Simulating urban growth by two state modeling and connected network

Results and Discussion

Land Use Analysis
As explained land use analysis was performed using GMLC, with ground truth collected from various field sur-veys and supplemented by google earth. Land use analy-sis shows that urban area had phenomenally increased to 29.33% in 2012 of the entire region. Whereas, vegetation decreased drastically to about 33.68% for the entire area. Land use statistics are as tabulated in Table 4 below and the output of land use analysis is as described in Fig. 3. Urban land use change can be seen in form of clumped in filling growth in the city limits. Urban area mushrooms near the regions that have an urban class, and develops as a leapfrog along the highways and roads that connect the city with the periphery. Classically it can be observed that the urban land use spurt in the northeast. It can be attrib-uted to developments of amenities such as International airport and satellite towns. Southeast defines the develop-ment based on connecting corridor to another mega city indicating that transport corridor plays an important part in urban sprawl. Land use analysis highlights that the per-centage of urban land is increasing in all directions due to the policy decisions of (1) industrialization (2) boost to information technology and biotechnology sector in late 1990s and consequent housing developments in the periph-ery and unplanned outgrowth.
Accuracy assessment Accuracy assessment of the land use analysis was performed by generating the reference image through the 30% signatures developed using known ground truth data. The results of the accuracy assessment are tabulated in Table 5. Accuracy assessment indicated a good classification result with an overall accuracy of 80–90% and kappa close to 0.8. Land use data is then used as base layers for training the model, validation of model and prediction to the year 2020.
Modelling the spatial change of the urban extent in 2020
Land use data was then recoded into built-up and non built-up class, using the module r.reclass and further recoded to two classes and a binary image was generated as output from the GRASS. The drivers that are responsible for urban growth were digitised from google earth and manual data collection. These drivers were then used to derive the site suitability maps based on influence and dependence values of each driver. Multi-criteria evaluation is considered to develop the allocation suitability for urban growth based of a variety of attributes such as urban growth etc., These criteria’s are standardised to yield weight(s) for each of the




drivers. These weights act as input criteria for the influence of each driver on urban growth. Higher the weight rate of a particular driver urban growth is higher, lower the weight of a particular driver urban growth is lower. Further to obtain the realistic influence and weight characteristics the land use images were used as references. Based on recoded land use image of 2008, the influence of each driver in order to urbanise the area was tested keeping the reference images of 2010 and 2012. The condition was applied such that the spatial extents of current water bodies (2012) do not change during the modelling. The weights for each scenario was then obtained based on validation per pixel basis so that the developed semantics match the original land use. The land use supervised classified data and the generated simulated data for the years 2010 and 2012 is as shown in Fig. 4. The accuracy assessment was done using the land use classi-fied data of 2012 as a reference and simulated data for 2012 as actual data. The accuracy assessment and Kappa values calculated are as tabulated in Table 6. Once the validation reaches a threshold overall accuracy and kappa. Then 2012 is used as a base image to predict 2014–2020 for business as usual scenario. Population is then changes to match the other two scenarios and prediction was performed for 2014–2020. The results of each scenario is as shown in Fig. 5 and quanti-fied in Table 7.
Results of GEOMOD indicate that there would be an extensive growth along with various drivers across the study regions. Roads including the major highways such as state highway and national highway influence the majority of outgrowth or urban sprawl. Whereas there would be an extremely high growth density towards the International air-port. Results also indicated by 2018, the city infilling would reach a threshold value and the growth would start spread-ing extensively from the centre towards the rural regions adjoining its boundary. This would affect the rural dynamics considerably and provision of basic amenities such as pota-ble water and sanitation would become a challenge by 2020. Further, to understand the spatial arrangement of growth spatial metrics as indicators were analysed.

