See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/235428308

Land Use Land Cover Classification and Change Detection Using High
Resolution Temporal Satellite Data

Article · November 2012

CITATIONS READS

96 37,203

2 authors, including:

Sopan T Ingle
North Maharashtra University
62 PUBLICATIONS 1,168 CITATIONS

SEE PROFILE

All content following this page was uploaded by Sopan T Ingle on 29 May 2014.

The user has requested enhancement of the downloaded file.

 Kaul and Sopan et al / Journal of Environment (2012), Vol. 01, Issue 04, pp. 146-152 ISSN 2049-8373

Research Paper

Land Use Land Cover Classification and Change

Detection Using High Resolution Temporal Satellite Data
Harshika A. Kaul 1* and Ingle Sopan2
1
Department of Environmental Sciences, Moolji Jaitha College, Jalgaon, Maharashtra
2
School of Environmental and Earth Sciences, North Maharashtra University, Jalgaon, Maharashtra
*E-Mail: hakaul@gmail.com

Abstract

Land use and land cover change has become a central component in current strategies for managing natural resources
and monitoring environmental changes. To maintain the present natural resources and to understand the causes and
consequences of over exploitation of soil and water resources the land use, a land cover mapping and monitoring was
done in the study area i.e. Jalgaon District. In this study, satellite images for March and November 2007 were used for
LULC (Land Use/ Land Cover) Supervised classification. For the classification purposes, seven LULC classes
were decided. For Accuracy Assessment Classification error matrix and KAPPA analysis was done. Change Detection
between both the images for all the land use and land cover classes were computed. The most extensive land cover
category of the District in November is agricultural land i.e. 49.43%. The second most extensive land cover category is
barren land, 28.31% in March. Saline area shows significant difference in both seasons. The major change (+ 9.02%)
identified, in the study, was in Agricultural area from pre-monsoon to post-monsoon season. The overall accuracy of
classification methodology is 89% and 91.02% and Kappa statistics is 0.86 and 0.88 respectively for the March and
November images.

Keywords: Land Use, Land Cover, Supervised Classification, Change Detection, Accuracy Assessment

1. Introduction
Land cover refers to the physical characteristics of Earth's activities on his natural resource base over time. In
surface, captured in the distribution of vegetation, water, situations of rapid and often unrecorded land use change,
soil and other physical features. Landuse refers to the way observations of the Earth from space provide objective
in which land has been used by humans and their habitats information of human utilization of the landscape. Over
(such as agriculture, settlements, industry etc.). Although the past years, data from the Earth sensing satellites have
land use is generally inferred based on the cover, yet both become vital in mapping the Earth's features and
the terms land use and land cover being closely related are infrastructures, managing natural resources and studying
interchangeable. For example, settlement is cover but if we environmental change (Zubair, 2006).
include buildings whether it is being used for residence or
industrial activity, it shows the land use component Rao & Narendra (2006) and Boakye et al (2008) have done
(Chaudhary et al, 2008). the mapping of land use and land cover changes using
unsupervised classification methodology. They used
Land use and land cover change has become a central ERDAS imagine software for classification and final maps
component in current strategies for managing natural preparation. There are many studies found in which many
resources and monitoring environmental changes. Viewing researcher or scientists have used ERDAS imagine
the Earth from space is now crucial to understand man's software for the classification purpose. Asadi et al (2010),

Available online at www.scientific-journals.co.uk 146

 Kaul and Sopan et al / Journal of Environment (2012), Vol. 01, Issue 04, pp. 146-152 ISSN 2049-8373

created spatial digital database consisting of Land Use/ carried out.

