DETECTION OF RICE-PLANTED AREA USING MULTI-TEMPORAL ALOS/PALSAR DATA

A rice-planted area map is a basic information resource for rice production management. In this study, a method is proposed to detect the annual rice-planted area by using multi-temporal phased array type L-band synthetic aperture radar. This method addresses two problems: (i) the sigma(0) value of planted paddies is not consistent; and (ii) annual data fails to reveal time series change of sigma(0) value in rice-planted area. Data were acquired from both planted and non-planted paddies in the dry season. Accuracy assessment was performed based on field survey data. As a result, the proposed method was effective at revealing the annual rice-planted area. It was also found that data acquired in the rice growth stage contributes to determining the rice-planted area with high accuracy. Further investigation is required that sets more detailed classes to reduce the rate of misclassification and eliminates small regions in detected images.