International Journal of Environment, Ecology,

Family and Urban Studies (IJEEFUS)

ISSN (P): 2250-0065; ISSN (E): 2321-0109
Vol. 9, Issue 4, Aug 2019, 1-10
© TJPRC Pvt. Ltd.

CATEGORISATION OF SEASONAL VARIATION IN CULTIVABLE LAND

UTILIZATION USING LANDSAT IMAGES - A CASE OF NAYAGRAM

BLOCK, JHARGRAM DISTRICT, WEST BENGAL, INDIA

SUBIR KUMAR MOYRA1 & DR. SUKLA HAZRA2

1
Research Scholar, East Calcutta Girls’ College (WBSU), Lake Town, Kolkata, West Bengal, India
2
Principal, East Calcutta Girls’ College (WBSU), Lake Town, Kolkata, West Bengal, India
ABSTRACT

Categorisation of cultivable land utilization is a key element for the monitoring and estimation of crop
producing area as well as to establish a programme of cropland management. Such categorisation can be done either in
temporal span or in spatial difference. In this study attention is focused on assessing the prevailing difference in
cultivable land utilization with respect to different seasons. The object oriented general classification of satellite data only
leads to common information on the status of cultivable land utilization. However specified classification approach
(SCA) can be effectively used to facilitate multi seasonal crop and non-crop area delineation. The newly developed

Original Article
approach of classification has further enhanced by giving new dimension to training set collection. The application of RS
and GIS technique leads to estimate the seasonal variation in cultivable land utilization (SVCLU) based on multi spectral
satellite imagery of the different seasons i.e. Kharif, Ravi and Zaid for the year 2010 - 2011.

KEYWORDS: Classification, Categorization, Specified Classification, Seasonal Variation, Land Utilization & Crop Area

Received: May 10, 2019; Accepted: May 30, 2019; Published: Jun 12, 2019; Paper Id.: IJEEFUSAUG20191

1. INTRODUCTION

Production is the function of land, labour, capital and organization. However, in agricultural sector the
importance of land is rated top most as the production of crop directly depends on availability of cultivable land.
At present the rapid increase of population has resulted additional utilization of agricultural land to fulfil their
miscellaneous demands contrasting to the fact that the supply of land is limited and cannot be stretched according to
demands. It is proved by the fact that the availability of per capita arable land in India has decreased alarmingly
from 0.34 hectares in 1951 to 0.17 hectares in 2001 (Sharma & Ram 2009) [I]. The matter of providing food to all
along with the issue of shrinkage in arable land has given birth to a relevant challenge which can never be addressed
unless the utilization of land resource is maximized.

The concept of looking to the cultivable land utilization from the perspective of spatial change (Pierce
2007) [II] in net cropped area with changing season is known as seasonal variation in cultivable land utilization
(SVCLU). On the basis of advent and retreat of south west and north east monsoon, three agricultural seasons are
practiced in Indian subcontinent i.e. Kharif, Ravi and Zaid. There exists mark dissimilarity of seasonal cropped area
in most of the agro producing areas. The area of Kharif is still largest as our peasants are more interested to grow
crops with the help of nature gifted monsoonal rain water. In spite of infrastructural advancement and several
campaigns in irrigation sector a large portion of cultivated land still devoid of irrigation. A proper assessment of

www.tjprc.org [email protected]
2 Subir Kumar Moyra & Dr. Sukla Hazra

cultivable land on the basis of no. of crops grown in a particular year i.e. single or double or multi is of fundamental
importance in the present world (Roy 2012) [III].

The most precise method of determining SVCLU through base level data collection by agricultural field visit is
indeed very costly and requires skilled manpower. Beside collection of such a large amount of data within a stipulated time
period i.e. span of a particular season is considered to be a mammoth task. However in the present work emphasized are
given on integrated studies using remotely sensed satellite image and GIS tools. Satellite data in conjunction with
sufficient ground truth is considered as a reliable source of information for any geospatial analysis. This kind of study not
only save time and money but also increase the accuracy of the result as the biasness from the part of data collector is
minimized.

