Development of Hyperspectral Data Utilization Technology by Using Data Mining Method for Paddy, West Java, Indonesia

摘要

This report describes the results of the study on the development of hyperspectral data utilization technology for paddy field in Indonesia such as growth stage classification for estimating harvest time, in collaborate with the Agency for the Assessment and Application Technology (BPPT) in Indonesia. The study areas are Indramayu and Subang of West Java in Indonesia. We acquired HyMAP data over both areas and carried out ground measurement in 2008. Also, we used MODIS time series data (MOD 13Q1). We used both the traditional approach like SAM and supervised classification, and new approach with data mining technique includes sparse linear discriminant function and LASSO regression. Because hyperspectral data has more than 100 bands, the data could work more effectively by using the method with data mining technique. This result indicates the potential of hyperspectral data. The hyperspectral and multispectral sensors is being developed by the Ministry of Economy, Trade and Industry (METI) of Japan toward its planned launch in 2014. This study result is one of the potential future operational applications for the new sensor.