EXPERT CLASSIFICATION BASED LAND USE / LAND COVER CHANGE DETECTION : A CASE STUDY OF DEHRADUN, INDIA KUMAR MINAKSHI 1 A. HARINDA LAKMAL2 , A.M.K.B. ABEYSINGHE 3 and V.K.DADHWAL 1 1 INDIAN INSTITUTE OF REMOTE SENSING , 4 KALIDAS ROAD , DEHRADUN UTTARAKHAND, INDIA, 91-135-2524118,
[email protected] ,
[email protected] 2 SURVEY DEPARTMENT , SRILANKA,
[email protected] 3 EARTH RESOURCE DEPARTMENT UNIVERSITY OF MORATUWA , SRI LANKA,
[email protected] KEY WORDS: Change Detection, Land use / land cover Classification , Expert Classification ABSTRACT This paper investigates the Spatio-Temporal change of Dehradun city area to determine the land use / land cover changes with in the six years of period from 2000 to 2006. Dehradun was a part of Uttar Pradesh, India. Upon the formation of the new state of Uttarakhand, Dehradun popularly known as “Doon valley” was declared as a capital of this new state. This study is an attempt for assessment of impact of capitalization of a small hill town using remote sensing tools. Imagery have to be registered, normalized and subset for input into the change detection process. In general two types of change detection techniques have been applied; the first one is those detecting binary change / non-change information, for example, using image differencing, image rotating, vegetation index differencing and PCA, and second are those detecting detailed ‘form-to’ changes, for example, using post-classification comparison, Change Vector Analysis and hybrid change detection methods and Expert Classification. A comparison and identification of the best results is achieved through accuracy assessment. It is observed that urban area increased by 47 % after the formation of the new state at the cost of fallow land and grass land. It is also observed that agriculture area has also considerably decreased. 1.0 INTRODUCTION Land-cover refers to the physical characteristics of earth’s surface, captured in the distribution of vegetation, water, soil and other physical features of the land, including those created solely by human activities e.g., settlements. Land-use refers to the way in which land has been used by humans and their habitat, usually with accent on the functional role of land for economic activities. The ability to forecast land-use and land-cover change and, ultimately, to predict the consequences of change, will depend on our ability to understand the past, current, and future drivers of land-use and land-cover change. These factors as well as other emerging social and political factors may have significant effects on future land use and cover. Patterns of land use, land-cover change, and land management are shaped by the interaction of economic, environmental, social, political, and technological forces on local to global scales A variety of change detection techniques have been developed, and many have been summarized. Due to the importance of monitoring change of Earth’s surface features, research of change
detection techniques is an active topic, and new techniques are constantly developed. The objective of change detection is to compare spatial representation of two points in time by controlling all variances caused by differences in variables that are not of interest and to measure changes caused by differences in the variables of interest (Green et al. 1994). The basic premise in using remotely sensed data for change detection is that changes in the objects of interest will result in changes in reflectance values or local textures that are separable from changes caused by other factors such as differences in atmospheric conditions, illumination and viewing angles, and soil moistures. Because digital change detection is affected by spatial, spectral, thematic and temporal constraints, and because many change detection techniques are possible to use, the selection of a suitable method or algorithm for a given research project is important, but not easy. Remote sensing can provide an important source of land use land cover information through which land use extract with ease through new generation satellite systems. The traditional change detection procedure is the postclassification comparison, aiming to find out the difference between the classified images of two different dates (Singh, 1989). Some author, however, proposed to perform this detection by image differencing (Jensen et, al. 1982; Singh 1989). This study is an attempt for assessment of impact of capitalization of a small town using remote sensing tools. Objective of the study is to determine the land use / land cover changes during the specific period, in a particular area by using remote sensing techniques. This paper investigates the Spatio-Temporal change of Dehradun city area to determine the land use / land cover changes with in the six years of period from 2000 to 2006. 2.0 STUDY AREA AND DATA USED The Study area lies between 3361281.41 to 3550992.91 N coordinates and 211069.819 to 222726.319 E coordinates with respect to the projection of zone no. 44N UTM (Universal Transverse Mercator) on WGS 84 (World Geodetic System) datum. It is covering approximately 120 Sq. km. of Dehradun city area. Satellite data of LANDSAT 7 ETM+ of the year 2000 (Path 146 Row 039 Date 25 November 2000 )and IRS ID of 2006 ( 2nd December 2006) of the same season were used for the above case study . The study area falls in Sheet No. 53 F/15 and 53 J/3 with the Scale of 1:50000 published by Survey of India. The study area is illustrated in Figure 1. 3.0 METHODOLOGY The temporal data of the study pertained to two different satellite and sensors .Hence the Imageries had to be co-registered, normalized and subset for input into the classification and change detection process. Training samples were collected from the field and input into the system for supervised classification. Rules were developed for expert classification for improving the results of supervised classification. The classified result was finally input into the change detection process to quantify the land use change that has occurred over a period of six years after the formation of the new state. The methodology flowchart is illustrated in Figure 2. 4.0 ANALYSIS AND RESULTS The training samples were analyzed for the signature separability. Both the graphical and statistical methods were used for separability analysis. The signatures which could not be separated were deleted and new pure homogenous samples were selected. The average transformed
