LANDSCAPE PATTERN ANALYSIS THROUGH METRICSwgbis.ces.iisc.ernet.in/energy/lake2010/Theme 1/vishnu_b.pdf · year was divided into 8 directions and then each direction into 13 concentric

Lake 2010: Wetlands, Biodiversity and Climate Change

22nd-24th December 2010 Page 1

LANDSCAPE PATTERN ANALYSIS THROUGH METRICS

Vishnu Bajpai1 , Ramachandra T. V.1,2,3,*

1. Centre for Ecological Sciences, Indian Institute of Science.

2. Centre for Sustainable Technologies, Indian Institute of Science

3. Centre for infrastructure, Sustainable Transportation and Urban Planning, IISc

*Corresponding author: [email protected]

Abstract

Land-use and land-cover (LULC) refers to the changes due to activities of humans, linked with global

Environment and climate change. Conditions in environmental systems with large measures of urbanization

are correlated, such as population density with built-up area .Urban sprawl refers to the unplanned,

uncontrolled spreading of urban area adjoining the edge of a city caused as a result of urbanization.

This study investigated the spatio-temporal dynamics of Bangalore’s landscape during the period (1999-

2008) using multi-temporal satellite images. Bangalore Administrative Boundary along with 10 km circular

buffer was taken as the study area. A maximum likelihood classification algorithm was used to classify the

images. After classification land cover statistics were evaluated for each class. The classified image of each

year was divided into 8 directions and then each direction into 13 concentric circles. Landscape metrics

were computed to describe the landscape pattern used in this study. The landscape metrics enabled the

description of the spatial regularities and trends, and constitute useful indirect indicators of the impact of

abrupt landscape changes. The study shows that there has been a 13.35% growth of pervious area during

the study period.

Keywords: Urban Sprawl, Urbanization, Landscape metrics

1. Introduction

Urbanization comprises population immigration to an already existing urban area that results in expansion

of that area at the cost of natural land.Urbanization has resulted in increased density in affected areas,

most notably in growing cities, and is often accompanied by a multitude of problems or consequences. One

of the most notable of these problems is urban sprawling. The extent of urbanization is called the sprawl

that drives change in land use patterns. Various studies were attempted on urban sprawl (The Regionalist,

1997; Sierra Club, 1998). In the developed countries (Batty et al., 1999; Torrens and Alberti, 2000; Barnes

et al., 2001, Hurd et al., 2001; Epstein et al., 2002) and also in developing countries for example China



(Yeh and Li, 2001; Cheng and Masser, 2003) and India (Jothimani, 1997 and Lata et al., 2001; Sudhira et

al..,2003) have studied urban sprawl.The parameter of quantifying urban sprawl is generally the built-up

area (Torrens and Alberti, 2000; Barnes et al., 2001; Epstein et al., 2002). Urban sprawl is also referred as

irresponsible, and often poorly planned development that destroys green space, increases traffic, contributes

to air pollution, leads to congestion with crowding and does not contribute significantly to revenue, a major

concern. Analysing the sprawl over a period of time will help in understanding the nature and growth of

urbanization.

Tools including geographic information systems (GIS) and remote sensing enable land planners, managers,

and ecologists to display and quantify changes in landscape structure that result from disturbances(Turner

and Carpenter, 1998).The objective of this paper is to assess land use change in study area from 1999 to

2008 and finding the role of landscape metrics for describing the changes in urban areas over time The

study demonstrates that the application of GIS and remote sensing coupled with statistical analyses, such as

arriving at Shannon's entropy helps in studying the sprawl and identifying regions having potential for

subsequent sprawl.

2. Study area

The study area encompasses Greater Bangalore with 10 km circular buffer (fig 1). Bangalore is one of the

fastest growing cities in India and is branded as ‘Silicon Valley of India’ for heralding and spearheading the

growth of Information Technology (IT) based industries in the country.(Sudhira et al.,2007).It is capital

city of Karnataka with an area of 741 sq. km., between the latitudes 12°39’00’’ to 13°13’00’’N and

longitude 77°22’00’’ to 77°52’00’’E. The population of Bangalore is about 8 million (Ramachandra and

Kumar, 2008). It is the fifth largest metropolis in India.



Figure 1: Study area

3. Data processing and methodology

In this study Landsat TM data were used.For the study data were preprocessed using GRASS software

which involves line correction, Geometric correction. Since unsystematic geometric errors remain in

commercially available remote sensing data, geometric rectification was needed to reduce the error before

land use classification .Maximum Likelihood classifier (MLC) was then used to classify temporal RS data

into 4 Land cover (LC) classes such as urban, vegetation, water and others. This method constitutes a

historically dominant approach to RS-based automated LC derivation (Gao et al., 2004) and has become

popular and widespread in RS because of its robustness (Hester et al., 2008). After classification land cover

statistics were evaluated for each class. Detailed analysis is done by dividing the city into eight zones

North-West-West (NWW),North-North-west (NNW) North-North-East (NNE), North-East-East (NEE),

South-East-East (SEE), South-South-East (SSE), South-South-West (SSW), South-West-West (SWW),and

then 13 concentric circles of 2km were drawn from the centre. In this way classified images was divided

into 8 directions and then 13 concentric circles were cropped and finally 104 regions were obtained from

each classified images. The cropped region obtained from each classified images were used for further

computation of spatial metrics.