2. STUDY AREA

Nayagram, a community development block having an area of 501.44Km2, located in south western part of West
Midnapur district of West Bengal state, bounded by 22 o 44 / N to 22 o 74 / N latitude and 88 o 08 / E to 88 o 13 / E longitude
has been selected for on-going research rationale. The Study area possesses an undulating topography having a gentle to
moderate slope from north-west to south east direction. Elevation of the area varies between a maximum of 110 m. in
central north to a minimum of 10 m. in south east. The drainage of the area is controlled by Subarnarekha river system.
The Subarnarekha basin forming a part of metamorphic terrain of Chota Nagpur Plateau has attracted the attention of
eminent earth scientists for past years. But the area under consideration has received mere attention in this regard.

Figure 1: Location of the Area

The area falls under tropical monsoon climate with hot and humid summer. Average annual rainfall for the basin
is around 1615 mm. Mean minimum temperature of coolest and mean maximum temperature of warmest months is 13 o C
and 35 o C respectively. The study area can be divided into 4 soil categories namely, coarse loamy typic ustifluvents, coarse
loamy typic haplsfalfs, fine loamy ultic paleustalfs, fine loamy aeric ochraqualfs. The plateau is covered with a variety of
different habitats of which Sal forest is predominant. Beside forest a large parts is covered with bamboo grasslands and
shrubs. The selected study area has a total population of 142199 among which 40 % are tribal population (Census of India,
2011).

Impact Factor (JCC): 4.6869 NAAS Rating: 3.58

Categorisation of Seasonal Variation in Cultivable Land Utilization 3
using Lands at Images - A Case of Nayagram Block,
Jhargram District, West Bengal, India
3. METHODOLOGY
3.1. Software and Data Used

For the derivation of band combination statistics and image classification TNT Mips 2012 and for mapping and
presentation Arc GIS 10.1 has been used. Landsat satellite images for the month of November, February and April
covering the selected area were used. Images are collected for the agricultural season 2010 to 2011. The specifications of
the products are described in table 1.

Table 1: Satellite Data Used

Season Kharif Ravi Zaid
Digital image Landsat 7 ETM Landsat 5 TM Landsat 5 TM
Path / Row 139 / 44 &139 / 45
L5010473 & L6206755 &
WOID L5007727 & L5007980
L5010471 L5996999
Date of acquired 07.11.2011 13.02.2010 02.04.2010
Resolution 30 m. 30 m. 30 m.

3.2. Specified Classification Model

All the calculations and experiments are done for the images of Kharif, Rabi and Zaid season to show the
distribution of cultivable land and its seasonal variation. The methodology of the present appraisal was primarily based on
an integrated approach backed by data analysis based on Geographical Information system.

3.2.1 Classification Approach

Literatures inspected prior to this study have revealed that researcher commonly used indices approach to
discriminate cropped area from satellite images (Bannari 1995, Choudhury 1987) [IV, V]. This discrimination was done on
the mechanism of significant difference in both absorption and reflectance of electromagnetic radiation in Red and Near
Infra-Red (NIR) region due to the high amount of chlorophyll content in crop land. NDVI, SAVI, TSAVI are some
renowned vegetation index which are frequently used in this regard for their high sensitiveness to vegetation and
insensitiveness to soil-background (Ghulam et al. 2007) [VI].This approach is indeed very much challenging and bound to
suffer from lack of accuracy (Nash 2009) [VII] while applying it in an area mix with dense vegetation and agricultural tract
as found in the case of present study.

In order to mitigate such challenges, a unique non parametric approach of multispectral image classification,
specified classification approach (SCA) was developed in this regard. This has been introduced not only to minimize the
mixing of crop class with vegetation class but also to achieve a universal acceptance as a powerful cropped area
discriminating approach. Rees (2001) [VIII] described image classification as the process of making quantitative decisions
from image data, grouping pixels of the image into classes intended to represent different physical objects or types. It is the
most popularly used information extraction technique in digital remote sensing (Bhatta 2013, Moyra 2014) [IX, X]. In SCT
pixel sorting is done in order to cluster the pure crop pixel and reduce errors from target class.

