Predicting Urban Expansion in Arak Metropolitan Area Using Two ...

World Applied Sciences Journal 18 (8): 1124-1132, 2012 ISSN 1818-4952 © IDOSI Publications, 2012 DOI: 10.5829/idosi.wasj.2012.18.08.1217

Predicting Urban Expansion in Arak Metropolitan Area Using Two Land Change Models 1

Mozhgan Ahmadi Nadoushan, 2Alireza Soffianian and 3Alireza Alebrahim

Young Researchers Club, Majlesi Branch, Islamic Azad University, Isfahan, Iran Department of Environmental science, Faculty of Natural Resources, Isfahan, 84156-83111, Iran 3 Department of Agriculture, Payame Noor University, 19395-4697, Tehran, I.R of Iran 1

2

Abstract: Accurate and up-to-date information about urban expansion and changes in urban areas is needed for urban planning. Besides detecting changes occurred in an area, predicting future changes is important for long-term planning. Arak is one of cities which has undergone fast urban expansion during recent decades due to industrialization and population growth. The main objectives of the present study are to detect changes in urban area of Arak during 1956-2006 and to predict urban changes to 2025 using 2 land change model: CAMarkov and Geomod. To detect and simulate urban changes, aerial photos and Landsat TM and IRS-P6 LISS-III images were used. Visual interpretation of aerial photos and neural network classification of satellite images were employed for generating land use maps with two main classes: built-up or non-built-up. To predict urban changes for 2025, two models, land use map of 2006 and suitability map derived from driver maps such as slope map, distance to urban areas and distance to roads maps were used. Land use map for 2006 also was predicted and the capability of model in projecting this map was validated using real map of 2006. Results showed that the model had good predictive power in Arak city. Key words: Urban changes

CA-Markov

Geomod

INTRODUCTION Urban studies are becoming important tools for planners knowing that in 2015 more than half world’s population will be living in cities [1]. Urbanization is a major trend in recent years all around the world. Rapid urbanization is often the cause of enormous pressure on rural and natural environments [2] and cause changes in other land cover or land use types of an area. Over 80% of worldwide urban population growth comes from Africa and Asia, whose cities will see their populations double in the next 30 years [3]. As a developing country, Iran is now witnessing an almost continual large-scale urbanization. The number of towns and cities has also increased significantly, such that a total of 199 towns in 1956 reached 825 in 2001 and increased to 1016 in 2006 [46]. Today, with concern to population growth and people‘s tendency to live in cities, most of changes occur in urban areas. Arak is one of cities in Iran which has undergone fast urban expansion during recent decades due to industrialization and population growth.

Predictive power

Arak

Land-cover and land-use change analyses and projection provide a tool to assess ecosystem change and its environmental implications at various temporal and spatial scales [5]. Satellite remote sensing in conjunction with GIS has been widely applied and been recognized as a powerful and effective tool in detecting land use and land cover change [7]. Remotely sensed imagery provides an efficient means of obtaining information on temporal trends and spatial distribution of urban areas needed for understanding, modeling and projecting changes [8]. There are various classification methods. In this study, Artificial Neural Networks (ANNs) classifier was used to generate land use maps from satellite images. Artificial Neural Networks classifier has been widely used for land use/land cover mapping during recent decades [9]. One of the most widely used change detection methods is post-classification comparison method which uses separate classifications of images acquired at different times to produce difference maps from which ‘‘from-to’’ change information can be generated [7,8 ,10].

Corresponding Author: M. Ahmadi Nadoushan, Young Researchers Club, Majlesi Branch, Islamic Azad University, Isfahan, Iran. Tel: +98-9131697106, Fax: +98-3116643569.