Figure 2: Data Processing and methodology used in this study.

Table 1: Description of metrics used in this study

Sl

No.

Indicators Formula Description

1. Mean Patch Size

It measures the average area of

all patches in the landscape.

MPS > 0

2. Mean shape

Index)

Mean shape index (MSI)

measures the average patch shape

or the average perimeter-to-area

ratio, for a particular patch type



(class) or for all patches in the

landscape

MSI >=1

3. Mean Nearest

Neighbor

Distance

where h =nearest edge to edge distance

n =total no of patches of same type

Mean Nearest Neighbor

measures average patch to patch

distance.

MNN > 0, without limit.

4. Mean

Proximity

Index)

The mean proximity index

measures the degree of isolation

and fragmentation of the

corresponding patch type

MPI >=0.

5.

Interspersion

and

Juxtaposition

Index

0 < IJI <=100

IJI =0 when the corresponding

patch type is adjacent to only 1

other patch type and the number

of patch types increases.



eik =total length (m) of edge in landscape

between patch types (classes) i and k.

m =number of patch types (classes)

present in the landscape, including the

landscape border, if present.

IJI = 100 when the

corresponding patch type is

equally adjacent to all other patch

types.

4. Results

During the study period (1999 to 2008) urban density increased in almost every direction in each circle

which showed clear urban growth in the study area. On the other hand, there has been a decrease in the

vegetation, water bodies and others. The classified images are shown in figure 3 and the statistics are listed

in table 2. Several Landscape metrices were computed in this paper (Mean Nearest Neighbor distance,

Mean Patch Size, Mean Proximity Index, Mean Shape Index and, Interspersion and Juxtaposition Index.).



Figure 3: Classified Images from 1999 to 2008

Table 2: land use statistic



Figure 4: Mean Neareast Neighbour distance



Mean Nearest Neighbour measures average patch to patch distance. The value of this index decreased

from1999 to 2008.Less value of Mean nearest neighbor distance in landscape indicates that patches are less

insular in the landscape. From 1999 to 2008 values decreased, this showed 1999 to 2008 urban patches are

more aggregated. For circle 11 from 1999 to 2000 and 2003 in SWW direction value decreased, however in

2006 value increased but again in 2008 value decreased, which showed from 1999 to 2003 the urban

patches were aggregated but in 2006 some fragmentation happened but again in 2008 decreased value of

this index showed that urban patches are more compact in 2008, this pattern is seen in almost all direction

for all circles.

Figure 5: Mean Proximity Index

MPI = 0 if all patches of the corresponding patch type have no neighbors of the same type within the

specified search radius. MPI increases as patches of the corresponding patch type become less isolated and

the patch type becomes less fragmented in distribution.The general trend is that the value of this index

increased.In 1999 in circle 2 in NNE value is low which increased in all the years. In circle 3 in NNW



direction value increased from 1999 to 2008and it is max in 2008.Increased value of this index showed less

fragmentation in the landscape from 1999 to 2008.In circle 4 in SWW value is low in 1999a slight increase

is there in 2000 and 2003 but in 2006 there is a decrease which showed fragmentation and again in 2008

value increased which showed aggregation of urban patches in 2008.



Figure 6: Mean Patch size

Mean patch size, measures the average area of all patches in the landscape. If MPS is small it indicates a

fragmented landscape.In circle 2 i.e. near the city center, the NNE value increased in 2000,2003,2006 and

it is maximum in 2008(fig 6) which showed that urban patches are fragmented in earlier years but they are

becoming more compact in 2008. For circle 4 in SWW values decreased from 1999 to 2003 which showed

the fragmentation of patches but in 2006 and 2008 value increased and maximum in 2008,which showed

the aggregation of urban patche in 2008.



Figure 7: Mean Shape Index

Shape index measures the complexity of patch shape compared to a standard shape.MSI = 1 when all

patches of the corresponding patch type are circular (vector) or square (raster); MSI increases without limit

as the patch shapes become more Irregular. In circle 1 value increased from 1999 to 2000 in all directions

and in 2003,SWW showed a drastic increase which showed that sprawl has happened there but in NNW



value decreased and in 2006 SWW value decreased in 2008 value decreased in almost every direction

(fig.7) so from 1999 to 2006 patch shapes are very irregular but in 2008 urban patches become more

regular which showed the aggregation of urban patches in 2008.For circle 5 in 2006 in SSW value goes

upto 3.5 which showed urban patches become more irregular in this direction.for circle 4 in 1999 in SSW

value is high which showed in that direction patch shapes are irregular.

