Evaluation of thematic mapper data for mapping the rubber area under small holdings in Kerala

Evaluation of thematic mapper data for mapping the rubber area under small holdings in Kerala

G.Gopinathan & K. Abdul Samad
Specialists, Kerala State Land Use Board, India

Abstract
More than 90 percent of the Rubber produced in India is from Kerala. More than 80 percent of the area Under rubber in Kerala is accounted by small holdings. The present Rubber Board of India statistics are based mainly on the details available from the registered growers which contribute 4only less than 50 percent of the production. Hence an attempt was made to map out the extent of area under rubber in Kerala using Remote Sensing Data.

A pilot study to map out the extent of rubber in Panathur and Chittarikal area of Kasaragod district was taken up. Multi temporal false color composites of thematic mapper data was visually interpreted to map out the rubber area. Black and white aerial photographs of the same area was also interpreted to assess the effectiveness of TM data in mapping out the area under small holdings. The results revealed that multi temporal. TM FCC can be effectively used to map out the rubber in small holdings.

Introduction
Natural rubber can be extracted from the latex of over 895 species of plants, among which Have a brasiliensis is the most important commercial source of natural rubber. This is a quick k growing and a tall plant which is a native of Brazil and was introduced to tropical Asia in 1876. Recently rubber is attaining more and more importance in the world market. as a result India, especially Kerala State, is showing an increasing trend in area under rubber during the recent years. According to Rubber Board of India (RBI) \, at present 88 percent of the total area under rubber belongs to Kerala State and 90 percent of India's production is also from this state. It is also worth mentioning that in Kerala convention of area under other crops to rubber is attacking place with a high pace, probably due to high returns, the attractive incentives offered by the rubber Board of India and socio0economic reasons,. as per Robber Board of India estimate in Kerala nearly 20,000 hectares are being converted to robber crop annually since 1980.

The estimates of RBI show that during the past three decades there is a rapid increase in area and production of rubber in Kerala. The details of the last five years are furnished in the table below:

Table-i Estimate of area and production of Kerala as per Rubber Board of India

YEAR	AREA (hectares)	PRODUCTION (tonnes)
1983-84	2,94,290	1,62,212
1984-85	3,11,892	1,72,092
1985-86	3,26,710	1,84,563
1986-87	3,37,700	2,02,129
1987-88	3,50,881	2,16,562

During 1987-88 the area was 3,50,881 hectares as, against 64,700 hectares in 1955-56. The production has increased from 21,170 tones in 1955-56 to 2,16,562 tones in 1987-88. This shows that the area has increased to more than five fold and production to more than ten fold. This disparity of increase in production is due to the contribution from the increasing number of small holdings which are not accounted for in the total area accounted for in the total area estimates since they are not registered growers. It is estimated that there are more than 3,60,000 rubber holdings in Kerala during 1987-88. This is scattered throughout the State. Out of this there are only 254 estates covering a total area of nearly 47,000 hectares. This shows that more than 80 percent of the area under rubber in Kerala comes under small holdings. Previously the expansion of area under rubber took place only in the southern region of Kerala which is more or less saturated and during the last two decades this expansion has tan place in the Northern Kerala also. The Idukki Land use survey conducted by the Kerala State Land Use Board using aerial photographs in 1978-79 also revealed that the actual area under rubber was 22,000 hectares where as the Rubber Board figures showed only 15,400 hectares. This proved that the technique of remote sensing can be advantageously employed in the area estimation.

Rubber area delineated from TM FCC (2,3,4) Scale 1:50,000 (original)		Rubber area delineated from aerial photograph(B & W) Scale 1:50,000 (original)

	LOCATION MAP

This is a pilot study to evaluate the usefulness of Thematic Mapper (TM) False Color Composite (FCC) data, for mapping the area under rubber in Kerala especially small holdings to a reasonable level of accuracy. The Arial photography though yield more accurate data, it is very expensive, time consuming and difficult to procure. Munshi (1982) made an attempt for the determination wheat crop statistics using lands at data, pant etal (1988) used T, FCC on scale 1:2,50,000 for mapping different vegetation/land cover through visual interpretation. Yadav etal (1988) found TMM, FCC of bands 2,3, & 4 on scale 1:62,500 effective for monitoring and mapping of vegetal cover by visual interpretation. Singh (21988) used TN, FCC for visual interpretation in an attempt to map the land use/land cover.