1124

World Appl. Sci. J., 18 (8): 1124-1132, 2012

Post-classification comparison method provides a complete matrix of change information [11, 12]. In addition to detection of past changes, prediction and modeling of future changes are important for understanding quality and quantity of changes with passing time. Prediction of future land-use and land-cover changes is crucial for long-term planning. Numerous methods are used to build scenarios of the future. Land use change models are tools for understanding the causes and consequences of land use dynamics. There are some models for predicting land cover and land use changes such as CA-markov, Geomod, etc. Using these models, policy makers can analyze different scenarios of land use change and evaluate their effects, so models can support land use planning and policy [13]. After modelling and prediction of changes, validation of model is needed. Pontius and Malanson (2005) compared two land change models, CA-Markov and Geomod in terms of predictive power for modeling land change in central Massachusetts, USA [14]. Cabral and Zamyatin (2006) investigated urban dynamics of Sintra-Cascais municipalities (Portugal) between 1989 and 2025 using three land change models: two CA-based models and Geomod. They used Landsat TM and ETM for 1989, 1994 and 2001. Classified maps and suitability maps were used as inputs of models to predict urban area changes [1]. Rahdary et al. (2008) used satellite images to prepare land use map of Mouteh Wildlife Refuge and to detect changes occurred in study area during 1972-2005. Land use maps were produced using hybrid image classification method and changes were detected by post classification comparison method [15].

Fan et al. (2008) explored land use and land cover changes in the Core corridor of the Pearl River Delta (China) from 1998 to 2003 using TM and ETM+ images and the post-classification method. They also used Markov chain and Cellular Automata models to predict urban expansion in 2008 and 2013 and concluded that Markov chain modelling is an effective way to monitor and predict land use change and urban expansion [16]. The main objectives of the present study are to detect changes in urban areas of Arak from 1956 to 2006 using post-classification comparison method and to predict urban expansion of the study area for 2025 based on CA-Markov and Geomod models. MATERIALS AND METHODS Study Area: The study area covered approximately 8800 hectares located between latitudes 34°02'-34 °09' N and longitudes 49°36'-49 °49' E in the center of Iran (Figure 1). It contains Arak, Capital city of Markazi Province and its surrounding area. The population of Arak increased from nearly 59000 in 1956 to 446760 in 2006 [6]. Arak has experienced rapid expansion due to population growth and industrialization during recent decades. Average annual temperature of Arak is 13.8° C and average annual rainfall is 316 mm. Data: A time series of remote sensing data including aerial photos and satellite images spanning 5 decades was used to generate land cover maps. Aerial photographs from 1956 and 1972 at a scale of 1:50000 and 1:10000 were employed for producing land cover maps. Landsat TM and IRS-P6 LISS III images (path 68, row 46)

Fig. 1: The study area 1125

World Appl. Sci. J., 18 (8): 1124-1132, 2012

Fig. 2: The flowchart of methodology for the years 1990 and 2006 were also used in this study. A Digital Elevation Model (DEM) was produced from the digital topographic maps at a scale of 1:25.000 and Slope map was derived from DEM. Finally, an IRS-1C PAN image was used to enhance the spatial resolution of IRSP6 LISS-III image. Data Preprocessing: To prepare data for generating land use maps and detecting its changes, following procedures were employed. Figure 2 depicts the methods applied in the present study. Geometric Correction: All images and aerial photographs were rectified to UTM zone 39 N with at least 25 well distributed ground control points. At first geometric correction was carried out using topographic maps with the scale of 1:25000 to geocode aerial photos. Also for geometric correction of the 2006 IRS-1C PAN image, topographic maps with the scale of 1:25000 were used and then this rectified image was employed to register the 2006 LISS-III image. Geometric correction of Landsat TM image

of 1990 was carried out by the use of IRS-P6 LISS-III image. Finally, all data were resampled to a 30 m pixel size using the nearest neighbor method. After geometric correction of aerial photos, all photos for each year were mosaicked to prepare one image for land use mapping. Image Enhancement: The goal of image enhancement is to improve the visual interpretability of an image by increasing the distinction between features [17]. In this study, two false color composites (FCC) were produced for selecting training samples. Also image fusion was done to increase spatial resolution of the LISS-III image. LISS-III image was fused with IRS-1C PAN image to generate an image with high spatial resolution. Land Use Mapping: For generating land use maps from aerial photos and satellite images, 2 main land use classes were selected, namely built-up or non-built-up. For this study, built-up areas involve residential, commercial, industrial, educational, recreational establishments and transportation systems. 1126

