax and an applica- Planning and Design, 26(3), 393-408. tion for Michigan’s Saginaw Bay Wa- Kwartler, M., & Bernard, R. N. tershed. Online: [http://www.ncgia. (2001). CommunityViz: an integrat- ucsb.edu/conf/landuse97,] [Accessed ed planning support system. In: R. K. 20 Agust 2012]. Brail, R. E. Klosterman ed. Planning Governorate of Sakarya, Provincial Support Systems: Integrating Geo- Directorate of Environment and For- graphic Information Systems, Models estry, 2008. Sakarya province environ- and Visualization Tools. ESRI Press, mental report, Sakarya. Redlands, California, 285- 308. Sakarya Büyükşehir Belediyesi (Sa- Landis, J., & Zhang, M. (1997). karya Grater Municipality) (S.G.M.), Modeling urban land use change: the 2011. 2008 yılı Çevre Düzeni Planı, Sa- next generation of the California Ur- karya. ban Futures Model. Paper presented Sivakumar, A. (2007). Modelling at the Paper submitted to the Land transport: a synthesis of transport Use Modeling Workshop, USGS EROS modelling methodologies. Imperial Data Center. College of London. Landis, J. D. (1992). BASS II: a new Snyder, K. (2003). Tools for commu- generation of metropolitan simulation nity design and decision-making Plan- models: University of California at ning support systems in practice (pp. Berkeley, Institute of Urban and Re- 99-120): Springer. gional Development. Torrens, P. M. (2000). How cellular Landis, J. D. (1995). Imagining land models of urban systems work. WP-28. use futures: applying the California Centre for Advanced Spatial Analysis urban futures model. Journal of the (CASA), University College, London. American Planning Association, 61(4), Torrens, P. M. (2006). Simulating 438-457. sprawl. Annals of the Association of Lautso, K. (2003). The SPARTACUS American Geographers, 96(2), 248-275. system for defining and analyzing sus- Turkish Statistical Institute-TSI. tainable urban land use and transport 2012. Population Census, . policies Planning support systems in Veldkamp, A., & Fresco, L. (1996). practice (pp. 453-463): Springer. CLUE: a conceptual model to study the Li, X., & Yeh, A. G.-O. (2000). Mod- conversion of land use and its effects. elling sustainable urban development Ecological modelling, 85(2), 253-270.

ITU A|Z • Vol 12 No 1 • March 2015 • F. Terzi 201

Waddell, P. (2000). A behavioral and Design, 24, 235-246. simulation model for metropolitan White, R., Engelen, G., & Uljee, I. policy analysis and planning: residen- (1997). The use of constrained cellular tial location and housing market com- automata forhigh resolution modeling ponents of Urban Sim. Environment of urban land-use dynamics. Environ- and Planning B: Planning and Design, ment and Planning B: Planning and De- 27(2), 247-264. sign, 24(3), 323-343. Waddell, P. (2002). UrbanSim: Mod- Xie, Y. (1996). A generalized model eling urban development for land for cellular urban dynamics. Geograph- use, transportation, and environmen- ical Analysis, 28(4), 350-373. tal planning. Journal of the American Yang, X., & Lo, C. (2003). Modelling Planning Association, 68(3), 297-314. urban growth and landscape changes White, R., & Engelen, G. (1997). in the Atlanta metropolitan area. Inter- Cellular automata as the basis of in- national Journal of Geographical Infor- tegrated dynamic regional modelling. mation Science, 17(5), 463-488. Environment and Planning B: Planning