Objectives

Mapping of rubber area using 1:15,000 black & white aerial photographs and also using 1:50,000 multitemporal TM FCC data.
Comparison of the area mapped from aerial photographs and TM data
Evaluation of multitemporal TM FCC for mapping the rubber area under smallholdings.

Study area
The study area covers Chittarikal and Panathur area of Kasarakode District in KeralaState, India with a geographical area of 22,000 hectares. It is situated between north of Nileswaram river (12o 17'N) to 12o30'N and 75o20'E to 75o25'E longitude. This area has been selected as it is in this area that rapid extension of area under rubber has taken place recently. Prior to 11980 there was no rubber cultivation in between Chittarikal and Pamathur estates. Hence the increase in area under rubber in the study area has taken place recently and that taken place recently and that too by small holdings.

The area has undulating to rolling topography within average annual rainfall of 3600 mm. The land use/land cover of the area is agriculture, habitation, forest, streams and rivers. The major crops of the area are rubber, coconut, cashew etc.

Data used
Multi-temporal LANDSAT FCC of Thematic Map per Data, bands 2,3, & 4, path-row No 145-52 dated 25-3-1986 & 19-10-1986 of scale 1:50,000 was used for visual interpretation. In addition 1:15,000 black & white aerial photographs of 1985 belonging to the same area were also used. Survey of India topographical sheet of scale 1:50,000 (No. 48 p/7) was used for the preparation of base maps.

Methodology
Visual interpretation of 1:50,000 TM FCC of the study area was carried out for demarcating the rubber pockets. The image interpretation elements namely tone and texture, were mainly considered at the time of interpretation. Black and white aerial photographs on 1:15,000 were stereoscopically interpreted and the rubber areas mapped out separately. Here in addition to tone and texture, the element pattern was also taken into account at the time of interpretation. Two base maps for transferring the interpreted details were prepared from the 1:50,000 survey of India top sheet. Thus two Thematic Maps on scale 1:50,000, one from TM FCC and another from rubber area were prepared with the help of a transferring instrument, Random ground checks were carried out to verify the accuracy of interpretation. The area under rubber was computed from two thematic maps separately using millimeter square grid. From the computed area the percentage of area under rubber were calculated separately from the two thematic maps. Then a comparative evaluation was carried out. The study took ten man days for the interpretation and mapping from TM data and fifty three Mondays for interpretation and mapping from aerial photographs.

Results and discussions
The details of the area under rubber computed from the two thematic maps is presented in the table below :

Table-ii

Sl. No.	Details studied	As per TM FCC	As per Aerial photo
1.	Total study area (hectares)	22,000	22,000
2.	Area under rubber (hectares)	2,388	3,842
3.	Percentage of area under rubber	10,85	17,46
4.	Area of smallest mapping unit (hectares)	5,00	0.25

The study revealed that it is possible to demarcate the rubber area from the non-rubber area using multitemporal TM FCC on scale 1:50,000 with a reasonable level of accuracy and within a limited time frame. The overall interpretation accuracy for TM data and aerial photographs were found to be 60 percent and 95 percent respectively based on the field checks conducted.

Evaluation of TM FCC data for mapping the rubber area especially under small holdings (less than 20 hectares) was done with respect to aerial photographs. The study revealed that multi temporal TM FCC provided new possibilities in using satellite based data for mapping the area under rubber in Kerala. This is due to the fact that the spectral resolution of the lands at TM FCC is very suitable for vegetation studies. The smallest mapping unit aerial photographs was 0.25 hectares. Moreover mapping of fresh plantations was possible from aerial photos due to the peculiar pattern of terracing and count our planting. This is due to the advantage of accounting majority of small holdings due to the peculiarity of growing the rubber crop in Kerala.