World Appl. Sci. J., 18 (8): 1124-1132, 2012

Aerial Photos Interpretation: Land use pattern was interpreted visually on black and white aerial photographs and simultaneously digitized with the Arcmap software. Identifying features in aerial photos was performed based on tone, texture, pattern, size and shape. Image Classification and Accuracy Assessment: All land use maps were generated through artificial neural networks classifier. In order to get precise classification results, the training samples were selected from false color composite (FCC) images and topographic maps. To map land use information, a three-layer-perceptron neural networks were employed which contained one input layer, one hidden layer and one output layer. Input layer included spectral bands and training samples and output layer had two nodes. Experiments were conducted for selecting the best number of nodes in the hidden layer to improve the classification accuracy. The number of nodes in the hidden layer was selected equivalent to the number of nodes in the input layer. The parameters of momentum and learning rate were set at 0.5 and 0.2 based on experimental results. The Overall accuracy of land use maps has been assessed by error matrices. The Ground truth data were derived from GPS, topographic maps and false color composite images and error matrix was generated for each land use map. Post-classification Change Detection: Post-classification comparison change detection algorithm which is the most common approach for change detection process was used to determine changes in urban areas in 5 decades from 1956 to 2006. Change Prediction: Simulation models of land-use andcover change (LUCC) typically examine a landscape at initial points in time t0 and t1 and then predict the change from t1 to some subsequent point in time t 2. The predicted map of t2 is usually compared to a reference map (i.e., a truth map or observed map) of t2, in order to evaluate the performance of the simulation model [18]. Cellular Automata Markov (CA_Markov) and Geomod are the two models, which have similar options to allow for specification of the predicted quantity and location of land categories. Geomod is a model which uses exactly two categories of land use or land cover and can simulate only the transition from the first category to the second category and can not simulate an additional simultaneous transition of the second category to the first [14]. Geomod has the advantage that it does not

require large amounts of data for calibration and validation compared to other complex dynamic models [19]. This model has a method to create empirically and automatically a suitability map based on several driver maps and examines a suitability map to find the pixels with the largest suitability value and then predicts conversion of new urban at the non-urban pixels that have the largest suitability for urban [14]. Land cover map of 1990 was used to calibrate Geomod model and after that the model predicts urban changes from 1990 to 2006. Projecting urban area changes for 2025 carried out using the suitability map and land use of 2006. Markov and CA-Markov models were processed for each pair of dates and transition probability and transition area matrices were produced. The transition probability matrix determines the likelihood of change in each pixel from one land use or land cover class to each other category between the two dates. The markovian model outputs an image which reports the probability that each land use or land cover class would be found at each location in the next step [20]. Markov Chain Analysis describes land use change from one period to another and uses this as the basis to project future changes. This is accomplished by developing a transition probability matrix of land use change from time one to time two, which will be the basis for projecting to a later time period. To predict land use changes in the study area, Markov chain analysis and CA-Markov model were employed. Transition probability matrices were derived from Markov chain analysis and predicted land use maps were generated through CA-Markov model. Land use maps from 1972 and 1990 were used for model calibration and then the models used to predict changes from 1990 to 2006. This operation was repeated for 1990-2006 and a land use map for 2025 was simulated. A real map for 2006 has been produced through neural network classification of satellite images, to be used as reference map in the model validation process. Preparing Suitability Maps: The suitability maps determine which pixels will change according to the largest suitability [1]. In this study, for modelling and prediction using CA-Markov, the suitability maps that predict the transitions of built-up and non-built-up areas were created by Geomod model. These suitability maps were generated using driver maps and Geomod module. The driver maps were slope, distance to urban areas and distance to roads maps. Projection of land use changes using CA-Markov and Geomod was carried out for 2025.