Land Cover (LULC) using IRS (Indian Remote Sensing)-
ID PAN and LISS (Linear Imaging Self Scanning Sensor) 2. Methodology
III merged satellite data of Municipal Corporation of
Hyderabad, India. Data input and conversion was done in 2.1. Study Area
ERDAS image processing software. Database creations
and analysis were done in Arc/Info and ArcView GIS The study area i.e. Jalgaon, is located on the North-West
softwares. Yu et al (2007) used Landat MSS Image (1976), region of the state of Maharashtra, India and bounded by
Landsat TM Image (1990) and IRS-ID LISS III (2005) longitude 74°45”-76°28” East and latitude 20°-21° North.
images for landuse/ cover mapping by visual interpretation It is situated in the Tapi River basin stretching nearly in
and further analysis was carried out in ERDAS IMAGINE breadth from 128-144 km. The total area under irrigation is
in Birahi Ganga Sub-Watershed of the Garhwal Himalaya, 97000 ha. Out of this area, surface irrigation accounted for
India. 96000 ha and well irrigation for 1000 ha. Three major
irrigation projects, 11 medium irrigation projects, 62 minor
Unlike unsupervised classification many scientists like projects and 315 percolation tanks are completed in
Rao & Narendra (2006), Remi et al (2007), Chaudhary et Jalgaon District. As per 2001 census the total population of
al (2008) and Kim et al (2008) have applied supervised the district was 36,79,936. The annual rainfall is less than
classification in their studies. They reported that land use 500 mm. The maximum summer temperature varies from
land cover classification with maximum classification 39-43 °C and minimum winter temperature is found to be
accuracies were acquired by using Maximum Likelihood 12-15 °C. Relative humidity varies between 12% and 87%
Classification (MLC) decision rule. (Kaul & Ingle, 2011).
Some studies on sustainable watershed management have
2.2. Remote Sensing Data
been done in areas where resources are optimally utilized
for the benefit of the people and development of a region, Digital data for the present study, Indian Remote Sensing
as a whole. To achieve optimal utilization of resources, (IRS- P6) Linear Imaging Self Scanning Sensor (LISS-III)
studies were undertaken to decide alternative land use images, for March 2007 and November 2007 were
options in a watershed by visual interpretation techniques procured from National Remote Sensing Centre, Hydera-
using GIS. A similar study was undertaken in National bad, India. The images were used for identification and
Capital Region (NCR) Delhi, where urban LULC change classification of pre and post monsoon surface waterlogged
detection was done by Mohan (2005), under planning for areas and salt affected lands in the study area. As shown in
rural and urban communities. Table 1 each satellite data scene has informat-ion of four
spectral bands which correspond to Green (G), Red (R),
A project done by NRSA, Hyderabad (2007) on Geoinfor- Near Infra Red (NIR) and Short Wave Infrared (SWIR).
matics based Landuse and Landcover change in Khamman The spatial resolution for the G, R, NIR and SWIR bands
District of Andhra Pradesh, India, by Social Forestry is 23.5 m. In the present study, False Colour Composite
Institute of ITC Limited, PSPDC Unit, and Bhadrachalam. (FCC) constituted by four bands SWIR, NIR, R and G
The study was carried out using IRS ID and IA satellite bands corresponding to both pre and post- monsoon
data of LISS III, LISS IV and LISS I sensors using visual seasons were used in all the methodologies applied in this
interpretation technique. Some scientists like Thornton study related with satellite data.
(2002) and Herold et al (2006) also used visual interpretat-
ion technique to prepare a detailed land use land cover 2.3. Collateral Data
map.
Survey of India (SOI) Toposheets at 1:50,000 scale was
State Forest Report Maharashtra (2005) shows that total used for the preparation of the base maps and for remote
geographical area of Maharashtra is 30,771,000 hectares, sensing data interpretation. Further, irrigation project maps
in which land unavailable for cultivation is 19.27%, land were procured from Tapi Irrigation Development Corpora-
under cultural waste is 2.94%, fallow lands other than tion (TIDC), Jalgaon.
current fallows is 3.81% and land under current fallows is
3.87%. 2.4. Land Use and Land Cover

To maintain the present natural resources and to underst- The temporal satellite data IRS P-6 LISS III of the study
and the causes and consequences of over exploitation of area is pertained to two different seasons, pre-monsoon
soil and water resources the land use and land cover and post-monsoon. Hence the images had to be co-
mapping and monitoring was required essentially in the registered normalised and subset for input into the
study area i.e. Jalgaon, Indian, hence the current study was classification and change detection process. Maximum

Available online at www.scientific-journals.co.uk 147

 Kaul and Sopan et al / Journal of Environment (2012), Vol. 01, Issue 04, pp. 146-152 ISSN 2049-8373