Senaryo tabanlı arazi kullanım tah- lar, ‘bilgi’, ‘model’ ve ‘görselleştirmedir’ mini: Sakarya örneği (Klosterman 1999). Kaynakların ve bil- Arazi kullanım modellemesi çalış- ginin kullanımında, plancıların sınırla- maları, karmaşık kentsel sistemlerin maları göz önüne alındığında, mevcut dinamik yapılarının anlaşılması ve veri tabanlarının kullanılmasıyla, PSS, çevreye olan etkilerinin değerlendi- geleceği tam olarak tahmin etmek ye- rilmesi açısından büyük önem taşı- rine, alternatif senaryo tabanlı tahmin- maktadır. Günümüzde arazi kullanım leri geliştirmek üzere kurgulanmıştır modellemesi yaklaşımları, mekânsal (Klosterman 1998). ve zamansal süreçleri kapsayan ve di- Bu çalışmada, gerek doğal kaynaklar namik modelleme tekniklerini ortaya açısından gerekse depremsellik açısın- koymaktadır. Bu teknikler, belirsizlik dan hassas bir bölge olan Sakarya’nın unsurunu içeren ve karmaşık sistemler hem kentsel fonksiyonlarının ken- teorisine dayanılarak geliştirilen tek- di içinde, hem de bu fonksiyonların niklerdir. Bu nedenle arazi kullanım birbirleriyle etkileşimleri sonucunda ve bunun gibi kentsel modelleme ça- ortaya çıkan karmaşık sistemlerin se- lışmalarının günümüzde karşı karşıya naryolara bağlı olarak modellemesine kaldığı en önemli problemlerden biri- çalışılmıştır. Sakarya’nın arazi kulla- si ‘tahmin edebilirliktir’. Batty (2010), nımının modellenmesinde What-if? belirli kentsel büyüme biçimlerinin, modelleme yaklaşımı benimsenmiştir. uygun ölçeklerdeki modellerle kestiri- Bu yaklaşımın en önemli özelliği, farklı lebileceğini ve bu türdeki ‘bilinmezli- mekansal stratejilere bağlı olarak se- ğin’ kısmen ortadan kaldırabileceğini naryolar kurgulama ve bu senaryolar söylemektedir. çerçevesinde Sakarya’nın gelecekteki Günümüzde, arazi kullanım mo- arazi kullanımı ve mekansal büyüme delleme çalışmaları, ‘what if’ senaryo- biçiminin tahmin edilerek doğal çev- larının kurgulanmasına ve bu senar- reye olan etkilerinin analiz edilmesine yoların artık çoğu model oluşturma olanak sağlamasıdır. Bu model, ayrıca, çabalarına hakim olması yönündeki kullanıcıların model sonuçlarına göre bir değişime tanıklık etmektedir (Bat- model öncesi varsayımların ve mekan- ty ve Torrens, 2005). ‘Planlama Destek sal büyüme stratejilerinin kolaylıkla Sistemleri’ (PSS), planlama için kulla- revize edilerek yeniden modele sokul- nılan bugünkü ve gelecekteki bütün masına fırsat vermektedir. bilişim teknolojilerini bütünleştiren Model iki amaca hizmet etmektedir. bir bilişim çatısıdır. PSS, daha öncele- Birincisi, arazi kullanım planlamasına ri planlama sürecinin farklı yönlerini girdi vermek, ikincisi ise elde edilen desteklemek için geliştirilmiş geo-bilgi projeksiyonlar çerçevesinde, gelecekte teknolojisi ile ilgili araçları içermekte- alternatif yerel arazi geliştirme strate- dir (Geertman ve Stillwell, 2004). PSS, jileri oluşturmaktır. What-if? modeli üç kavram ile ifade edilmektedir. Bun- literatürde çokca