1127

World Appl. Sci. J., 18 (8): 1124-1132, 2012

Model Validation: Validation is the process of comparing the model’s prediction for time t2 to a reference map of time t2 [21]. To evaluate the overall performance of models for predicting urban changes, 2 methods were used, error matrices and calculation of agreement and disagreement between predicted and real map at the same time through validation process. Examine the Goodness-of-Fit of Validation and Kappa Indexes: A good fit of validation is the appropriate indicator of the model’s predictive power. It means that a high goodness-of-fit of validation shows that model has good ability or predicting changes [22]. We used VALIDATE module of Idrisi software which examines the components of agreement and disagreement between the comparison and reference maps of the same time. Real land use map derived from neural network classification was regarded as the reference map and the comparison map was the result of simulation. Validity of the predicted map was assessed using reference map. VALIDATE module also computes the Kappa index of agreement and its variants. The kappa index of agreement has several variations, each of which measures different characteristics of agreement. Kno gives the overall accuracy of a simulation run. Klocation indicates the level of agreement of location, given a specified quantity. Kquantity indicates the level of agreement of quantity, given the model's ability to specify location. These variations complement the standard Kappa. Kappa equals 1 when agreement is perfect and equals 0 when agreement is as expected by chance [23]. Error Matrices: Evaluation of the model’s predictive power also was performed by calculating the overall accuracy of predicted maps. For this purpose, Ground truth data obtained for classification accuracy assessment were used and the overall accuracy of simulated land cover maps were calculated. RESULTS AND DISCUSSION Accurate per-pixel registration of multi-temporal remote sensing data is essential for change detection because registration errors may lead to an overestimation of actual change [24]. The root mean square errors (RMSE) for all aerial photographs were between 0.2 and 0.7 pixels. RMSE for PAN, LISS III and TM images were determined to be 0.42, 0.48 and 0.58 pixels, respectively. The RMSE for urban areas was lower than other areas because finding more control points in urban areas was

easier than other areas. Image fusion was carried out and resulted in producing an image with high spatial resolution equal to 5.8 meter. The fused image enhanced the capability of identifying features and selecting training samples. Land use mapping was carried out using visual interpretation of aerial photos and neural network classification of satellite images and both methods generated land use maps with high accuracy. A standard overall accuracy for land-cover and land-use maps is set between 85 and 90 percent [19]. In this study, the overall accuracy of the land use maps for 1956, 1990 and 2006 were 95.03%, 95.53% and 90.6%, respectively. The Kappa index for the 1956, 1990 and 2006 land use maps were found to be 0.93, 0.92 and 0.81. Change Detection: At first, area and percentage area of non-built-up to built-up classes were calculated which is shown in Table 1. Non-built-up to built-up transformation from 1956 to 2006 occurred at a rate of 71.6 ha per year. Transformation of nearly 3581 ha from non-built-up to built-up areas was due to industrialization and population growth. The population of Arak city increased from nearly 59000 in 1956 to approximately 446760 in 2006. During this time period, a lot of companies and factories were constructed in Arak and this factor attracts people from other cities of Iran and also some rural areas of Markazi province to migrate to this city. Table 2 shows that approximately 31 ha of built-up areas were converted to non-built-up areas from 1956 to 2006. Observation indicated that these changes are due to classification errors. Main parts of these areas were urban areas such as highways and streets that were classified as green spaces mistakenly. Highways and streets are generally classified as urban (built-up) class, but when trees along the streets grow, they may be classified as green spaces. Yuan et al. (2005) addressed this problem in their study on land cover classification and change analysis of Twin cities metropolitan area. They mentioned transformation of about 6200 ha of urban area to forest from 1986 to 2002 [8]. Change Prediction Using Models and Model Validation: Geomod and suitability map derived from driver maps were used to predict changes in urban area for 2006 and 2025 (Figure 3). Predicted land cover map of 2025 derived from CA-Markov model is shown in figure 4. Real land use map for 2006 has been generated through neural network classifier and because of validation, it was predicted.