In Kerala rubber is generally grown in the midlands and high lands and major portion of small holdings are adjacent area. The small holdings under rubber in Kerala are mainly homestead planting. This will include buildings other crops etc. These homestead plantations are lying adjacent to each other. But they may not have uniformity in age and management. Mapping of major portion of these individual holding was possible from the aerial photographs. Mapping of these individual units was not possible with TM data due to the limitation of scale and resolution. Here the scale of TM used was 1:50,000 and so the smallest map able unit was above5 hectares. Mapping of well maintained pure plantation was very easy from TM data. They showed bright reddish pink tone with smooth texture in the imagery. This is due to the uniformity in age, management etc. of the plants. Mapping of small holdings was not so easy compared to pure plantations. Due to the variation in age, management etc.

These units did not yield very prominent tone and texture as in pure plantation. But it was possible to differentiate these units from forest area and other vegetation groups. A group of small holdings is seen as a single units in TM data due to the limitation of scale and resolution when compared with aerial photographs. Fresh plantations les than three years of age could not be mapped put from TM data due to their poor canopy coverage. But this was possible from aerial photographs. This is one reason for the variation of area computed from data and aerial photographs. The study concluded that the above aspects will not be a bottleneck in including the small holdings also in the estimation of average under rubber in Kerala.

As per this study the percentage of area under rubber in the study area (22,000 hectares_ according to aerial photographs and TM data are 17:46 and 10.85 respectively. This variation is not of much significance when compared with the advantage of TM data over aerial photographs such as easy availability, cost involved, synoptic and repetitive coverage, speed in interpretation etc.

Conclusion
The study revels that multi temporal FCC of TMM data on 150,000 scale can be effectively used for mapping the area under small holdings (les than 20 hectares) to a reasonable level of accuracy. Even though aerial photographs provide accurate information, it is expensive, time consuming and less feasible for large area studies. But TM FCCs' are readily available, cheap and needs only less effort sand man power for interpretation and mapping. The present study had taken 10 Mondays for interpretation and mapping of TM FCC about 53 Mondays for interpretation and mapping for aerial photographs. The time required for TM FCC can be still required for TM FCC can be still brought down to six or seven Mondays with experience.

Acknowledgement
The authors wish to express their sincere thanks to the Land Use Commissioner, Kerala State Land Use Board for providing the necessary facilities and also to their colleagues who helped them in this study.

References

kerala state land use board and space Applications Centre 1980, Land Use Survey OF Idukki District.
Munshi, M.K., 1982, A study on the determination of wheat crop statistics in India through utilization of lands at data Ph. D. Thesis, Department of Applied Mechanics, IIT Delhi, p. 147.
Rubber Board of India, Rubber and its cultivation 1985.
Indian society of Remote Sensing 1987, Proceedings of national symposium on remote sensing in land transformation and management held on December 21-22, 1987 at Hyderabad, India.
Pant, D.N. Saxena, K.G., parwal , M.C., Singh, I.J. and Ro, P.S., 19188, Analysis of fuel wood and fodder resources in Saharanpur district using remote sensing techniques. Proceedings national symposium on remote sensing in rural development. November 17-19.
Yadav, V.K., Mehrotra. A, Adinarayana. J and Gopal Rao. K, 1988, Pasture land management-A Remote sensing approach. Proceedings national symposium on remote sensing in rural development, Nov. 17-19.
Inderjit Singh,1988, Evaluation of landsat TM data for forest cover type and land use classification in sub tropical forests of Kumaon Himalaya (U.P.). Journal of Indian Society of Remote Sensing, Vol.16 No.4 (Dec.1988) p.43-52.
Department of Economics and Statistics, Government of Kerala., 'Statistics for Planning'