Thumbnail
Access Restriction
Open

Researcher NOBIN CHANDRA PAUL
Advisor Sahoo, Prachi Misra
Source KrishiKosh-Indian National Agricultural Research System
Content type Text
Educational Degree Master of Science (M.Sc.)
Publisher ICAR-Indian Agricultural Statistics Research Institute ICAR-Indian Agricultural Research Institute New Delhi
File Format PDF
Language English
Subject Domain (in DDC) Technology ♦ Agriculture & related technologies ♦ Animal husbandry
Subject Keyword Statistical Techniques For Discrimination and Acreage Estimation of Fruit Crops Using Hyperspectral Satellite Data ♦ Agricultural Statistics and Informatics
Abstract Horticultural crops plays a unique role in India’s economy therefore reliable and timely estimates of area under horticulture crops are of vital importance. Present methods of crop acreage estimation rely heavily on sample survey approach which is time consuming for a diversified and large country like India. Modern space technology with the advance tools of Remote Sensing, GIS and GPS may be an alternative option for estimating area under horticultural crops. The advantage of using satellite data is that it provides both synoptic view and the economies of scale, since data over large areas could be gathered quickly from such platforms. Generally till now, multispectral remote sensing data has been used for effectively discriminating field crops and estimating area under crops. But major limitation of multispectral data is lesser number of bands which may not be able to discriminate fruit crops. Hyperspectral data has relatively large number of bands which helps in discriminating fruit crops easily and more effectively. This study has been undertaken to investigate statistical techniques for discrimination of fruit crops and to estimate the acreage and map existing orchards of fruit crops using hyperspectral satellite data. The four tier hierarchical statistical techniques has been proposed for discrimination of fruit crops which includes one way ANOVA, CART, JaffriesMatusita (J-M) Distance and Linear Discrimination Analysis (LDA). Initially starting with 2151 band, after ANOVA these bands were reduced to 1876 and were further reduced to 10 after using CART. J-M distance analysis was performed on these ten wavebands to check whether these wavebands could discriminate different fruit crop pairs. The overall accuracy was assessed using LDA. The study has been conducted on three areas which include Sabour in Bihar, Meerut in Uttar Pradesh and IARI, New Delhi. Further the area under mango orchards was estimated using hyperspectral image of Meerut district. The estimates were compared with actual area under mango orchards measured using Global Positioning System (GPS) and the total area under mango was predicted. This study concludes that hyperspectral data has more discriminative power than the multispectral data for discriminating fruit crop. The proposed four tier statistical method can be utilized efficiently for discrimination of fruit crop orchards. The study also reveals the scope of hyperspectral remote sensing in acreage estimation of fruit crops.
Educational Use Research
Education Level UG and PG
Learning Resource Type Thesis
Size (in Bytes) 4.06 MB