1128

World Appl. Sci. J., 18 (8): 1124-1132, 2012 Table 1: Area and area percentages of land use classes for 1956 and 2006 Area percentages (2006)

Area (2006)

Area percentages(1956)

Area (1956)

Land use classes

45.7

4011

5.2

461

54.3

4772.2

94.8

8322.2

Built non-built

100

8783.2

100

8783.2

Total

Table 2: Matrix of built-up and non-built-up area changes from 1956 to 2006 1956 --------------------------------------------------------------------------------------------------------total

non-built

Built

4011

3580.6

430.4

built

4772.2

4741.6

30.6

non-built

8783.2

8322.2

461

Total

Fig. 3: Predicted land cover map of 2025 derived from Geomod model

2006

Fig. 4: Predicted land cover map of 2025 derived from CA-Markov model

Table 3: Components of agreement and disagreement between real and simulated map of 2006 using Geomod model Disagreement

Percentage

Agreement

Percentage

Disagreement due to location

7

Agreement due to chance

33

Disagreement due to quantity

3

Agreement due to Quantity

2

-

-

Agreement due to location

55

Overall disagreement

10

Overall agreement

90

According to the results of both models, built-up area tend to increase in future, as Geomod predict built-up area to be 5973 ha and CA-Markov predict it to be 5162 ha in 2025. The two most important components of disagreement between real and predicted maps are quantity disagreement and location disagreement, which sum to the total disagreement [21].

Table 3 and Figure 5 show agreement and disagreement components between real and simulated land use map using Geomod model. The value of overall agreement is 90 % and it indicates that Geomod had high potential in predicting changes during 1990-2006. Disagreement due to location is 7 % and disagreement due to quantity is 3 %. It means that the model had better performance in simulating the quantity of pixels rather

1129

100

100

90

90

80 70 60 50 40 30 20

disagreement due to quantity

Percent of Landscape

Percent of Landscape

World Appl. Sci. J., 18 (8): 1124-1132, 2012

disagreement due to location agreement due to location agreement due to quantity agreement due to chance

80

disagreement due to location

60

agreement due to location

50 40

agreement due to quantity

30

agreement due to chance

20

10

10

0

0

Fig. 5: Components of agreement and disagreement between real and simulated map of 2006 using Geomod model

disagreement due to quantity

70

Fig. 6: Components of agreement and disagreement between real and simulated map of 2006 using markov model

Table 4: Kappa indexes of agreement for 2006 map predicted using Geomod Kstandard

0.85

Klocation Khisto

0.89 0.95

Table 5: Components of agreement and disagreement between real and simulated map of 2006 using Markov model Disagreement Disagreement due to location Disagreement due to quantity Overall disagreement

Percentage

Agreement

7 4 11

Percentage

Agreement due to chance Agreement due to Quantity Agreement due to location Overall agreement

33 2 54 89

Table 6: Kappa indexes of agreement for 2006 map predicted using CA-Markov Kstandard

0.84

Klocation Khisto

0.89 0.95

7000 6000

Urban area(ha)

than the location of pixels. The results of kappa indexes confirm the fact mentioned above as Klocation which reveals the level of agreement of location is 0.89 and Kquantity which shows the level of agreement of quantity is determined to be 0.95 (Table 4). Table 5 and Figure 6 show agreement and disagreement components between real and simulated land use map using Markov model. The value of overall agreement is 89 %. Disageement due to location is 7 % and disagreement due to quantity is 4 %. It means that the model had better performance in simulating the quantity of pixels rather than the location of pixels. The results of kappa indexes reveals the level of agreement of location is 0.89 and Kquantity which shows the level of agreement of quantity is determined to be 0.95 (Table 6). In the present study, Markov and Geomod models has similar overall agreement between real and simulated maps. The value of overall agreement between real and predicted map derived from Geomod model is 90 % and this value for Markov model is 89 %.

