#TITLE_ALTERNATIVE#

• Main Author: SASKIA PUSPITASARI (NIM : 25111001) ; Pembimbing I : Prof. Ketut Wikantika, Ph.D ; Pembimbing , ALIA
• Format: Theses
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/17892
• Institution: Institut Teknologi Bandung

Rice is a stapple food for Indonesia. Therefore a good agricultural planning requires estimation of paddy productivity. To predict rice productivity is not an easy thing because it often occurs miscalculation of areas, time of harvest and it’s production that affecting in making decision related to food security policies. To solve that problem is needed an alternative approach that able to provide information quickly, continuously and accurately. Hyperspectral remote sensing technology is one of geospatial technology which providing detailed information to distinguish the growth stage of rice crop and its production. This study uses Ciherang varietas that growing at Karawang area, West Java, Indonesia. Hymap data acquired by airborne sensor are used for this study. This paper discusses booting growth stage monitoring of paddy. The Hymap data are processed using spectral mixture analysis method (SMA) to identify and determine spatial proportion called endmember. It uses reference endmember and image endmember as an input for SMA method thus providing two image classification in booting stage based on image endmember and reference endmember. The data has been classified are used to estimate harvest productivity. The Results showed that a classification model from reference endmember have 45 % of overall accuracy and 44% of kappa accuracy. A classification model from image enndmember have 35% overaal accuracy and 21% of kappa accuracy. The regression model for reference endmember is Yield = 1434.860 - 811.589SMA_RE, and for image endmember is Yield = 1386.942 - 854.806SMA_IE. SMA method can be used to estimate harvest productivity.