Crop Variable Value of Paddy Rice Prediction in Hyperspectral Data Using Support Vector Machine Method

Abstract

Hyperspectral is a new technology in remote sensing which exploits hundred of bands. Pusat Teknologi Inventarisasi Sumber Daya Alam Badan Pengkajian dan Penerapan Teknologi (PTISDA BPPT) applies hyperspectral in agriculture for yearly yield prediction. In this research, Leaf Area Index (LAI), number of chlorophyl (SPAD), and yearly yield has been predicted with hyperspectral data using support vector machine (SVM) method. SVM currently aroused many attentions due to its robustness in dealing with data with high dimensionality, including hyperspectral data. In principle, SVM works as a binary classifier to find the best separating hyperplane in the feature space. By modifying its loss function, SVM can be applied to regression problem in hyperspectral and Hymap hyperspectral remotely sensed data Region used are Indramayu and Subang; the growth period’s of paddy are vegetative, reproductive, and ripening, while the height’s of the spectral acquisition are 10 cm, 50 cm, and Hymap (2000 m). The data is owned PTISDA BPPT in cooperation with ERSDAC Japan. This research got prediction R2 maximum for LAI 0.96, SPAD 0.55, and yield 0.485. By using SVM, LAI values can be predicted quite well, but not for the SPAD values.This could be due to measurement error in the instrument of SPAD, while error’s of yield are probably due to measurement which using the example of a small area to represent a large area.