3.2.2 Training Sites for Classification

The geometric relations in the spectral feature space of bands are also regarded as a good indicator for the
selection of training sites. During the selection of training sites for second order classification correlation of Red and
near-IR bands are used to identify cropped area (Figure 2). Soil reflectance in Red and near-IR represents high correlation

www.tjprc.org [email protected]
4 Subir Kumar Moyra & Dr. Sukla Hazra

with a positive correlation coefficient and are distributed along a line, called soil line (Liang 2004) [XI]. This soil line is
characterized by a linear equation (1). The scatterplot in the Red and near-IR spectral space often constitutes a triangle
shape. Those pixels with higher near-IR reflectance locate above the soil line are identified as a full grown crop.

pn = (b + γ.pr) (1)

(Where, b and γ are the intercept and slope of the soil line. Baret et al. (1993) [XII] derived a standard coefficients
value of b = 0.037 and γ = 1.176 for the delineation of soil line.)

Figure 2: Correlation of Red and Near-IR Bands in Different Seasons

3.2.3 Adoption of Decision Rule for Successive Classification

The decision rule is mathematical algorithm that allows the pixels to group into distinct class values on the basis
of data contained in the training set signature (Rahaman 2001, Sahu 2008) [XIII, XIV]. In this process spectral appearances
of individual pixel in the image are compared with the estimated cover type signatures. If it satisfies the predetermined
criteria fixed for a cover type, respective pixel is assigned to the cover type that matches to set criteria. There are several
types of decision rule exist in the arena of GIS field.

In this decision rule, both variance and covariance of the training class pixels are computed. As the classifier
delineates ellipsoidal contours with the equal probability in the scatter diagram (Lillesand and Kifer 1994) [XV], their
variance about the central value is represented by mean vector and covariance method. On the basis of these two
parameters probability of a candidate pixel belonging to each signature class is computed and compared with other
alternative classes. Ultimately pixel is assigned to the class which it actually belongs or it is left as unidentified.
The algorithm for the classifier is as follows:

D = log (ac) - [ 0.5 log (Covc) ]-[ 0.5 (X-Mc) T (Covc-1) (X-Mc)] (2)

(Where, D = likelihood distance; c = a particular class; X = the measurement of the candidate pixel; Mc = the
mean vector of the sample of class; ac = probability of pixel to be a member of class; Covc = the covariance matrix of the
pixels in the sample of class; T = transposition function.)

Impact Factor (JCC): 4.6869 NAAS Rating: 3.58

Categorisation of Seasonal Variation in Cultivable Land Utilization 5
using Lands at Images - A Case of Nayagram Block,
Jhargram District, West Bengal, India
3.3 Linear Combination

Here seasonal cropland layers have been converted to binary raster and each of them was assign with the coded
value 1 and 2. Numerical value 1 and 2 were used to represent cropland and fallow land respectively for each seasons.
Thus one can easily identify the seasonal variation in cultivable land utilization (SVCLU). At last all three raster layers
were combined in a linear fashion following the equation 3 to categorize the study area in terms of seasonal variation in
cultivable land utilization (Thompson 1980) [XVI].

SVCLU = ∑Cr * n (3)

(Where, Cr = each season; n= no. of seasonal raster)

4. RESULT AND DISCUSSIONS

The Landsat 5 TM satellite image for the month of February was classified to monitor the cultivable and
non-cultivable segment. The total cultivable area was assumed as 297.365 Km. 2. In the present study, determination of
cultivable area is done by the process of reclassifying of land use groups based on general land use identification. The
cultivable area has been used for three seasons by means of polygon extraction.

4.1 Specified LULC

Unlike the general LULC, the spectral variability in the Red and near-IR region can be considered as the
determining factor for specified LULC which is basically prepared with the goal to determine how effectively crop pixels
are grouped. Based on this fact specified LULC of cultivated area is prepared for three different seasons having 6 classes
each (figure 3). What is most important to note that among the 6 classes only one class have identified as crop class based
on spectral triangle concept and rest 5 classes were consists of non-cultivable spectra.

Figure 3: Specified LULC Raster of Three Seasons

In case of specified LULC generation, there must present some loss of information. Hence accuracy assessment is
always required while generalizing with RS data (Asraf1992, Panigrahy 1998) [XVII, XVIII]. With the help of Kappa
coefficient the classified is quantitatively determined with the help of table 2. Accuracy of full grown crop is found 99%
and above for Ravi and Zaid season resembling very high level of accuracy. Accuracy of Kharif is found 91.43%, slightly
below but above the overall Kharif accuracy of 91.20 %.