Figure 8: Interspersion and juxtaposition Index

Interspersion and juxtaposition index measures the extent to which patch types are interspersed, higher

values result from landscapes in which the patch types are well interspersed (equally adjacent to each

other), whereas lower values characterize landscapes in which the patch types are poorly interspersed

(disproportionate distributionof patch type adjacencies).The general trend showed the decrease in the value

of this index from 1999 to 2008.From 1999 to 2000 not much change is visible..In 2003 all values are

between 60 and 100. In 2006 value decreased up to 30 but in 2008 value of this index showed a drastic

decrease indicating urban patches are poorly interspersed.In circle1 the values decreased in all the

directions from 1999 to 2008, which showed the aggregation of urban patches at the city center in 2008 as

we conclude from other metrics also.



5. Conclusion

In this study we used multi-temporal analysis combined with landscape metrics for identifying the changes

of landscape pattern caused by the urbanization. Landscape metrics were very useful to detect landscape

pattern and its changes. There has been a 13.35% urban growth in study area from 1999 to 2008.Urban

patches were more dispersed in earlier years but patches are aggregating which showed that the study area

becoming more compact in 2008.

References

[1.] Batty M., Xie Y., and Sun Z., 1999. The dynamics of urban sprawl. Working Paper Series, Paper 15,

Centre for Advanced Spatial Analysis, University College London.

[2.] Barnes K. B., Morgan III J. M., Roberge M C., and Lowe S, 2001. Sprawl development: Its patterns,

consequences, and measurement. Towson University

[3.] Cheng, J. and Masser, I., 2003. Urban growth pattern modeling: a case study of Wuhan city, PR

China. Landscape and Urban Planning, 62, pp. 199 – 217.

[4.] DiBari, J. M., (2007), “Evaluation of five landscape-level metrics for measuring the effects of

urbanisation on landscape structure: the case of Tucson, Arizona, USA” Land. & Urban Plann., 79,

308-313

[5.] Epstein, J., Payne, K., and Kramer, E., 2002. Techniques for mapping suburban sprawl.

Photogrammetric Engineering and Remote Sensing, Vol. 63(9): pp 913 – 918.

[6.] Gao, J., Chen, H. F., Zhang, Y., Zha, Y. (2004), “Knowledge-based approaches to accurate mapping

of mangroves from satellite data.” Photogramm. Engg. & Remote Sens. 70(12), 1241–1248.



[7.] Hester, D. B., Cakir, H. I., Nelson, S. A. C., Khorram, S. (2008). “Per-pixel Classification of High

Spatial Resolution Satellite Imagery for Urban Land-cover Mapping.” Photo. Engg. & Remote

Sens. 74(4), 463–471.

[8.] Hurd, J. D., Wilson, E. H., Lammey, S. G., and Civco, D. L., 2001. Characterisation of forest

fragmentation and urban sprawl using time sequential Landsat Imagery. Proceedings, ASPRS

Annual Convention, St. Louis, MO. April 23-27, 2001.

[9.] Jothimani, P. 1997. Operational urban sprawl monitoring using satellite remote sensing: excerpts

from the studies of Ahmedabad, Vadodara and Surat, India. Paper presented at 18th Asian

Conference on Remote Sensing held during October 20-24, 1997 in Malaysia.

[10.] Lata K. M., Sankar Rao C. H., Krishna Prasad V., Badrinath K. V. S., and Raghavaswamy, 2001.

Measuring urban sprawl: a case study of Hyderabad. GISdevelopment, Vol. 5(12).

[11.] Li, L., Sato, Y., Zhu, H., 2003. Simulating spatial urban expansion based on a physical process.

Landscape Urban Plan. 64, 67–76.

[12.] McGarigal, K., and Marks, B. J., (1995), “FRAGSTATS: spatial pattern analysis program for

quantifying landscape structure.” USDA Forest Service General Technical Report, PNW-351.

[13.] McGarigal, K., Cushman, S.A., Neel, M.C., Ene, E., 2002. FRAGSTATS: spatial pattern analysis

program for categorical maps, www.umass.edu/landeco/research/fragstats/fragstats.html (last

accessed 21 June 2005).

[14.] Ramachandra T. V., and Kumar, U. (2008), “Wetlands of Greater Bangalore, India: Automatic

Delineation through Pattern Classifiers.” The Greendisk Environmental Jour., 1(26), 1-22

[15.] Sierra Club, 1998. The Dark Side of the American Dream: The Costs and Consequences of

Suburban Sprawl. (http://www.sierraclub.org/sprawl/report98/)

[16.] Torrens P. M., and Alberti M., 2000. Measuring sprawl. Working paper no. 27, Centre for Advanced

Spatial Analysis, University College London

[17.] Turner, M.G., Carpenter, S.R., 1998. At last: a journal devoted to ecosystems.Ecosystems 1, 1–5.

[18.] The Regionalist, 1997. Debate on Theories of David Rusk. Volume 2, Number 3.

[19.] Yeh, A. G. O., and Xia Li, 2001. Measurement and monitoring of urban sprawl in a rapidly growing

region using entropy. Photogrammetric Engineering and Remote Sensing, Vol. 67(1): pp83



Documents

LANDSCAPE PATTERN ANALYSIS THROUGH METRICSwgbis.ces.iisc.ernet.in/energy/lake2010/Theme 1/vishnu_b.pdf · year was divided into 8 directions and then each direction into 13 concentric