Application of Multispectral Remote Sensing data and GPS for area Estimation of Minor Millet - Kolli Hills, Tamil Nadu
L. Gnanappazham, M. Navamuniammal, R.
Rengalakshmi, K. Balalsubramaniam, D. Dhanapal, E. D. Israel Oliver
King and S. Krishnakumar
M. S. Swaminathan Research Foundation, Chennai – 600 113
Introduction
Survey of the existing land use / Land cover pattern, its spatial distribution & changes in the land use pattern is a pre-requisite for planning, Utilisation and formulation of policies and development of the Society. The capacity to provide real-time information makes it possible to show how the changes take place. The technique of Remote Sensing gives information of large area with less time, cost and efforts than the conventional surveys. It is also possible to prepare land use maps at various levels and gives quantitative results with the help of high-resolution musltispectral data like IRS 1D LISS III Remote Sensing data. GPS gives accurate latitude, longitude and altitude of any position on the earth. Identification of different vegetation types mapped from remote sensing imagery, in real world is found to be very easy with the help of GPS.
The
accurate & precise land use mapping can be done through field surveys but it would be a time consuming & expensive task. Instead, this study tried to apply the advantages of integrating Remote Sensing and GPS to map the land use of Kolli hills with limited and proper ground truth survey. Use of Remote Sensing and GPS has found to be greatly advantageous in biodiversity conservation in terms of mapping vegetation cover and area estimation in an area with heterogeneous distribution of crop varieties like Kolli hills.
The study area
Kolli Hills is bounded by geo-coordinates 11 10' to 11 30' N and 78 15' to 78 30' E. Kolli hill is situated in the Namakkal district of Tamil Nadu. Some part of the eastern portion of the hill lies in the Perambalur district. Kolli Hill has an area of 482 Sq.Km. It stretches 29 Kms from north to south and 19 km from east to west. Kolli hill is a part of the Talaghat stretch and eastward of the hill lies Pachamalai. (Fiqure 1) The highest point in the Kolli hill is 1380 m above MSL, but the general level of the upper surface of the hill is not more than 1000 m.
Methodology
Data used
The study area fall in one quadrant of IRS 1C LISS III scene, Path - Row No. 101 - 65 acquired on October 19, 1998 were used for this study. Topographical maps published by SOI on 1: 50000 scale was used to derive base details like administrative boundaries of Revenue Forest, Taluk boundaries, Location of Hamlets and Villages. Ground Truth information was collected in different Land use/cover locations with the help of GPS by getting the latitude, longitude and elevation details of the Land use/cover.
Data analysis
Preliminary interpretation was done using FCC on 1:50000 scale for delineating various land use classes by using image characteristics like tone, texture, pattern, shape & size. On the basis of this preliminary map, ground truth work was carried out to correlate interpreted categories on imagery with the ground information, which is a critical component in land use classification. The GPS locations for each land use classes, which could be identified on satellite data were collected later used as training sites for digital analysis of October 1998 and November 2001 data.
The 1998 and 2001 data were geometrically corrected by taking 20 ground control points (i.e.) permanent features from toposheets (SOI 1970). The area of Kolli hills under cultivation other than reserve forest area are separated from the whole data for ease of analysis and to avoid misclassification with forest vegetation. The image was spectrally enhanced to derive normal false color composite which could help in identifying the training sites on the image.
The data were analysed by supervised classification with maximum likelihood classifier with all 4 spectral bands (Green, Red Infra red and Middle infrared). The known regions of different land use types like Rock out, Scrub, Plantation, Tapioca, Millets, Paddy, Mixed and Fallow/Harvested are marked on the image, which are called “Training sites”. The statistics of these training sites were derived. The training sites having standard deviation less than 0.02 were selected first. The left out training sites are corrected for the limit 0.02 to get the classification of all training sites given. Then the supervised classification was done with maximum likelihood classifier. Final classified image was verified in the field for accuracy assessment with 70 points covering all classes. Overall accuracy of classification was found to be 85 % for 1998 and 80 % for 2001.
|