bahsedilen senaryo Scenario-based land use estimation: The case of Sakarya 202 tabanlı modelleme türlerinden olan rından şu sonuçlar çıkarılmıştır: California Urban Futures (CUF)(Lan- Kontrollü gelişme senaryosuna göre, dis,1994; Landis, 1995) ve SOAP mo- yapılaşmış alan miktarında gözlenen deli ile benzerlik göstermektedir (San artışta ilk üç ilçe sırasıyla, Adapazarı, Diego Assoc. of Governments, 1994). Hendek ve Karasu ilçeleri olmuştur. İlk defa Klosterman (1999; 2001) Yayılarak gelişme senaryosuna göre ise, tarafından geliştirilen What-if model- Karapürçek, Akyazı ve Sapanca olmuş- lemesi aşağıdan yukarıya (bottom-up) tur. Her iki senaryonun birbirinden doğru bir yaklaşım sunar ve 3 aşama- tamamen farklı sonuçlar vermesi, hem dan oluşur: yoğunluk değerinin 2. Senaryoda 1. Se- a. Arazi uygunluk analizi: Bu adım- naryodan %20 daha az kabul edilmesi, da her bir arazi kullanımı, arazi hem de şehirsel büyüme biçiminde ana örtüsü ve doğal kaynakların türleri ulaşım akslarının belirleyici olmasın- tanımlanır. Tüm bu türler kent- dan kaynaklanmaktadır. sel gelişmenin yönlendiricileri Kontrollü gelişme senaryosunda, olan faktörler olarak belirlenir ve Sakarya’nın gelecekteki gelişme biçi- bunlar ağırlıklandırılır. Buna ek minde üç önemli koridorun öne çıktı- olarak, mekânsal büyüme süre- ğı görülmektedir. Bunlardan birincisi, cinde dönüşüme izin verilen arazi Adapazarı merkez olmak üzere kuzeye kullanım türleri tanımlanır. doğru Söğütlü ve Ferizli aksıdır. İkin- b. Büyüme: İkinci aşamada sena- ci aks, Adapazarı merkez olmak üzere ryolara bağlı olarak, gelecekteki Karapürçek Akyazı aksıdır. Bu aksın (hedef yıl) arazi kullanım talepleri gelişme deseni her ne kadar 2025 yılı hesaplanmaktadır. için dağınık bir görüntü verse de güçlü c. Arazi tahsisi: Son aşama ise, arazi bir odak olma potansiyeli taşımaktadır. uygunluk analizi sonuçlarına ve Üçüncü aks ise, Adapazarı- Sapanca projeksiyon hesaplarına bağlı aksıdır. olarak, gelecekteki arazi kullanım Araştırma alanında, yapılaşmış ala- türlerinin en uygun yerlerde tahsis nının morfolojik biçimlenmesinde, edilmesi ve doğal kaynaklardaki doğal sınırlayıcı unsurların büyük rol değişimin analizi aşamasıdır. oynadığı ortaya çıkmıştır. Sakarya’nın Bu çalışmada uygulanan model so- mekânsal büyüme biçiminin geleceğe nucunda, iki farklı senaryo ile önü- dönük modellenmesinde ortaya çıkan müzdeki 15 yıl boyunca arazi kulla- dikkat çekici özellik, yeni yapılaşma nımı ve mekansal büyüme biçiminin alanları bazı ilçelerde mevcut yapıya tahmin edilmesi ve söz konusu kentsel eklenerek büyüyebilirken, bazı ilçe- büyümenin doğal kaynaklar üzerinde- lerde ise doğal sınırlayıcılardan ötürü ki etkileri ortaya konmuştur. Birinci doğal eşiklerle birbirinden ayrılmış senaryo doğal kaynakların korunma- yerleşmeler biçiminde bir gelişme biçi- sını destekleyici politikalardan olu- mi ortaya çıkmaktadır. Bu ayrışma bir şan, daha kontrollü büyüme sunan, olumsuzluk gibi görünse de, bu alanlar mevcut gelişme dinamiklerini aynen bütüncül bir planlama anlayışı ile ele kabul eden ve yine mevcut yoğunluk alınıp, birbirilerinden doğal eşikler- değerlerini kullanarak gelecekteki yeni le ayrılmış ve