Spatial metrics as indicators of growth
Landscape metrics as described were calculated on a binary class land use data to understand the arrangements of urban class from 2008 to 2020. Landscape spatial metrics were selected based on its various properties, such as area, shape, contagion, edge. The results of each metric are discussed in the following sections:
A number of urban patches (NP) NP quantifies the num-ber of patches of the corresponding class. It does mean that if the value of this metric is 1 the entire landscape has only one class, if not any higher number indicates fragmentation of landscape. NP quantified for the study region showed that during 2008 and 2010, there were relatively larger patches in the landscape, when visually examined it was mainly in the outskirts (Buffer) of the city about 5000 patches. Which increased to an extent of 25,000 patches and was dominated by patch urban land use in 2012. Further, towards 2020 num-ber of patches consistently decreased indicative of the fact that the many urban patches in the neighbourhood expanded converting other land uses to urban land use, forming single urban patches, therefore reducing the count of the patches as infilling increases between the patches. This growth can be visualized in various parts of the study area such as the northeast and the southern part. Some parts of the city that show extensive urban growth are near the International airport, the emergence of the taluks such as Hoskote and Doddaballapura as developed urban corridors and the close proximity to the IT Corridor of the Silicon Valley, compared to other areas of the city. This argument remains the same in all scenario though scenario with 3% shows lesser infill-ing between patches but higher sprawl. The scenario with 7% shows that the dominant class would be urban by 2020 (Fig. 6).

Fig.5 Results of analysis for the years 2012, 2014, 2016, 2020 considering the rate of growth as 5%. b Results of analysis for the years 2012, 2014, 2016, 2020 considering the rate of growth as 3%
Normalized landscape shape index of urban class (NLSI)
NLSI metric signifies the shape of the class in the landscape. Simple shape implies that the land use class has become clumped and is the dominant urban class. Complex shape signifies the higher number of patches in the landscape of the particular class. Values close to 1 indicates that the class in fragmented and complex shape and have various land use in the neighborhood. Values close to zero indicates simple shape with the dominant class under consideration. NLSI calculated results show that the 5% growth scenario resulted in a value of 0.1 signifying complete degradation of all other class and dominance of urban patch by 2020, and with 7% scenario it showed the worse value of 0.05. But in the scenario of 3%, the value was significantly lower of 0.15. The simulated land use output for 2020 clearly brings out the fact that if the green spaces, open spaces are not well managed, Bangalore would face the crisis of these by 2020.
Total edge present in urban class This metric signifies again the edges formed by the patches, as patches increase and again coagulate form a large number of edges. As the number of edges increases, the presence and dominance of the class under consideration also increases. It can be seen that by 2020 all scenarios project a huge number of edges in the landscape. This again points out the fact that the land use will be dominated by urban (Fig. 7).
Clumpiness index Clumpiness index is a measure of aggregation or disaggregation. Values of this index range from 0 to 1.Higher the values higher the clumped patch. Lower the val-ues indicate the presence of other land use class. The quanti-fied results point out the fact that the entire landscape will have a dominant paved urban patch by 2020. The values close to 1 in all scenario (0.91 in a scenario with 7% growth, 0.83 in sce-nario 5% growth and 0.79 in the scenario of 3% growth) show a clumped urban growth in 2020. The discussion highlights that the development during 2008 to 2020 was phenomenal in North East and South West of the city due to Industrial devel-opment, planning of new infrastructure such as airport and due to setting up of Information technology and biotechnology arena in this regions that attracts the rural–urban migration for jobs from across the state and country and consequent spurt in housing colonies in the nearby localities.



 



 

 

TOP  » 
Citation : Bharath H. Aithal, S. Vinay, T. V. Ramachandra, 2018. Simulating urban growth by two state modelling and connected network. Modeling Earth Systems and Environment. © Springer Nature Switzerland AG 2018. Received: 30 April 2018 / Accepted: 11 August 2018. https://doi.org/10.1007/s40808-018-0506-1
* Corresponding Author :
Dr. T.V. Ramachandra
Energy & Wetlands Research Group, Centre for Ecological Sciences, Indian Institute of Science, Bangalore – 560 012, India.
Tel : +91-80-2293 3099/2293 3503 [extn - 107],      Fax : 91-80-23601428 / 23600085 / 23600683 [CES-TVR]
E-mail : tvr@iisc.ac.in, energy.ces@iisc.ac.in,     Web : http://wgbis.ces.iisc.ernet.in/energy, http://ces.iisc.ernet.in/grass
E-mail    |    Sahyadri    |    ENVIS    |    GRASS    |    Energy    |      CES      |      CST      |    CiSTUP    |      IISc      |    E-mail