Likelihood classifier was used in the supervised classifica- After Supervised classification initial LULC map is edited
tion method in this particular study. Many literatures on the basis of ground verification of doubtful areas and
reviewed for the selection of the appropriate and best some classes were recoded into their respective classes.
classification method. Rao & Narendra (2006), Remi et al Clump and Eliminate process were run to get the final
(2007) and Chaudhary et al (2008) have widely used LULC map in ERDAS IMAGINE. Ground truthing/
popular supervised classification technique, the maximum verification were done on those particular areas, which
likelihood classifier for the LULC classification in their were not clear in the classification. In this process, land
respective studies. use and land cover maps were prepared initially and the

Table 1. Satellite Images and Their Characteristics Used in the Study

Spectral Composition Swath Spectral Resolution/

S. No. Image Type Sensor Source
(No. of Bands) (km) Precision Scale (m)
G-0.52-0.59
R– 0.62-0.68 141
1 IRS P6 Mar. 2007 LISS-III 23.5 (all bands) NRSC, Hyderabad
NIR– 0.76-0.86
SWIR–1.55-1.70
G– 0.52-0.59
R– 0.62-0.68 141
2 IRS P6 Nov. 2007 LISS-III 23.5 (all bands) NRSC, Hyderabad
NIR– 0.76-0.86
SWIR–1.55-1.70

Jonathan et al (2007) decided that the Maximum Likeliho- confusing spots were identified. For each class i.e.
od Classification Algorithm (ML) would be an appropriate agriculture, forest, water bodies, settlement and barren
choice, since it is an extremely simple and easily land, at least ten points were marked. An exhaustive
implemented algorithm, but at the same time, is very well ground truthing of study area was carried out with the help
known and has already been successfully applied to a of GPS and made some corrections subsequently. The
broad range of remote sensing problems (Lillesand & ground reference data used for the image classification was
Kiefer, 2000). Therefore in this particular study, supervi- further applied in accuracy assessment purpose.
sed classification (Maximum Likelihood Classifier) is
done on the image processing software ERDAS Imagine After ground truthing, Accuracy assessment was carried
9.1 for the land use land cover classification. out. It is most important aspect to assess the reliability of
map. No image classification is said to be complete unless
For the classification I level, seven LULC classes i.e. its accuracy has been assessed. To determine the accuracy
Barren land, Settlement, Salt Affected Area, Forest land, of classification, a sample of pixels is selected on the
Waterbody, Harvested land and Agricultural land were classified image and their class identity is compared with
classified. The reason for using these seven classes is the ground reference data.
that, at this more generalized first and second level, an
accuracy in interpretation can be attained that will Evaluating the quality of a classification result is of high
make the land use and land cover data comparable in importance in remote sensing since it gives evidence of
quality to those obtained from other methods used in this how well the classifier is capable of extracting the desired
study. For land use and land cover data needed for the objects from the image. In this study the classification
identification and mapping of Salt affected and error matrix is used which is the common means of
waterlogged areas, the choice of these classes are expressing classification accuracy. The second technique
appropriate and the accuracy of interpretation at the used for accuracy assessment is the KAPPA analysis.
generalized first level is satisfactory when the interpreter KAPPA analysis is discrete multivariate technique which
makes the correct interpretation 85-90% of the time. And calculates the producer’s and user’s overall accuracy, as
generally, information at level I is of interest to users who well as the Kappa accuracy level. After accuracy assessm-
desire data on a nationwide, interstate or district wide basis ent and correction the class wise area of March 2007
(Anderson et al, 2001). image and November 2007 image was calculated (Table
2).
The classification is done for all the 15 major and medium
irrigation projects of Jalgaon District and the whole district In land use land cover classification change detection is
area. For Supervised classification total of 123 training very important process which is done after the completion
sites were made by demarcating a polygon or an area of of calculation and analysis of the land use land cover
interest for all the seven land use and land cover types. classes from pre and post monsoon images i.e. March 2007

Available online at www.scientific-journals.co.uk 148

 Kaul and Sopan et al / Journal of Environment (2012), Vol. 01, Issue 04, pp. 146-152 ISSN 2049-8373

and November 2007. By using Change Detection option in 3.1.1. Extent of Land Cover Categories in District Area
ERDAS IMAGINE software, the differences between two
images of pre and post monsoon for all the seven land use Jalgaon District comprising 11,765 sq. km geographical
and land cover classes were computed. The detailed area, have 15 Talukas (Local Jurisdiction areas falling
methodology is illustrated in flowchart in Figure 1. under the administrative powers of District Magistrate) and