kendi içinde bütünlüğü yerleşme alanlarını tahmin etmeye ça- olan kentsel alanlar gibi planlanırsa bu lışan senaryodur. Bu senaryoya kont- durum avantaja çevrilerek, kontrollü rollü gelişme senaryosu adı verilmiştir. ve kompakt bir gelişme yapısına çevri- İkinci senaryo ise piyasa odaklı gelişme lebilir. Bu senaryoda gelişme biçiminde temelleri üzerinde kurgulanmış yayı- en etkili kontrol parametresi merkeze larak gelişme senaryosudur. Adından yakınlık olmuştur. Yani, yeni yerleşme da anlaşılacağı üzere bu senaryo, daha taleplerinin ilçe merkezlerine müm- düşük yoğunlukla gelişmeyi öngören kün olan en kısa mesafelerdeki büyük (mevcut durumdan %20 daha düşük arazilerde karşılanması olmuştur. yoğunluk değerleri kabul edilmiştir) Yayılarak gelişme senaryosunda ve kentsel gelişmede ana aksları belir- ise, Sakarya’nın gelecekteki biçimlen- leyici ve yönlendirici olarak kabul eden mesinde temel belirleyici olarak ula- senaryodur. Her iki senaryoya bağlı şım aksları etkili olmuştur. Buna ilave olarak elde edilen simülasyon sonuçla- olarak her ilçedeki yoğunluk değeri,

ITU A|Z • Vol 12 No 1 • March 2015 • F. Terzi 203 mevcut yoğunluk değerlerinden %20 bir örüntü sergilese, deprem gibi yı- daha düşük alınmıştır. Aslında bu se- kıcı bir doğal tehlike ile karşı karşıya naryoda yapılan her iki kabul, özellikle olan kente alternatif bir gelişme deseni Amerika’da şehirsel saçaklanma (ur- sunmaktadır. Bu senaryo, 2025 yılında ban sprawl) olarak tabir edilen düşük elde edilen yerleşme deseninde ortaya yoğunluklu, otomobil bağımlısı yaygın çıkan yeni yerleşme alanlarının yeni yerleşme dokularında etkili olan fak- uydu kentler gibi değerlendirilmesine törlerden ikisidir. Simülasyon sonuç- olanak tanımaktadır. Burada önemli larından Sakarya’nın gelecekteki yer- olan 2025 yılında ortaya çıkması bek- leşme biçiminin mevcut kent merkezi lenen yeni yerleşme taleplerinin yine etrafında kümelenmeyen, çoğunlukla plan bütünlüğü ve koordinasyonu lineer ve mevcut yerleşmeden kopuk içinde uydu kentler ve çok merkezlilik bir yerleşme düzeni ortaya çıkmıştır. konsepti ile planlanması gereğidir. İlçeler bazında dikkat çeken bulgular- Görüldüğü üzere farklı varsayımlara dan birisi, Adapazarı’ndaki yeni geliş- sahip ve farklı senaryolardan elde edi- me alanlarının kentin kuzeybatısına len farklı gelişme desenleri, farklı dina- uzanan karayolu boyunca lineer ola- miklere, planlama eylemlerine, uygu- rak yer seçtiği, bir önceki senaryodan lama araçlarına sahip olsa bile, her iki farklı olarak mevcut yerleşmeye eklen- şehirsel büyüme biçimi uygun planla- mediği, tersine mevcut yerleşmeden ma eylemleri ve kontrol mekanizma- kopuk olarak geliştiği görülmektedir. ları ile kent için uygun gelişme deseni Bir diğer ilginç bulgu ise, Akyazı’da olma potansiyeli taşıyabilmektedir. Bu mevcut yerleşmeye alternatif ikinci bir da, bu tür modelleme yaklaşımlarının kentin ortaya çıkmasıdır. Her ne ka- kentler için alternatif gelecekleri görme dar bu senaryoda elde edilen yerleşme olanaklarını, olası sorunları ve fırsatları düzeni mevcut yerleşmelerden kopuk ortaya koyma becerisini sunmaktadır.

Scenario-based land use estimation: The case of Sakarya