| dc.description.abstract | Rice is fundamental for most Indonesian people. Monitoring rice field is therefore important in order to obtain information required by food or agriculture planning. This task could be accomplished by using satellite-based remote sensing technology, including ALOS. In this research, ALOS capability in provision rice field information are explored through its two main sensors, i.e. optical-based AVNIR-2 and PALSAR, which is a SAR sensor, primarily focused on fully polarimetric data. ALOS PALSAR data used in this research were acquired in 2007 and 2009, while ALOS AVNIR images were attained in 2008 and 2009. This research shows that NDVI values based on AVNIR-2 fluctuate during both observations. In 2008 and 2009 images, the NDVI range was 0,238-0,549 and -0,209-0,516 respectively. The diversity of NDVI value in rice field was caused by varying growth phases such as vegetative, generative, mature, and pre- harvest. In this research, rice growth phase were studied using linear backscatter analysis using fully polarimetric (PLR) PALSAR image. In many phases, HH consistently produced higher backscatter coefficient compared to other linear polarization components, i.e. HV and VV. Analysis showed that HH is the most suitable polarization for monitoring rice field. This is due to the fact that HH polarization is more sensitive to variations of rice structure in different age levels. Classification using decision tree was implemented by exploiting QUEST and CRUISE algorithms. Using PALSAR image, classification accuracy based on CRUISE and QUEST was 56,36 % and 40 % respectively. Accuracy on AVNIR-2 was higher than PALSAR, i.e. 94,74 % and 90,91 % using CRUISE and QUEST respectively. | en |