www.tjprc.org [email protected]
6 Subir Kumar Moyra & Dr. Sukla Hazra

Table 2: Accuracy of Specified LULC

Kharif Ravi Zaid
Specified LULC
Producers Users Producers Users Producers Users
Class
Accuracy Accuracy Accuracy Accuracy Accuracy Accuracy
Dry Fallow 97.33% 99.37% 95.27 % 98.94% 98.91% 100.00%
Wet Fallow 91.71% 74.68% 92.40% 88.89% 82.47% 63.24%
Barren Land 100.00% 52.54% 100.00% 87.21% 96.45% 96.20%
Full Grown Crop 91.43% 96.48% 99.00% 85.34% 99.68% 100.00%
Wet Channel 63.76% 96.94% 100.00% 100.00% 81.75% 82.35%
Sand 100.00% 100.00% 86.59% 100.00% 100.00% 91.98%
Overall Accuracy 91.20% 95.09% 97.79%
Kappa Statistics 0.8812 0.9269 0.9306

4.2 Seasonal Variation in Cultivable Land Utilization

In order to distinguish cropped area from fallow area five specified LULC groups i.e. Dry fallow, Wet fallow,
Barren land, Wet channel and Sand classes are merged for three different seasons. Thus raster layers having spatial
information of cropped and fallow characteristics are prepared.

Figure 4: Seasonal Cropped and Fallow Area

Studying the figure 4 and go through with the generated statistics of seasonal variation as shown in table 5, it was
shown that there exist a mark seasonal variation in spatial extension of cropland. It increases in the Kharrif season with
favorable agro climatic condition and reaches its base in Zaid season with prevailing adversity due to dryness of soil and
lack of irrigation facility. In figure 9 efficiency of land utilization in terms of different seasons is represented by percentage
value.

Table 5: Seasonal Variation

Cropping Cropped Seasonal Mean Deviation from Standard
Season Area (Km2) Area (Km2) Mean (Km2) Deviation
Kharif 107.385 +34.98
Ravi 75.912 72.400 +3.512 30.090
Zaid 33.913 -38.487

Impact Factor (JCC): 4.6869 NAAS Rating: 3.58

Categorisation of Seasonal Variation in Cultivable Land Utilization 7
using Lands at Images - A Case of Nayagram Block,
Jhargram District, West Bengal, India
4.3 Categorization of SVCLU

Identified both classes, i.e., crop and non-crop area of each raster for Kharif, Ravi and Zaid season was assigned
with the code 1 and 2 respectively. Code 1 is represented by crop area and code 2 signifies fallow area. Finally the linear
combination of the three binary raster of the respective seasons following the equation 3 has produced the categorization
map of SVCLU (figure 5). From which different code type of values has depicted typical crop lands like values having
single 1 and double 2 are either Kharif or Ravi or Zaid crop lands which are of single crop land type; double 1 and single 2
are either Kharif and Rabi crop lands or Kharif and Zaid or Ravi and Zaid which are double crop land type and 111 is
Kharif, Ravi and Zaid crop land which are triple or multiple crop land.

Figure 5: Categorization of SVCLU

5. CONCLUSIONS

Variation in cropped area is regarded as an important parameter to measure the availability of food. Due to
unfavorable condition some time the cropped area reduced heavily. On the basis of field data collection it is quite
impossible to calculate the exact figure of shrinkage or increase. Present specified classification model will enable to
classify the full grown crop area from Landsat image. During the modelling precaution has been taken to collect all three
season image with full grown crop.

Overall analysis of our study explore that multi seasonal analysis enables to find out the seasonal variation in
cultivable land utilization as well its categorization. Total of 8 agriculture land use class has been derived which helped to
trace the land utilization dynamics of the area. It is evident from the findings that the area is mono cropped area with

www.tjprc.org [email protected]
8 Subir Kumar Moyra & Dr. Sukla Hazra

maximum weightage on Kharif class. Specified classification approach provides effective options to truly group the
agricultural feature. With the present approach a new methodology of monitoring the SVCLU e.g. single crop to double,
double to triple and vis. have been developed.

6. FINANCIAL AND ETHICAL DISCLOSURES

Funding: This study was conducted without having contribution of any funding agencies.

Conflict of interest: The authors declare that they have no conflict of interest.

REFERENCES

1. Sharma, D D, Ram N (2009). Environmental implication of landuse / landcover changes in Solan district, Himachal Pradesh,
Geographical Review of India, vol. 71, no. 3, pp. 287-296.