Table 2. Land Use Land Covers Classes of Jalgaon District

March November
S. No. Classes
Sq. km % Sq. km %
1 Salt Affected Area 588.25 5.00 291.70 2.48
2 Barren Land 3330.60 28.31 2927.13 24.88
3 Harvested Area 1764.75 15.00 1176.50 10.00
4 Water Body 211.77 1.80 258.83 2.20
5 Agricultural Area 4754.23 40.41 5815.44 49.43
6 Forest Area 941.20 8.00 1120.02 9.52
7 Settlements 175.29 1.49 175.20 1.49
Total 11765.00 100.00 11765.00 100.00

*Percentage is calculated from the total geographic area of Jalgaon District

Figure 1. Methodology for Land Use/ Land Cover Map Generation

3. Results and Discussion 1519 villages. According to 1951 census the population
was 14,71,351 which became more than double in 2001
3.1. Land Use and Land Cover Mapping and Change census i.e. 36,82,690. The forest area in the district in
Detection 1956-57 was about 1620 sq. km i.e. 13.92% (Dist. Gazett-

Available online at www.scientific-journals.co.uk 149

 Kaul and Sopan et al / Journal of Environment (2012), Vol. 01, Issue 04, pp. 146-152 ISSN 2049-8373

eer, 1956-57) but in 2005-06 it was estimated about (5.0%) and 291.7 sq. km (2.48%) in November
10.90% of the total geographic area (State of Forest which shows significant difference and will be discussed
Report, 2005). in forthcoming sections of the paper. The above mentioned
results indicate that there is significant difference in the
The supervised classification conducted on the March and pre-monsoon season and post-monsoon season, which
November, 2007 images (Figure 2 a-b), the forest area in can be clearly seen in the Figure 2 (a and b). As shown in
the district specifically shows greater variations. As shown Table 2, water body covers 258.83 sq. km (2.2%) in post-
in Table 2, the forest area calculated in November 2007 monsoon season and 211.77 sq. km (1.8%) in pre monsoon
image was 1120.02 sq. km i.e. 9.52% but in March 2007 season. The last land cover/ land use class classified
image the forest area was decreased to 8.0% because, the by the supervised classification is settlement. Settlement is
forests on the Satpuda mountains become naked due to an area of human habitation, which has a cover of
leaves fall off the trees in this season. So, the mountains buildings and network of transport and other civic
appear barren in this season. As there will not be much amenities. The total area under settlements is estimated
difference between the two seasons of a year as confirmed about 175.2 sq. km which is 1.49% of the total district
by ground truthing, but as compared to State of Forest area.
Report 2005, the forest area in the year 2005 was estimated
10.90% which has decreased by 1.38% within two years 3.2. Land Use/ Land Cover Changes and Accuracy
(2005-2007). Assessment of Jalgaon District

Figure 2. Land Cover Map of Jalgaon District, Maharashrta

Table 2 shows that the most extensive land cover category Table 3 shows the changes in various land use land cover
of the District as in November image (Post-monsoon) and categories (in sq. km and percentages) during the pre-
March image (Pre-monsoon) was agricultural land i.e. monsoon and post-monsoon period, in the year 2007. The
49.43% and 40.41% respectively. The difference between major change identified in the study was in Agricultural
the two seasons here in the same class was due to the area. This area has been increased by 9.02% from pre-
harvesting process that generally took place in March. monsoon to post-monsoon season, which is followed by
That’s why the harvested land in March is 1764.7 sq. km the next major change in harvested land which has been
or 15% of the total area, while in post monsoon season, it decreased by 5.0% from pre-monsoon to post-monsoon
is reduced to 1176.5 sq. km or 10% of its total area. But favouring the fact that farmers harvest their crops in pre-
according to district gazetteer of 1956-57, the agricultural monsoon season indicating the cropping pattern of India
area mentioned was 7769.8 sq. km which shows that there which is in accordance with the country's climatic
was slight reduction in agricultural practices during 50 conditions.
years (1957-2007) which may be due to changing cropping
pattern or people have shifted to other occupation. Also, in post monsoon season barren and saline areas
had significant reductions due to availability of water
As shown in the Table 2, the second most extensive land which lead to revegetation in barren lands, and the salts of
cover category is barren land, which had covered about saline lands also dissolve in water and leaches out in to
3330.6 sq. km (28.31%) in March and 2927.13 sq. km the ground and mixes with underground water or if the
(24.88%) in November. This is followed by harvested land water level is high it dissolve with water and saturates on
and forest area. Saline area covers 588.25 sq. km in March the upper level of ground but cannot be remarkably