5000 4000 3000 2000

CA-markov

1000

Geomod

0 1950 1960 1970 1980 1990 2000 2010 2020 2030 Time(year)

Fig. 7: Urban changes of Arak during about 7 decades Figure 7 shows urban changes of Arak from 1956 to 2025 and differences between Geomod and CA-Markov prediction.

1130

World Appl. Sci. J., 18 (8): 1124-1132, 2012

CONCLUSION The results indicate the capability of multi-temporal data which were used in the present study for mapping urban land use, detecting its changes and also predicting changes in future. Outputs of this study could be useful in land management and land use planning. In the present study, post-classification comparison was successfully employed for detecting and monitoring land use changes and Geomod and CA-Markov models were helpful for predicting land use changes. The outcomes of this research indicate that Landsat TM and IRS-P6 LISS-III images can be effectively used for generating accurate land use maps as the overall accuracies of all generated land use map were found to be over 90%. This study showed that the urban area of the study area has expanded significantly during 5 decades between 1956 and 2006 due to industrialization and population growth. A lot of people immigrated to Arak because of job opportunities during recent years. In the present study, post-classification comparison was successfully employed for detecting and monitoring urban expansion and changes occurred in urban areas. Urban expansion and subsequent degradation of farmlands and natural vegetation has been occurred and is occurring in different places in Iran and also in all over the world such as Bankok in Thailand and Shijiazhuang in China [25]. In the present study, CA-Markov and Geomod models had similar overall agreement between real and simulated maps. Both models had high potential in predicting urban changes in the city of Arak and its periphery. In general, Markov models have shown the capability for predicting land cover/land use change trends, regardless of whether or not the trend actually persists. Modelling methods provide useful information about probable changes in future and can be used to prevent some adverse effects on environment. The most important difference between Geomod and CA-Markov is that Geomod predicts only a one-way transition from one category to one alternative category, whereas CA-Markov has the ability to predict any transition among any number of categories. Therefore if the important dynamics of a landscape involve simultaneous gain or loss of a category or involve substantial interactions among more than two categories, then it would seem that CA-Markov is the better choice based on qualitative characteristics. Alternatively, it might be advisable to simplify the base data in order to focus on the major signal of change, which frequently is the

conversion from a single non-disturbed category to a single disturbed category, especially in the case of urban landscape modeling [14]. REFERENCES 1.

Cabral, P. and A. Zamyatin, 2006. Three land change models for urban dynamics analysis in Sintra-Cascais area. In the 1st EARSeL Workshop of the SIG Urban Remote Sensing, Berlin. 2. Weber, C. and A. Puissant, 2003. Urbanization pressure and modeling of urban growth: example of the Tunis Metropolitan Area. Remote Sensing of Environment, 86(3): 341-352. 3. http://undp-usa.blogspot.com/2007/06/state-ofworlds-populationunleashing.html 4. Rafiee, R., A. Salman Mahiny, N. Khorasani, A.A. Darvishsefat and A. Danekar, 2009. Simulating urban growth in Mashad City, Iran through the SLEUTH model (UGM). Cities, 26: 19-26. 5. Lopez, E., G. Bocco, M. Mendoza and E. Duhau, 2001. Predicting land-cover and land-use change in the urban fringe a case in Morelia city, Mexico. Landscape and Urban Planning, 55: 271-285. 6. Statistical center of Iran, http://www.sci.org.ir/ 7. Weng, Q., 2002. Land use change analysis in the Zhujiang Delta of China using satellite remote sensing, GIS and stochastic modelling. Journal of Environmental Management, 64: 273-284. 8. Yuan, F., K. Sawaya, C. Loeffelholz and E. Bauer, 2005. Land cover classification and change analysis of the twin cities (Minnesota) metropolitan area by multi temporal land sat remote sensing. Remote Sens. Environ, 98(2-3): 317-328. 9. Mas, J.F., 2003. An artificial neural networks approach to map land use/cover using Landsat imagery and ancillary data. In the proceedings of the International Geosciences and Remote Sensing Symposium IEEE IGARSS, Vol. 6, France. 10. Wu, Q., H. Li, R. Wang, J. Paulussen, Y. He, M. Wang, B. Wang and Z. Wang, 2006. Monitoring and predicting land use change in Beijing using remote sensing and GIS. Landscape and Urban Planning, 78: 322-333. 11. Alphan, H., 2003. Land-use change and urbanization of Adana, Turkey. Land degradation and development, 14(6): 575-586. 12. Lu, D., P. Mausel, E. Brondizio and E. Moran, 2004. Change detection techniques. International Journal of Remote Sensing, 25(12): 2365-2401.