2. F.J. Pierce, D. Clay (Eds.) (2007). GIS Applications in Agriculture, CRC, Boca Raton, p. 224

3. Roy P K (2012). Economic geography – a study of resources, New central book agency, pp. 147.

4. Bannari, A., Mori D, E. Bonn (1995). A review of vegetation indices. Remote Sensing Reviews 1995 (13), 95–120.

5. Badhe, D. K., & Saiyad, A. S. (2011). Constraints faced by the brinjal growers in adoption of recommended productions
technology. Agriculture Update, 6(2), 8-10.

6. Choudhury B J (1987). Relationships between vegetation indices, radiation absorption and net photosynthesis evaluated by a
sensitivity analysis, Remote Sensing of Environment 22 (2), 209–233.

7. Ghulam A, Li Z-L, Qin Q, Tong Q, Wang J, Kasimu A, Zhu L (2007). A method for canopy water content estimation for highly
vegetated surfaces—shortwave infrared perpendicular water stress index, Sci. Chin. Ser. D: Earth Sci, vol. 50, pp. 1359–1368.

8. Mourya, S., & Chakraborty, S. Assessment and Impact of Ambient Air Quality in Sonepur-Bazari Opencast Project–AN
Approach towards Sustainable Environment, Raniganj Coalfield, Barddhaman, West Bengal, India.

9. Nash, E., Korduan, P., Bill, R. (2009). Applications of open geospatial web services in precision agriculture, A review, Precis.
Agric. 10(6), 546–560

10. Rees W G (2001).Physical principles of remote sensing, 2nd edition, Cambridge University Press, 269p.

11. Satty T L (1980). The analytic hierarchy process: planning, priority setting, resource allocation, McGraw-Hill, New York,
p287.

12. Bhatta, B. (2013). Remote sensing and GIS. Oxford university press, pp. 360-368.

13. Moyra, S. K., Hazra, S. & Roy, R. (2014). Application of multi criteria decision analysis for the identiofication of artificial
ground water recharge zone, International journal of Remotesensing and GIS, vol. 3, no. 5, pp. 28-37.

14. Liang S (2004). Quantitative remote sensingof land surfaces, A John Wiley & sons. Inc. publication, pp. 248-252.

15. Baret F, Jacquemoud S, Hanocq J F (1993). The soil line concept in remote sensing, Remote Sens. Rev. 7: 65-82.

16. Panja, S., Mandi, V., Bhattacharya, C., & Sarkar, K. K. (2016). Performance of True Potato Seed (TPS) Hybrids in Gangetic
Alluvial Zone of West Bengal. IASET: International Journal of Agricultural & Bio-Chemical Science (IASET: IJABS) ISSN
(P): Applied.

17. Rahman Md R, Islam A H M H (2001). Wheat area and yield estimation using remotely sensed data, The Journal of Geo-
Environment, 1: 91–99

Impact Factor (JCC): 4.6869 NAAS Rating: 3.58

Categorisation of Seasonal Variation in Cultivable Land Utilization 9
using Lands at Images - A Case of Nayagram Block,
Jhargram District, West Bengal, India
18. Sahu K (2008) Text book of Remote Sensing and Geographical information Systems, Atlantic publishers, pp. 377-380.

19. Lillesand T M, Kiefer R W (1994). Remote sensing and image interpretation,” John Willey and sons,Newwork.

20. Chakraborty, S., & Basu, D. (2018). Homestead Gardening: An Emerging Venture Towards Achieving Food Security &
Nutritional Security-A Study of Selected Areas of West Bengal. International Journal of Applied and Natural Sciences (IJANS)
ISSN (P), 2319-4014.

21. Thompson D R, Wehmanen O A (1980). Integrating multi-criteria evaluation with geographical information systems.
International Journal of Geographical Information Systems, vol. 5, no. 3, pp. 321-339.

22. Asrar G F, Myneni R B, Choudhury B J (1992). Spatial heterogeneity in vegetation canopies and remote sensing of absorbed
photosynthetically active radiation: a modeling study, Remote Sensing of Environment 41, 85–103.

23. Panigrahy S, Chakraborty M (1998). An integrated approach for potato crop intensification using temporal remote sensing
data. ISPRS Journal of Photogrammetry and Remote Sensing, 53: 54–60

www.tjprc.org [email protected]