Available online at www.scientific-journals.co.uk 150

 Kaul and Sopan et al / Journal of Environment (2012), Vol. 01, Issue 04, pp. 146-152 ISSN 2049-8373

identified. As soon as water evaporates in summer season irrigation projected area separately.
or pre-monsoon season salts again become visible as
white patches on the soil surface (WALMI, 1995). This study successfully shows that the change detection
Therefore, the change recorded for saline areas is about - techniques can applied to land use land cover change using
4.5%. The barren land shows about 3.43% decrease from remote sensing data. The study shows that the major
March to November due to revegetation in post monsoon change identified in the study area is in Agricultural area
season. 9.02% increased from March to November. There are

Table 3. Change Detection Statistics and Accuracy Assessment for Classification in the District Area

Change Producers Accuracy Users Accuracy

Classes
Sq. km % March November March November
Salt Affected Area -296.55 -2.52 75.00 100.00 100.00 100.00
Barren Land -403.47 -3.43 95.65 98.00 95.65 89.09
Harvested Area -588.25 -5.00 92.31 82.86 92.31 96.67
Water Body 47.06 0.40 50.00 75.00 100.00 100.00
Agricultural Area 1061.21 9.02 100.00 91.80 82.35 90.32
Forest Area 178.82 1.52 85.71 97.44 85.71 90.48
Settlements 0.00 0.00 89.66 69.57 83.87 88.89

March- (Overall Classification Accuracy) = 89.00% (Overall Kappa Statistics 0.86)

November-(Overall Classification Accuracy) = 91.02% (Overall Kappa Statistics 0.88)

The statistics of change detection shows (Table 3), there is some changes also reflected in the study in the saline areas
only 0.4% increase in the water body from pre-monsoon to and water body from pre-monsoon to post-monsoon
post-monsoon and no change has shown in the settlement season. The overall accuracy achieved in the study is 89%
class. for March image and 91.02% for November which is
satisfactory. Both of the accuracy assessment indicate
The results in Table 3 show that the performance of the Kappa coefficient nearer to one.
classification methodology is in general quite good,
reaching an overall accuracy of 89% and 91.02% and The study shows that lack of management and spatial
Kappa statistics is 0.86 and 0.88 respectively for the March information in irrigation projects, which may be alleviated
and November images. Both of the accuracy assessments with remote sensing data that can provide opportunities for
indicate high Kappa values which are close to one (1). It periodical survey of land use/ land cover changes and their
means that the true (i.e. observed value) agreement spatial distribution.
approaches to 1 and chance of agreement approaches to 0.
Considering individual classification accuracies, the References
"Saline Area" class in November is most accurately
classified with 100% both for producer’s and user’s Anderson, J.R., Hardy, E.E., Roach, J.T., and Witmer, R.E.
accuracy. (2001) A Land Use and Land Cover Classification
System for Use with Remote Sensor Data. Geological
4. Conclusion Survey Professional Paper 964, A Revision of the Land
Use Classification System as Presented in U.S. Geological
The results obtained from the study confirm that IRS P-6 Survey Circular, 671, USGS (U.S. Geological Survey).
LISS III data are indeed appropriate for performing
regional scale I level LULC assessments, due to its Asadi, S.S., Rao Hanumantha, C.H., Prasad Lakshmi, T.,
moderate spatial resolution and excellent over all quality. and Reddy Anji, M. (2010) Evaluation of physical charact-
The methodology used for the study is capable of eristics using geomatics: A Case Study. Indian Journal
generalizing quite well over the entire Jalgaon District, an of Science and Technology, 3(4), pp. 450-454.
area of 11,765 sq. km accurately identifying areas of
Agriculture, Harvested, Barren, Saline, Forest, Water Body Boakye, E., Odai, S.N., Adjei, K.A., and Annor, F.O.
and Settlement. Moreover it was also capable of classifyi- (2008) Landsat Images for Assessment of the Impact of
ng the different classes in all the 15 major and medium Land Use and Land Cover changes on the Barekese

Available online at www.scientific-journals.co.uk 151

 Kaul and Sopan et al / Journal of Environment (2012), Vol. 01, Issue 04, pp. 146-152 ISSN 2049-8373