divergence figures for classification of ETM+2000 image was 1948.02 and for LISS III of 2006 was 1990.64. The overall accuracy achieved was 94.3 % and 96.1 % and kappa statistics was 0.93 and 0.95 respectively for the 2000 and 2006 images. Both of the Accuracy assessments indicate the high Kappa values which were close to one (1). It means that the true agreement (i.e. observed accuracy) approaches to 1 and chance agreement approaches to 0. The classified images are presented in Figure 3 and 4 and the change difference image is presented in figure 5. Expert classification could be performed by using the IMAGINE Expert Classifier. The expert classification software provides a rules-based approach to multispectral image classification. In essence, an expert classification system is a hierarchy of rules, or a decision tree, that describes the conditions under which a set of low level constituent information gets abstracted into a set of high level informational classes. A rule is a conditional statement, or list of conditional statements, about the variable’s data values and/or attributes that determine an informational component or hypotheses. Fig 6 illustrates the expert Classification Hypothesis and rules. Table 1 indicates the quantitative change of each classes and the graph in figure 7 illustrates the area change in each category. 5.0 CONCLUSIONS This case study successfully shows that change detection techniques can be applied to land use land cover change using remote sensing data. Results from the hybrid classification show that area of urban, degraded and dense vegetation have increased over the period whereas fallow, agriculture grassland and stream were decreased during the period of 2000 – 2006. There is a very large change in Agriculture and fallow areas in the study area, indicating that agriculture area has been sacrificed to sustain the pressure of urbanization. It is observed that urban area increased by 47 % after the formation of the new state at the cost of fallow land and grass land. 6.0 ACKNOWLEDGEMENTS The first author wishes to express her gratitude to Director National Remote Sensing Agency (NRSA) and DEAN, Indian institute of Remote Sensing (IIRS) for constant guidance and encouragement . The second and third author are also grateful to Director National Remote Sensing Agency (NRSA ) and DEAN, Indian institute of Remote Sensing for constant guidance and encouragement for giving them opportunity to undergo training programme at IIRS. 7.0 REFERENCES Coppin, P. & Bauer, M. 1996. Digital Change Detection in Forest Ecosystems with Remote Sensing Imagery. Remote Sensing Reviews. Vol. 13. p. 207-234. Furby, S. L., 2002. Land Cover Change: Specification for Remote Sensing Analysis. National Carbon Accounting System Technical Report 9, Australian Greenhouse Office, Canberra, Australia. http://www.greenhouse.gov.au/ncas/reports/pubs/tr09final.pdf Green, et al. 1994. Using Remote Sensing to Detect and Monitor Land-Cover and Land-Use Change. Photogrammetric Engineering & Remote Sensing. Vol. 60. No. 3. p. 331-337. James B. Campbell (1996, 2002); ‘Introduction to Remote Sensing (third edition); the Guilford press, a division of Guilford Publication, Inc 72 spring street New York. NY10012. John R Jensen. 1990. “Interdictory Digital Image Processing: A Remote Sensing Perspective”. 2nd Edition. Englewood cliffs, New Jersey.
Lillesand TM and Keifer W (1994); ‘Remote Sensing and Image Interpretation’; New York: John Wiley. Macleod & Congalton. 1998. A Quantitative Comparison of Change Detection Algorithms for Monitoring Eelgrass from Remotely Sensed Data. Photogrammetric Engineering & Remote Sensing. Vol. 64. No. 3. p. 207 - 216. MD. Rejaur Rahman, A. H. M. Hedayutul Islam, MD. Shareful Hassan (2004); ‘Change Detection of Winter Crop Coverage and the Use of Landsat data with GIS’; The Journal of Geo- Environment Vol.4, PP1-13. Singh .A (1989); ‘Review Article: Digital Change Detection Techniques using Remotely Sensed Data’; International Journal of Remote Sensing; Vol.10 PP 989- 1003. Steven M.D (1987); ‘Ground Truth: An Under view’; International Journal of Remote Sensing, Vol.8 PP 1033- 1038. Wallace and Furby 1994. Assessment of Change in Remnant Vegetation Area and Condition. Report to the LWRRDC project ‘Detecting and Monitoring Changes in Land Condition Through Time using Remotely Sensed Data’, CSIRO MIS Technical Report. (http://www.cmis.csiro.au/rsm/research/remveg/vegassess.htm) Thandar Htoon. & Minakshi Kumar 2007. Urban Spatial Pattern from High Resolution Data using Spatial and Spectral Properties. Project Report submitted to the IIRS, Dehradun, India. (Unpublished) Table 1 Comparison of Land Use / Land Cover Changes
Class Name
URBAN AREA DEGRADED DENSE VEGETATION STREAM FALLOW GRASSLAND AGRICULTURE
Landsat Data (2000) LISS 3 Data(2006) Difference No. of Area No. of Area No.of Area Percentage Pixel (Sq. m) Pixel (Sq. m) Pixel (Sq. m) (%) 33738 961533 49634 1414569 15896 453036 47.12 6600 188100 7555 215317.5 955 27217.5 14.47 25334 5974 37698 6948 31717
722019 170259 1074393 198018 903934.5
35171 5937 27496 3802 18481
1002373.5 9837 169204.5 -37 783636 -10202 108357 -3146 526708.5 -13236
280354.5 -1054.5 -290757 -89661 -377226
38.83 -0.62 -27.06 -45.28 -41.73
Figure 1 Location Map Of Study Area
Figure 2 : Methodology Flowchart
Figure 3 : Classified Image - Landuse Landcover 2000
Figure 4 : Classified Image Landuse Landcover 2006
Figure 5 : Change Image 2000- 2006
Figure 8 : Rules and Hypothesis for expert classification
Fig 9 Comparison of Land Use / Land Cover Changes