1131

World Appl. Sci. J., 18 (8): 1124-1132, 2012

13. Veldkamp, A. and E.F. Lambin, 2001. Predicting land-use change. Agriculture, Ecosystems and Environment, 85: 1-6. 14. Pontius, R.G. and J. Malanson, 2005. Comparison of the structure and accuracy of two land change models. International Journal of Geographical Information Science, 19(2): 243-265. 15. Rahdary, V., A. Soffianian, S. Maleki Najfabdai, S.J. Khajeddin and A. Pahlavanravi, 2008. Land Use and Land Cover Change Detection of Mouteh Wildlife Refuge Using Remotely Sensed Data and Geographic Information System. World Applied Sciences Journal, 3(1): 113-118. 16. Fan, F., Y. Wang and Z. Wang, 2008. Temporal and spatial change detecting (1998-2003) and predicting of land use and land cover in Core corridor of Pearl River Delta (China) by using TM and ETM+ images. Environmental Monitoring and Assessment, 137: 127-147. 17. Shalaby, A. and R. Tateishi, 2007. Remote sensing and GIS for mapping and monitoring land cover and land-use changes in the Northwestern coastal zone of Egypt. Applied Geography, 27(1): 28-41. 18. Pontius, R.G., D. Huffaker and K. Denman, 2004. Useful techniques of validation for spatially explicit land-change models. Ecological Modelling, 179(4): 445-461. 19. Echeverria, C., D.A. Coomes, M. Hall and A. Newton, 2008. Spatially Explicit Models to Analyze Forest Loss and Fragmentation between 1976 and 2020 in Southern Chile. Ecological Modelling, 212(3-4): 439-449.

20. Houet, T. and L. Hubert-Moy, 2006. Modelling and projecting land-use and land-cover changes with a Cellular Automaton in considering landscape trajectories: an improvement for simulation of plausible future states. EARSeL proceedings, 5: 63-76. 21. Pontius, R.G., W. Boersma, C. Castella, K. Clarke, T. Nijs, C. Dietzel, Z. Duan, E. Fotsing, N. Goldstein, K. Kok, E. Koomen, C.D. Lippitt, W. McConnell, A.M. Sood, B. Pijanowski, S. Pithadia, S., Sweeney, T.N. Trung, A.T. Veldkamp and P.H. Verburg, 2008. Comparing the input, output and validation maps for several models of land change. The Annals of Regional Science, 42(1): 11-37. 22. Pontius, R.G. and P. Pacheco, 2004. Calibration and validation of a model of forest disturbance in the Western Ghats, India 1920-1990. GeoJournal, 61: 325-334. 23. Pontius R.G., L. Claessens, C. Hopkinson, A. Marzouk, E. Rastetter, L. Schneider and J. Vallino, 2000. Scenarios of land-use change and nitrogen release in the Ipswich watershed, Massachusetts, USA, 4th International Conference on Integrating GIS and Environmental Modeling, Alberta, Canada. 24. Stow, D.A., 1999. Reducing misregistration effects for pixel-level analysis of land-cover change. International Journal of Remote Sensing, 20: 2477-2483. 25. Madhavan, B.B., S. Kubo, N. Kurisaki and T. Sivakumar, 2001. Appraising the anatomy and spatial growth of the Bangkok Metropolitan area using a vegetation-impervious-soil model through remote sensing. International Journal of Remote Sensing, 22(5): 789-806.

1132