Catchment in Ghana. European Journal of Scientific ative of ITC Limited, PSPD Unit, Bhadrachalam, A
Research, 22(2), pp. 269-278. Project under Clean Development Mechanism,
UNFCCC. Department of Space, Government of India,
Chaudhary, B.S., Saroha, G.P., and Yadav, M. (2008) Hyderabad.
Human Induced Land Use Land Cover Changes in
Northern Part of Gurgaon District, Haryana, India: Natural Rao, K.N., and Narendra, K. (2006) Mapping and Evaluat-
Resources Census Concept. Journal of Human Ecology, ion of Urban Sprawling in the Mehadrigedda Watershed in
23(3), pp. 243-252. Visakhapatnam metropolitan region using Remote Sensing
and GIS. Current Science, 91(11), pp. 1552-1557.
Maharashtra State Gazetteer (1956-57) District Gazetteer,
Jalgaon. Government of Maharashtra, Cultural Affairs and Remi, A., Huang, S., Jiren, L., Herve, Y., and Yvew-
Tourism Department, Mumbai. Louis, D. (2007) Large Scale Land Cover Map Generation
based on low and medium resolution, Envisat ASAR Data,
Yu, H., Joshi, P.K., Das, K.K., Chauniyal, D.D., Melick Application of Poyang Lake area (JIANG XI, P.R. China).
D.R., Yang, X., and Xu, J. (2007) Land Use/ Cover Proc. Envisat Symposium 2007, April 23-27, 2007,
Change and Environmental Vulnerability Analysis in Montreux, Switzerland (ESA SP -636, July 2007).
Birahi Ganga Sub-Watershed of Garhwal Himalaya.
Tropical Ecology, 48(2), pp. 241-250. State Forest Report (2005) State Forest Report. Forest
Department, Government of Maharashtra, India.
Herold, M., Latham, J.S., Di Gregorio, A., and Schmullius,
C.C. (2006) Evolving Standards in Land Cover Characteri- Thornton, L. (2002) 3rd Generation, Statewide, Detailed
zation. Journal of Land Use Science, 1(2-4), pp. 157-168. Land Use Land Cover for New Jersey. A Project by
New Jersey Department of Environmental Protection.
Jonathan, M., Meirelles, M.S.P., Berroir, J.P., and Herlin,
I. (2007) Regional Scale Land Use/ Land Cover Classifica- WALMI (1995) Applications of Soil Survey in
tion Using Temporal Series of Modis Data. MS/MT. Irrigation Water Management. 4th ed. Aurangabad
Revista Brasileira de Cartografia, 59, pp. 1-7. (Maharashtra), Water and Land Management Institute,
Publication no. 21.
Kaul, H.A., and Ingle, S.T. (2011) Severity Classification
of Waterlogged Areas of in Irrigation Projects of Jalgaon Zubair, A.O. (2006) Change Detection in Land Use and
District, Maharashtra. Journal of Applied Technology in Land Cover Using Remote Sensing Data and GIS (A
Environmental Sanitation, 1(3), pp. 221-232. case study of Ilorin and its environs in Kwara State).
M.Sc Project, Department of Geography, University of
Kim, M., Xu, B., and Madden, M. (2008) Object based Ibaden.
vegetation type mapping from an Orthorectified Multispe-
ctral IKONOS Image using Ancillary Information.
Commission: VI, WG VI/4.

Lillesand, T.M., and Kiefer, R.W. (2000) Remote Sensing

and Image Interpretation. 4th ed. New York, Wiley and
Sons.

Maharashtra State Gazetteer (2001) Maharashtra State

Gazetteer. Government of Maharashtra, Cultural Affairs
and Tourism Department, Mumbai.

Mohan, M. (2005) TS24.5 Urban Land Cover/ Land Use

Change Detection in National Capital Regional (NCR)
Delhi: A Study of Faridabad District. In: Pharaohs to
Geoinformatics, FIG Working Week 2005 and GSDI-8,
April 16-21, 2005, Cairo, Egypt. TS 24- Planning for
Rural and Urban Communities.

National Remote Sensing Agency (2007) Geoinformatics

Based Land Use and Land Cover Changes in Khamm-
an District of Andhra Pradesh by Social Forestry Initi-

Available online at www.scientific-journals.co.uk 152

View publication stats